TRANSCRIPTEnglish

Getting Started in ICS/OT Cyber Security - 20+ Hours - Part 4 (Secure Network Architecture)

1h 29m 35s14,201 words2,115 segmentsEnglish

FULL TRANSCRIPT

0:01

hello and welcome back in this section

0:04

we're going to talk about secure network

0:06

architecture so so far in the first

0:08

three parts of the course we've really

0:11

level set trying to get everybody

0:14

whether you come from an IT background

0:16

or an OT background a basic

0:20

understanding of the different

0:20

components that can make up a IC or OT

0:25

environment and looking at some of the

0:28

basics of cyber security but now we're

0:32

going to look at how we bring all of

0:35

that together and starting moving

0:37

forward really what we're talking about

0:39

is how do we especially from a a

0:42

practical perspective how do we secure

0:45

our critical infrastructure environments

0:48

how do we secure our OT

0:51

networks and so let's go ahead and jump

0:54

into this section and you can see what

0:57

we're going to look at at this idea of

0:59

what secure network architecture brings

1:02

us and this is where I always start when

1:04

when I talk about if I could only do one

1:07

thing to protect an OT Network what

1:09

would it be and it would be to put a

1:13

firewall between it and and OT and that

1:18

sounds very simple and simplistic to

1:21

many many people and some people think

1:23

of it as a joke and and and and others

1:27

realize the the beauty in that

1:28

Simplicity and I know and understand

1:30

it's not always possible but we'll have

1:32

those discussions um the idea is that we

1:35

start with a firewall and then we'll

1:37

look at how we build additional layers

1:40

of secure network architecture to help

1:43

protect the

1:45

environment a big part of that is what

1:48

we're going to use to help us do that is

1:52

the Purdue model or what we call really

1:54

the expanded Purdue model today in

1:57

2024 so so we'll talk about the the

2:00

expanded Purdue model and how we can do

2:04

additional Network segmentation around

2:07

zones and conduits based off of Isa

2:10

62443 right we've started to talk about

2:13

Isa 62443 as the golden standard of how

2:16

we build a cyber security program in

2:19

control system

2:21

networks we'll also talk about how

2:24

attackers get into the environment with

2:26

the killchain and whether we talk about

2:28

the it cyber killchain from loed Martin

2:30

so how do the attackers get into the

2:32

environment into the IT

2:35

Network versus well once they're in the

2:38

it Network how do they get into the

2:40

control system Network and then how do

2:44

they spread control and and gain access

2:46

and gain further control over the OT

2:49

Network and that's really what's

2:51

described in the IC silver cyber kill

2:53

chain that we're going to talk

2:55

about and then we'll wrap up with brief

2:59

introduction to the industrial internet

3:01

of things or iot where we're basically

3:05

allowing the OT Network to have internet

3:08

access but of course we want to do that

3:11

as securely as possible so we're going

3:12

to introduce that here and we'll come

3:14

back and talk about it in later sections

3:16

and same thing with secure remote access

3:18

which is it's this guess necessary evil

3:20

if you want to describe it as such to

3:23

allow remote parties like

3:26

vendors to be able to come in and do

3:28

things like maintenance Maybe some needs

3:30

to come in and maybe do a firmware

3:31

upgrade on a on a PLC during a

3:33

maintenance window we want to allow them

3:36

to be able to do that and especially if

3:38

it's in a dangerous environment we don't

3:40

necessarily want somebody standing there

3:42

in the plant doing the upgrade when they

3:44

can do it safely from

3:46

home we just want to make sure if we're

3:48

allowing them to do an upgrade safely

3:51

from home remotely that that secure or

3:54

that network connection is as secure as

3:58

possible you just have to remember it's

4:00

it's never 100% so if we're going to

4:03

have it we have to be extremely careful

4:08

and we have to make sure we do secure

4:11

secure remote access the right way so

4:13

that's what we're going to talk about in

4:16

this part so we already previously

4:19

started to talk about the five main ways

4:21

that attackers get into OT networks and

4:24

the primary route that we see is they

4:27

first get into the IT Network right

4:30

using a fishing attack which is the most

4:32

common way that attackers will get in

4:35

and we'll come back and and talk more

4:36

about that and that's really kind of the

4:40

the main example that we're going to

4:42

talk about in this section we'll come

4:44

back and talk about all of these as

4:46

we're going throughout the course but

4:48

the main focus in this part is really

4:50

talking about as attackers gain access

4:53

to the it Network then what do they do

4:56

how do they gain access to the OT

4:58

Network and then once they're in the OT

5:02

Network it's our job to ideally detect

5:06

them be able to find them and get them

5:08

off the network unfortunately we know

5:11

that a lot of OT environments at least

5:13

half don't have any type of network

5:16

security monitoring capabilities to find

5:18

attackers in the

5:19

environment so let's take a step back

5:21

and well where do we

5:27

start so taking that step back let's

5:29

talk about the

5:31

it kill chain or the it cyber kill chain

5:34

that locked Martin had created you can

5:36

see back in 2011 when uh they had

5:39

introduced this idea of this process on

5:42

explaining how attackers get into the

5:45

environment and this is not every

5:47

attacker and this is not every every

5:50

route that they can use to get into a

5:52

network but in general and this is the

5:55

the common kind of scenario that we're

5:57

going to use uh for our conversation for

6:01

this part at least when we talk about

6:04

attackers breaking into the IT

6:06

environment the most common route

6:09

they're going to use still today is

6:11

fishing attacks like we saw with

6:13

Colonial pipeline right attackers

6:15

whether they're it's an advanced

6:18

persistent threat trying to get into the

6:20

environment for Espionage or it's a

6:22

ransomware group like dark side and

6:24

Colonial pipeline that wants to infect

6:26

your computers just to get a couple of

6:28

bucks at the end of the day they all

6:31

realistically follow the same type of

6:35

pattern and how they break in into the

6:38

environment so the first step is they

6:40

perform reconnaissance or sometimes you

6:43

hear of things like ENT or open source

6:46

intelligence gathering but they they

6:48

perform reconnaissance right they want

6:50

to find out information about your

6:52

organization information that they need

6:55

to help their attack be more successful

6:58

so if I'm an attacker and I want to

7:00

break into your environment using a

7:02

fishing attack I want to send fishing

7:04

emails well I'm going to need to start

7:06

with email addresses so we'll

7:09

use the greatest open- source resource

7:13

in the world the internet typically

7:15

Google to be able to go ahead and find a

7:18

list of email addresses for employees

7:22

that work at your company and maybe I

7:25

don't have as many email addresses as I

7:27

want but if I understand the email

7:30

structure how are your emails created

7:33

meaning oh or when we have a new

7:36

employee right we create an email

7:38

address for that employee with their

7:40

first name followed by a period followed

7:43

by their last name followed by the at

7:45

sign followed by the the company

7:48

domain so I might have michael. holom

7:54

microsoft.com so if I go into LinkedIn

7:57

and see another Microsoft employee

8:00

even though I don't know their email

8:02

address if their name is say Jennifer

8:07

Salvador I know that more than likely

8:10

her email address is jennifer. Salvador

8:14

microsoft.com so understanding that

8:16

naming convention or that syntax I can

8:18

use that to go

8:21

through social media especially sites

8:24

like LinkedIn to gather names of

8:27

employees that work at companies

8:30

generate their email

8:32

addresses and so I have that list of

8:35

targets and there's other information we

8:38

can gather but that's a lot of what

8:40

we'll find in you kind of General

8:42

fishing attack

8:43

reconnaissance if you want to make you

8:45

know the fishing attacks more specific

8:48

to the organization you can see you know

8:50

look for different events that are

8:51

happening with the organization or maybe

8:53

you want to tailor those fishing emails

8:56

to specific individuals so you might

8:58

look them up in social media media maybe

8:59

not just LinkedIn but more like Tik Tok

9:02

and Facebook and insta to be able to see

9:05

more of their life so you can put more

9:07

personalized information into an email

9:10

just for

9:12

them this isn't necessarily a course on

9:15

creating fishing attacks but the idea is

9:17

we get as much information as we need to

9:20

make a realistic fishing attack to be

9:23

able to send to the email addresses that

9:26

we've gathered for for the organization

9:30

now the fishing email itself is is

9:32

pretty harmless we need to couple that

9:35

fishing

9:37

email with either and usually one of two

9:41

things it's either going to contain a

9:43

link to a malicious website and that

9:45

malicious website is going to host code

9:48

that'll infect the victim's

9:52

computer or we're going to put a

9:55

malicious or infected

9:57

attachment in the fishing email so when

10:01

the victim receives the fishing email if

10:04

they double click or launch that

10:05

attachment then it will infect their

10:07

computer so that's the idea of

10:09

weaponization now the key with these two

10:11

phases when we talk about reconnaissance

10:13

and weaponization is these are typically

10:16

almost always the two phases that we

10:18

cannot

10:20

detect from some type of network

10:22

security monitoring process and we can't

10:26

we can't see if an attacker is using

10:28

Google or to look up information about

10:30

our company or if they're in LinkedIn

10:32

looking up information about our

10:34

employees we

10:36

can't we can

10:40

uh we don't have an idea of what type of

10:43

malware are they creating right as an

10:45

attachment or a malicious file or or

10:48

site or what site that is right we have

10:50

we have no visibility unless we're

10:52

literally looking over their shoulders

10:53

or we have their systems down we have no

10:56

idea or visibility into what they're

10:58

doing

10:59

so these two phases of the kill chain

11:01

there there's not much we can do here

11:03

other than we assume that they're

11:06

happening so we're going to assume that

11:09

attackers are performing reconnaissance

11:11

on our environments that's why you

11:13

should be performing reconnaissance on

11:15

your own environments to see what the

11:17

attackers are going to see and that's

11:18

what we're going to come back and look

11:19

at in part seven especially when we talk

11:22

about using showand toine exposed OT

11:26

assets on the internet

11:29

so we want to perform reconnaissance on

11:31

ourselves we do the

11:33

ense on ourselves to see what the

11:36

attackers are seeing so if we can find

11:37

an issue before they do then we can fix

11:40

it before they can exploit

11:42

it so in those first two phases we

11:46

don't necessarily know 100% that they're

11:50

happening but we're assuming they're

11:53

happening

11:54

regardless now next Once the attacker

11:57

has their fishing email created and

11:59

they've created let's say in this case

12:01

they've created a malicious attachment

12:02

that's going to go with a fishing email

12:04

so they're going to send that to their

12:08

target employees and whether it's every

12:11

employee in the organization or maybe

12:13

it's just a select few you're going to

12:17

send those fishing emails right that's

12:19

the the delivery phase now when we get

12:22

an employee to fall victim so they get

12:25

the email they fall for it and they

12:27

doubleclick that attachment

12:29

that's where the exploitation comes in

12:31

where it starts so then that malware

12:34

installs on the machine during the

12:36

installation phase and once it's

12:39

installed almost always that system is

12:43

configured with the malware to go out

12:45

and make a connection to what we call a

12:46

command and control server out on the

12:48

internet that's a host that the

12:51

attackers have control over so when

12:54

let's say Jennifer Salvador at Microsoft

12:57

Falls victim to and I hope there's not a

12:59

real Jennifer alador at Microsoft but

13:02

there could be but when she falls victim

13:04

to that fishing email it opens up that

13:07

connection to the computer that the

13:10

attacker controls and it gives the

13:13

attacker then access to her computer as

13:16

her so the attacker now has access to

13:19

all of her data her application her

13:23

email and then potentially the attacker

13:25

can then take additional steps to gain

13:27

full administrative control control over

13:29

that system and then look at gaining

13:33

full administrative control over other

13:35

systems in the environment especially if

13:37

they're looking for sensitive

13:40

information so those are the first kind

13:42

of the six main phases and I it's

13:45

jumping ahead a little bit but then the

13:47

last phase is really once the attacker

13:49

has that foothold in the environment

13:51

once that machine's infected and they

13:53

have that connection to their command

13:54

and control

13:55

server then they want to be able to do

13:59

what they're going to do so depending on

14:01

what is their objective right what is

14:03

their mission is it for industrial

14:06

Espionage is it to plant ransomware is

14:08

it to the list goes on and

14:11

on but it's going to depend that's the

14:14

again that's the general it cyber kill

14:17

chain as far as explaining in general

14:19

most of the time how attackers get into

14:22

the

14:27

environment so building off of that

14:30

let's go back to the you know thinking

14:32

of colonial pipeline or the common

14:35

ransomware compromise and how that

14:38

impacts an an OT environment when it's

14:41

connected to an IT Network and there is

14:44

no security or we talk about a world

14:47

without secure network

14:50

architecture so again using this example

14:53

of a fishing attack let's say we have an

14:55

attacker that sends a fishing email to

14:58

an employee there's there's Jennifer

15:01

Jennifer gets the email address and she

15:03

falls victim to it and she double clicks

15:06

she opens up that attachment in this

15:08

case that ransomware is designed not

15:09

only to infect her computer but to also

15:12

spread to all of the other Assets in the

15:15

environment at least those that are

15:16

running Windows so anything in this case

15:19

running Windows is going to become

15:23

compromised rather easily rather quickly

15:26

and so you pretty much have the entire

15:28

it Network completely compromised we

15:31

talk about the entire environment

15:32

burning down to the to the

15:37

ground but if this is an IT environment

15:40

connected to an OT Network like let's

15:42

say our power plant it's this is the

15:44

back office to our power plant and if

15:47

there's

15:49

no segmentation there's no protection

15:52

between it and OT then that infection is

15:55

just going to move right into the power

15:57

plant and and in fact all the windows

16:00

assets that we have and as we've already

16:02

started to talk about most of our assets

16:05

that we see at least higher levels of

16:06

the Purdue model which we're going to

16:08

talk about in this

16:10

section most of those section most of

16:13

those assets they're just other

16:15

computers running Microsoft

16:19

Windows so there's no additional

16:24

protection and so this is where when we

16:27

have different types of compromise is

16:28

like Colonial pipeline again I don't

16:30

know all the details unless you probably

16:32

work at the FBI or at Colonial pipeline

16:35

you nobody knows the complete details of

16:38

what

16:39

happened but when the it environment

16:42

became infected with ransomware you know

16:45

did that infection start to move into

16:47

the OT environment because of a lack of

16:50

security control again I don't I don't

16:54

know I believe

16:56

so but I don't know

17:03

so how do we protect against this this

17:05

is where we were talking about firewalls

17:06

earlier not to jump

17:09

ahead to take a step to the side though

17:13

and talk about the Purdue model which we

17:15

started to mention right this idea that

17:17

we have this structured almost

17:20

framework that we can use to understand

17:23

how it and OT talk with each other and

17:27

then how the it and the OT networks are

17:31

comprised of their own layers that talk

17:33

amongst themselves and that we can use

17:36

this to start to have a conversation

17:40

about

17:42

security so the Purdue

17:44

model and you can see it was created in

17:47

in the early 90s I actually I'm actually

17:49

I get to nerd out about this because

17:50

there was a floor engineer that was on

17:52

the in in the working group that created

17:54

the the Purdue model and this is this is

17:57

not the original Purdue model this is

18:00

what we call that you see here this is

18:01

what we call the expanded Purdue

18:04

model but the idea is that the Purdue

18:08

model and the expanded Purdue model

18:11

right can be used to break up an itot

18:14

environment into these different levels

18:16

or

18:18

layers and that we look at how we can

18:22

secure the environment from a network

18:25

perspective and so this is where we

18:27

start when we talk about secure network

18:30

architecture in OT and you can see at

18:33

level four or five those top layers

18:35

that's the it environment that's the bo

18:37

the back office right that's where our

18:38

employees sit and they read their emails

18:41

and they browse the internet and they

18:42

get

18:44

infected but they're they're directly

18:46

connected to the internet right so

18:49

that's where we see the the most danger

18:51

now at the lower levels that's where we

18:53

have the the OT environment so you can

18:56

see levels 0 1 and two to really that

19:00

kind of make up those say the lower

19:02

levels of the OT and that's where we

19:04

have our things like our

19:06

plc's that bring in data from things

19:10

like sensors right and control different

19:13

systems like in our power plant the

19:14

turbines and the combustion Chambers and

19:16

the

19:18

generators and then we use different

19:20

systems like engineering workstations to

19:23

be able

19:24

to update programming on those plc's or

19:28

upgrade the firmware and we collect

19:30

process data and store that on data

19:32

historians and we're able to see or

19:35

visualize the process that's that's

19:37

happening in the environment on those

19:39

hmis and and so on and so

19:44

forth and then we still have more to the

19:46

OT environment because at level three we

19:49

have a whole other layer an upper layer

19:51

if you want talk about they have it

19:53

labeled here site operations and this is

19:55

this is based off of the the S version

19:58

of the expanded Purdue

20:00

model so we have an additional layer of

20:03

engineering workstations and data

20:05

historians which we're going to come

20:06

back and talk later about why we have

20:08

multiple

20:09

layers we can have things like active

20:11

directory for doing authentication for

20:13

our windows-based

20:16

assets might have some other type of

20:19

management

20:21

system that's specific to that different

20:23

type of environment so if I have a power

20:26

plant versus a Manufacturing facility

20:29

versus a railway right where each is

20:32

going to have its own different type of

20:34

management system or

20:36

systems that are specific to that type

20:38

of environment so we'll look at some

20:40

more examples of that later on as

20:43

well so really what we're doing is we're

20:45

saying levels four and five are the o or

20:48

sorry levels four and five up top that's

20:50

the it environment levels 0 1 2 and

20:54

three that's the OT Network

20:59

now between the two we create a DMZ or

21:03

what you see here is level

21:06

3.5 so just like we have if you're

21:08

familiar with an IT DMZ between the it

21:11

Network and the internet right we put

21:13

that that DMZ to help protect an extra

21:16

layer of protection between the it

21:18

Network and an untrusted Network like

21:21

the

21:22

internet now from an OT perspective

21:24

right we're not con connected directly

21:26

to the internet ID

21:29

but the one network we're going to be

21:31

connected to is the it

21:34

Network and that's the untrusted network

21:37

from the OT

21:40

perspective so we want to make sure

21:41

we're going to put a DMZ between us and

21:46

the IT

21:48

Network so that's what level 3.5 is so

21:52

we're going to create a DMZ and ideally

21:55

with two different layers of firewalls

21:59

and that when we have resources that

22:02

need to talk with it we're going to

22:05

place them in that

22:07

DMZ now remember we don't if we can get

22:10

away with it if we can if it's

22:12

technically possible we don't want it to

22:15

be able to originate connections and

22:17

talk with

22:18

OT we want ot to originate those

22:21

connections and send data to

22:25

it like in this example let's say I do

22:28

want to automate the installation of

22:30

Windows patches on things like my domain

22:33

controllers and my data historians and

22:35

my engineering workstations remember

22:36

those are just Windows servers and

22:38

workstations

22:41

laptops so what we can do is we can put

22:43

that patch management server in the DMZ

22:47

so we configure the OT window servers to

22:50

reach into the DMZ and get patches well

22:53

the patch management server in the DMZ

22:56

it reaches out to the it patch server

23:00

and pulls those in to the network so

23:04

everything's being pulled into the OT

23:06

Network all those connections they're

23:08

originating from OT right OT is going to

23:11

it asking for that data not the other

23:13

way around and if we it's the other way

23:16

around if we allow it to reach into OT

23:20

then if an attacker gets into the IT and

23:22

network and they will they can find that

23:25

route and they can use it to get into

23:27

the OT Network

23:28

we'll talk more about that as we go

23:32

on so even just by simply placing a

23:36

single firewall this is what robley even

23:38

even highlights he talks about in just

23:41

at the the last drus uh Summit uh he

23:45

mentioned this and and I've heard him

23:46

say it

23:47

before most OT they don't need zot trust

23:51

they don't need the the latest Shiny Toy

23:53

I always

23:55

say they just need a fireball

23:58

okay if there's one thing we can do to

24:01

protect OT is to install a

24:05

firewall now obviously there's a lot

24:07

more that we want to do on top of that

24:09

but if we could only do one thing it to

24:11

install a firewall because of this exact

24:15

example so if that attacker sends that

24:17

fishing email and Jennifer Falls for it

24:21

and that say that ransomware takes off

24:23

and starts infecting all the

24:25

windows-based resources in the

24:27

environment

24:28

and it goes to move to the OT

24:31

environment if that firewall is there

24:33

and the firewalls configured

24:35

correctly the ransomware infection is

24:38

stopped from spreading into the OT

24:40

environment we're

24:42

done so yes the it environment has

24:45

burned down to the ground that's

24:48

bad obviously but if we have a truly OT

24:53

resilient

24:55

environment we're still up and running

24:57

right our power plant is still up and

24:59

running and generating power the

25:02

business is running around scrambling to

25:04

try to fix everything so they can get

25:07

back up and running from a business

25:09

perspective but hey you know what we're

25:11

still generating power and that's

25:13

ultimately what matters that we're still

25:15

generating power and we're doing it

25:20

safely you can take the couple days or

25:23

couple weeks to rebuild the it

25:25

environment but we still want to make

25:26

sure that the OT environment is up and

25:28

running and it's doing things

25:32

safely so just by adding that one

25:36

firewall right we've taken a huge step

25:39

in protecting the OT

25:45

environment so let's talk about

25:47

firewalls for for a few

25:51

minutes and so just if you're not

25:54

familiar with a firewall right we

25:56

typically think of a firewall as a a

25:58

hardware Appliance and yes there's

26:00

there's virtual firewalls and and we can

26:03

talk about network based and host based

26:05

right now we're talking about a network

26:08

Appliance physical Appliance and that's

26:10

what we want to focus on for for right

26:12

now and that we're going to put it in

26:16

the network to filter traffic between

26:19

two you see two or more Network segments

26:22

for now we're talking about what we were

26:23

just looking at right having that

26:25

firewall that physical Network Appliance

26:27

between it and OT right filtering

26:29

traffic between it and OT because by

26:33

default that firewall says we're going

26:34

to block all

26:36

traffic between it and OT from it to OT

26:40

and ot to it and then we only come back

26:42

and open up those specific holes in the

26:45

firewall that we need to allow the

26:48

business to

26:50

operate so again if we can we don't want

26:53

to allow any traffic directly from it

26:56

into OT

26:58

remember we want the traffic connections

27:00

to originate from

27:01

OT and allow them to reach out to get

27:05

that

27:08

information there are also the idea of

27:12

OT specific firewalls like mards that

27:16

can understand not only protocols like

27:19

tcpip but other common OT protocols like

27:24

like bom buus the most commonly used OT

27:27

protocol

27:28

and we're going to come back and talk

27:29

about those later really what at this

27:31

level when we're talking about just

27:34

protecting it from OT or OT from it or

27:37

both right we're really just talking

27:39

about common old firewalls that we see

27:41

in in it environments cisos and poo alos

27:46

and

27:50

foret now there's a couple of main

27:53

different types of of firewalls the most

27:56

basic that you see and you see

27:59

originally they started out just as

28:00

simple packet filters and there's a lot

28:02

to unpack here but bear with me for a

28:04

few

28:06

minutes idea with the packet filtering

28:09

this is where you have the firewall and

28:11

it's going

28:12

to filter traffic or either block or

28:15

allow it based off of what we call the

28:17

five tupal and a tupal just means it's a

28:19

list of things so we have this list of

28:22

five pieces of information these are

28:24

five pieces of information that are very

28:26

important from a network intrusion

28:27

detection perspective so we're going to

28:29

come back and talk about this in one of

28:31

the last parts of the the

28:35

course but when we want to allow traffic

28:38

and we want to allow that traffic based

28:41

off of and make it as specific as

28:43

possible so we can say where it's coming

28:44

from and not only just the IP address of

28:47

where it's coming from but but what port

28:49

is it coming from and where is it going

28:51

to so what's the destination IP address

28:53

and again not just the IP address of the

28:56

The Host

28:57

that is going to but the

29:01

port and then what protocol are we using

29:03

in this case we're talking about the

29:05

transport protocol so is it TCP or UDP

29:08

typically icmp is typically the only

29:10

other protocol you'll see when you're

29:13

talking about the five tup at this

29:16

level and that was the most basic form

29:19

of packet or firewall where it's doing

29:22

packet filtering now you can even do

29:24

packet filtering on on switches and

29:25

we're going to come back and talk about

29:26

that layer

29:27

later but for now remember a switch is

29:31

not a firewall and a switch does not

29:34

replace a firewall I've gone into

29:35

environments where I remember going into

29:38

into one environment very large

29:40

environment and and the the gentlemen

29:43

that was in charge of of the site was

29:46

really proud of their new firewall that

29:47

they had installed and I had to break it

29:51

to him it wasn't a firewall it was it's

29:53

a

29:54

switch so switches can some emulate

29:58

firewalls but they're not firewalls and

30:00

they're not designed or meant to replace

30:02

firewalls so we'll come back and talk

30:04

more about that as well now all

30:07

firewalls over the last 20 years

30:09

probably at least do packet filtering

30:12

and stateful inspection the idea is with

30:14

stateful

30:15

inspection it tried to kind act as a

30:18

stop Gap to address an issue where if

30:21

you had a a previous or an older

30:23

firewall that you could have an attacker

30:26

try

30:28

to pretend to be return traffic saying

30:32

you know saying maybe let's maybe using

30:35

a common example would

30:37

be in our it environment we had that

30:39

firewall right between us and the

30:42

internet and we know that the employees

30:45

at that company go out and

30:48

visit websites out on the internet and

30:51

we almost know always right they're

30:52

going to Amazon and Google right there's

30:54

a lot of common websites so what if an

30:56

attacker could SPO Pro one of those IP

30:58

addresses and send packets to the

30:59

firewall pretending to be from let's say

31:03

google.com say hey oh I'm just return

31:05

traffic now it's all spoofed and it it's

31:08

kind of hard to pull off these different

31:09

attacks but it's not impossible so it

31:12

was a way for attackers to slip

31:14

through because the firewall didn't have

31:17

any idea of is that legitimate return

31:21

traffic or not so that's what stateful

31:23

inspection does it watches when your

31:25

employees leave goat when they go out to

31:27

the Internet so it says oh okay Mike

31:30

just went out to google.com and oh

31:31

there's the return traffic from

31:33

google.com if

31:35

there's someone else saying oh I'm

31:37

return traffic from google.com trying to

31:40

go to Mike the firewall says I don't see

31:43

where Mike went to see you at this one

31:45

point in time so I'm just going to drop

31:48

you that's the idea of stateful

31:51

inspection so that's what all firewalls

31:52

do these days the most all firewalls at

31:55

a bare minimum do stateful inspection as

31:57

well as packet filtering we're going to

31:59

talk a lot more about packet filtering

32:01

though now there's also deep packet

32:03

inspection this is where we came up with

32:05

you know NextGen firewalls um at this

32:08

point 10 15 years

32:09

ago the idea where you have a firewall

32:14

that not only right allows traffic in

32:16

and out and we filter it and does

32:17

stateful inspection but it essentially

32:20

acts like an intrusion detection sensor

32:22

and it looks at the

32:24

traffic and it can look for suspicious

32:27

or malicious

32:29

activity and whether it's at the TCP

32:31

level maybe it's an application layer so

32:33

maybe it's it's looking at HTTP traffic

32:36

or FTP traffic to see if there's some

32:38

type of attack there and then a lot of

32:41

times you can also use these to do

32:42

things like like U like URL filtering

32:45

right so don't allow your employees to

32:47

go to gambling websites and and porn

32:49

sites those types of things but deep

32:52

packet inspections really this idea of I

32:53

want to open up the packet and look to

32:55

see is there anything in there that

32:57

looks suspicious or malicious that's

33:00

also what we'll see when we come back

33:02

and talk about the OT specific firewalls

33:04

like an Mard where you can actually go

33:06

in and determine what specific commands

33:11

do you want to allow let's say between

33:13

an HMI and a PLC so if an attacker

33:17

gained access to an HMI and they tried

33:19

to use it for anything other than IL

33:21

legitimate command they would be

33:23

stopped great right so it's another

33:26

reason why we want to implement

33:28

firewalls and the more firewalls the

33:31

better almost always I was in one

33:33

environment they literally wanted to I

33:35

mean Implement hundreds of OT firewalls

33:37

which in a way I really appreciated but

33:41

even in that environment it was Overkill

33:43

so we were able to scale them back a

33:45

little

33:47

bit now this is the idea of an access

33:50

control list that we were

33:52

mentioning so if you've never seen one

33:54

and this is one from a standard Cisco

33:57

firewall looks the same on a firewall or

33:59

switch uh and these are very similar to

34:02

other firewalls so that's why I like use

34:04

using the Cisco

34:07

one now the Cisco ACL always starts with

34:11

at the very bottom even though you don't

34:13

see it actually has what we could talk

34:15

about as far as the the deny IP any any

34:18

so by default if you don't add any of

34:20

these additional Rules by default a

34:22

firewall right will block all traffic

34:25

insane block everything

34:28

but then we come back and we want to

34:30

open up individual

34:36

holes so we can talk about the ACLS here

34:40

but let's jump to the next slide so

34:43

hopefully this will

34:45

help give you a bit better idea of what

34:48

we're trying to accomplish with the ACL

34:50

so you can see the ACL that we were just

34:51

looking at in the upper right hand

34:53

corner and then that little line below

34:56

trying to show you how the the rules are

34:58

written how the acl's are are written

35:00

and then we have our example of our

35:02

power plant with the it and the OT

35:04

networks separated by a single firewall

35:07

so a simple example but it's a great one

35:10

to start

35:12

with so here you can see we have the it

35:15

environment and then down below we have

35:17

a number of control system assets with

35:20

IP

35:23

addresses so bear with me because in

35:27

this environment this is a let's say a

35:30

smaller power plant environment where

35:33

they started cyber security to have a

35:36

fireball and they want to allow certain

35:40

traffic from it into OT

35:43

so we're going to we're going to go with

35:46

the example right it's not ultimately

35:47

what we want to do but it's a it's a

35:49

good

35:50

start so you can see in this example

35:53

right they have the it environment if we

35:55

look at that first ACL

35:58

the ACL is written so that the first

35:59

thing you're going to see is permit or

36:00

deny right are we going to allow the

36:02

traffic or are we going to deny the

36:04

traffic the next thing we're going to

36:06

see is the protocol that we're going to

36:08

either allow or deny so you can either

36:10

see transport protocols like TCP or UDP

36:13

or mention

36:15

icmp which is what we use typically for

36:17

Ping even though it has other uses or if

36:21

you want to just block everything you

36:22

say block

36:25

IP so we can see TCP UDP icmp or IP

36:29

which means all of all of them and then

36:32

we're going to say where is the traffic

36:34

coming from that we either want to allow

36:36

or deny so where is it coming

36:39

from we want to specify the IP address

36:42

in the port and then we're going to say

36:43

where is it going to and then remember

36:45

we want to be as specific as possible we

36:48

want to specify the destination IP

36:50

address and the destination Port now in

36:53

this case the acl's are it's somebody

36:55

just starting off so they don't know to

36:57

make them as specific as possible so in

37:00

this case if I'm looking at that first

37:01

line is saying okay allow right permit

37:05

TCP traffic

37:06

from

37:08

any Source IP address and any port

37:13

basically on the it Network to

37:16

access the device at

37:19

101050 52 over Port

37:23

80 and if you're not familiar Port 80 is

37:25

used for H TP which is the unencrypted

37:29

web

37:30

browsing so more than likely if we

37:32

allowing somebody to talk with an OT

37:35

asset running a web interface is

37:37

probably going to be uh either some type

37:39

of management server where they're

37:41

running an application or an HMI so in

37:45

this case if I look at 101050 152 this

37:48

is my little icon for an

37:53

HMI so they're saying allow anybody in

37:56

it to talk with his HMI over 80 and you

37:59

can see the next line says oh permit TCP

38:02

any host to

38:05

101050 152 on Port 443 right the

38:08

encrypted version of web browsing so

38:11

we're allowing basically unencrypted and

38:13

encrypted web browsing to the HMI

38:18

interface now we can also see there's a

38:21

permit TCP any host 10.10 50150 same

38:25

host on Port 3389 which is used for

38:29

remote desktop which is for Windows host

38:33

so we know not only is this probably an

38:34

HMI but it's also an HMI that runs on

38:37

Windows and that you can remotely

38:39

connect to it with remote desktop so you

38:41

could log into it as if you could

38:43

remotely any other Windows

38:48

machine now the next line actually

38:51

references a different hose so you see

38:54

permit TCP remember coming from any host

38:57

any Source IP address any Source port

39:00

and going to the specific host at

39:02

101050

39:04

154 over Port 502 which is the port for

39:08

modbus so this is almost always just by

39:11

looking at we're allowing mod bus

39:13

traffic almost always going to be a PLC

39:15

and that's the little icon down at 1010

39:18

50154 that represents a a

39:22

PLC and you can see we're going to allow

39:25

icmp to that host as well so somebody

39:27

could ping it even though we might not

39:30

want to allow ping traffic cuz ping even

39:34

ping could potentially bring down an

39:37

older asset now we're talking 20 plus

39:40

years older but it still could happen

39:43

never know but we'll come back and talk

39:45

more about that

39:46

later and then the last thing you can

39:48

see is we allow permit UDP traffic from

39:53

any host from anywhere on the it Network

39:55

to that PLC at 10.10

39:58

50154 on Port 123 which is used for

40:02

Network time protocol with ntp now

40:04

there's other ways we can do time

40:05

synchronization in OT environments as

40:08

well but ntp will see most commonly you

40:12

usually in the windows and Linux and the

40:14

traditional it world so a lot of that

40:16

does creep in into to

40:18

OT so this is an environment where

40:21

they're just starting with security and

40:23

they know they need a firewall which is

40:25

great and they block all the traffic by

40:27

default but a couple things they're

40:29

missing one is they're allowing it to

40:31

talk with directly with ot not the other

40:33

way around right so that's a big

40:35

Miss and they're not being as specific

40:38

as possible with those ACLS and they're

40:40

saying oh just allow anybody in it to

40:43

talk with those resources over those

40:45

ports so again the ACL is somewhat

40:47

limited right it's saying which IP

40:50

addresses we're going to and which ports

40:52

we're going to but it doesn't say well

40:54

what IP addresses and what ports are we

40:55

coming from

40:56

from so remember we want to make it as

40:59

specific as possible so again it's a

41:02

good start there's just a bunch of

41:05

misses and this is very common when I

41:07

work with you especially smaller

41:09

environments that are just starting out

41:11

in in securing their their OT

41:12

environments this can be very similar to

41:15

what you might

41:18

see but we'll have the traffic now here

41:22

we have a slide that just lists some of

41:25

the common ports like we're just going

41:27

through that ACL list and so oh 80 and

41:30

443 and 3389 and 502 and like what what

41:33

do these mean right

41:36

so different ports are commonly

41:38

associated with different services or

41:42

protocols and that's not always the case

41:45

but there are some common ones so if you

41:47

see TCP 21 it's

41:51

99.999% of the time going to be used for

41:53

FTP for file transfer

41:57

TCP 22 is almost always going to be used

41:59

for SSH right to remotely log into that

42:02

host from a a command line perspective

42:05

and be able to make

42:07

changes and and you can see tell then

42:09

the list goes on and on so I try to list

42:13

some of the common it protocols on the

42:16

left hand side that you're going to see

42:18

in OT environments because you'll see

42:20

things like ldb on active directory

42:22

domain controllers or you'll see

42:24

Microsoft SQL Server running on a

42:25

Windows

42:27

server for that's acting as a data

42:29

historian because it's running SQL

42:31

Server which is where it stores all the

42:33

process

42:34

data and then we have some common IC or

42:38

OT protocols as well I put these in

42:41

numerical order so things like S7 mod

42:44

bus I think those are the always the

42:46

easy ones to remember because they're

42:48

kind of similarly numbered right 102 502

42:51

and then you get into other ones like

42:53

cod assist PC Works Omron tmp3 backnet

42:57

for our building management

42:59

systems if you want a more thorough list

43:02

um obviously Google is your friend but I

43:04

also have some cheat sheets in my GitHub

43:06

repository so uh check out GitHub go to

43:10

github.com

43:12

SEC and uh you'll see the uh go to Quick

43:15

Start guides and then there's I have

43:17

Quick Start guides for showd in as well

43:19

as nmap and then they're there for it as

43:22

well as OT so they have more

43:26

comprehensive list for it and and OT

43:30

separately

43:31

so uh but it's a great reference to

43:35

have so we can come and look at there's

43:38

also additional ways that we can

43:40

configure firewalls in the environment

43:42

this again is where we're just talking

43:44

about

43:46

protect I should I always keep saying it

43:49

the wrong way I I it makes me sound bad

43:53

we'll say in this case we're

43:55

implementing secur security between it

43:57

and OT we haven't even got into well how

44:00

do we Implement security within the OT

44:02

environment for now again we're still

44:04

focused on you between it and OT now

44:08

here's where we Implement not just one

44:10

physical firewall but two layers of

44:13

physical firewalls to form that itot DMZ

44:17

or level 3.5 of the expanded Purdue

44:21

model and so the idea then is we have

44:25

that lay layer of communication

44:27

separating the

44:28

two and so as we need to get information

44:33

from the it Network into OT we can pull

44:35

it into the DMZ first and then pull it

44:38

or move it into the OT Network or vice

44:42

vers if we need to push data like

44:43

process data from the data historians we

44:46

can take it from the OT data historian

44:48

push it to the data historian sitting in

44:50

the DMC which can then forward it to the

44:54

data historian in the it Network so the

44:56

business gets the data it needs but we

44:59

never allow the business to reach in and

45:01

gather that information

45:08

itself now there's the idea of the

45:10

triple home firewall I want to talk

45:12

about this real and then we're going to

45:13

go back to the last

45:15

one so remember in well we'll go back

45:18

there already so in the last one we talk

45:20

about a what I consider the true DMZ is

45:24

where we have

45:26

multiple layers of firewall physical

45:28

firewall

45:30

appliances you do see some environments

45:34

and this should only be an IDE and even

45:37

then this is just not a good

45:40

idea in order to usually save money

45:43

right which is not a bad thing but this

45:45

is the one place where you don't want to

45:47

save money you can see they have an IT

45:50

OT DMZ they have an IT Network and an OT

45:52

network but they're only using a single

45:54

firewall to do this so you get a

45:56

firewall appliance that has three or

45:58

more

46:00

interfaces the problem with this is what

46:03

if the attacker can take control over

46:05

the firewall from any of these segments

46:08

especially it if I'm in the it Network

46:10

and I take control over the N the

46:12

firewall then guess what I have access

46:14

to the it DMC and I have access to the

46:17

OT Network done

46:19

deal whereas at least if I'm back in the

46:23

traditional dual homed DMZ

46:27

the whole point of having those two

46:29

layers of

46:31

firewalls is that if an attacker gets

46:35

into the

46:38

environment can they gain access to that

46:41

firewall and this is where we come back

46:42

and say well let's say if they're both

46:44

Cisco ASA

46:47

firewalls and the attacker gets in and

46:50

they have a zero day attack for Cisco

46:52

ASA and remember zero days are not just

46:55

for nation states anymore see ransomware

46:57

groups and other types of attackers have

46:59

zero days as

47:02

well but let's say they have a zero day

47:05

attack for Cisco ASA so then they can

47:07

use that and they can blow right into

47:08

the IT DMC and then they can get right

47:10

into the OT

47:13

Network so the one caveat or the one

47:16

exception we want to make when we talk

47:17

about creating this dual homed DMZ not

47:21

only do we want two layers of physical

47:23

firewalls but we want those layers the

47:26

physical firewalls to be from different

47:29

vendors so let's say I have a Cisco ASA

47:31

layer up top between it and the DMZ and

47:34

then I have let's say for to net

47:35

firewalls between the DMC and OT

47:38

Network so even if an attacker has a

47:43

zero day for Cisco ASA and they can get

47:46

directly into the DMZ they're not going

47:48

to be able to bypass the foret to get

47:50

into the OT

47:52

Network now if the attacker has zero

47:55

days for both they're more than likely

47:57

going to be nation state and there

47:59

nothing you're going to do ever to to

48:01

keep them out 100% so this is one we get

48:05

to a point where we can only do so much

48:07

right before we need to shift our

48:09

priorities right and our resources to

48:12

focus on building up security elsewhere

48:14

like in network security monitoring so

48:17

when the attackers do get into the

48:19

environment then we know about

48:22

it but for now remember if we're going

48:24

to have those two of firewalls do it the

48:26

right way and have them from two

48:28

different vendors companies don't like

48:30

to do that because well you have to have

48:32

people that understand how to manage and

48:34

administer two different types of

48:37

firewalls and they can be different but

48:39

there's a lot of similarities between

48:41

them as well so it's not not the end of

48:43

the

48:44

world but that's the best approach is to

48:47

have two different layers of firewalls

48:49

from two different vendors actually just

48:50

to work with a client uh an outside

48:53

client from uh you know outside of my

48:55

job and they had two different layers of

48:59

firewalls from two different providers

49:03

absolutely impressive it's like that's

49:05

exactly the way you do

49:08

it all right so enough about that all

49:11

right so let's move on so this is what I

49:13

was alluding to earlier now Network

49:17

switches can say pretend to be be

49:20

firewalls because we can add Access

49:22

Control list to to switches

49:26

so that way as traffic comes into a port

49:29

we can say what traffic to allow or not

49:32

just like with a firewall and the nice

49:34

thing again is especially like in the

49:35

Cisco world or others is if you can

49:38

write a ACL for the the firewall you can

49:40

write an ACL for a

49:42

switch and you're just doing it on a

49:45

port by Port basis and this is what

49:46

we're going to want to come back and do

49:48

later on to

49:49

further limit the traffic that's flowing

49:53

between segments and devices within the

49:56

environment we want to do as much

49:58

segmentation as possible we want to do

50:01

as much filtering as possible because

50:03

that's going to limit what the attackers

50:05

can do when they get into the

50:06

environment because they will get into

50:08

the environment it's just a matter of

50:12

time so we want to apply ACLS and

50:16

switches that's great but again what

50:19

does mentioned earlier is we don't

50:20

replace firewalls with

50:24

switches so when we're talking about

50:28

segmenting different networks completely

50:31

one from the other like it and OT we use

50:34

physical

50:36

firewalls if you use a switch to do this

50:40

you're just increasing the chance that

50:42

an attacker a can take control over that

50:44

switch and then they can bypass the

50:46

security

50:47

features so just remember for now a

50:50

switch is not a firewall we're still

50:53

going to leverage all the security

50:54

capabilities of of a switch no doubt but

50:58

switches are not meant to replace

51:06

firewalls so next let's talk about data

51:08

diodes and we can also mention

51:10

unidirectional gateways but data diodes

51:13

have been around forever I remember the

51:15

class I took with Rob Lee he kept you

51:18

know talking about data diodes and that

51:20

they're one of the single greatest

51:23

assets you have in protecting your

51:24

environment and at the same time nobody

51:27

uses them and and that was practically

51:31

10 years ago at this point well little

51:33

little under but yeah it and it's people

51:36

use them even less and less um now they

51:38

have an alternative called a

51:40

unidirectional Gateway or something like

51:42

a unidirectional security Gateway um

51:44

that essentially does the same thing but

51:47

and let's talk about him and you'll see

51:49

what I mean so the idea is a data diode

51:51

is it's an appliance right kind of like

51:54

get a network firewall wall is an

51:56

appliance and you use it to connect

51:57

networks to together but with a data

52:01

diode it's not

52:03

electronics that Gap the the two

52:08

networks that that bridges that Gap I

52:10

should say between the two networks what

52:13

it does it it has a it uses physics

52:16

right it's a it's a a light system to be

52:19

able to transmit data and and it only

52:22

goes in One Direction just like if

52:23

you're sending uh data down

52:27

fiber so it literally can only send data

52:31

in one direction there's no possibility

52:34

of return traffic

52:36

zero so if we're going back to the

52:40

premise that we only want to allow ot to

52:42

send data to

52:46

it a data diode fulfills that that

52:51

need it's the perfect tool to be able to

52:54

do that job I think the problem is these

52:57

days is sometimes they're seeing just as

53:00

older systems that aren't that popular

53:03

right they typically now they tend to be

53:05

a lot more

53:07

expensive than unidirectional gateways

53:10

or or fire

53:11

walls and so I think you just don't see

53:14

a lot of

53:15

people rushing out to buy data diodes

53:18

and Implement them even if they're sold

53:20

on the fact of we only want ot to send

53:22

data to it and not the other way around

53:27

so if you don't want to go down the data

53:29

diode route and invest in those then a

53:32

unidirectional gayway is a you know I

53:37

guess a reasonable

53:39

alternative it's just not as physically

53:42

and and secure from a physics

53:44

perspective right it's it's it's more of

53:46

it's Hardware it's software to and it's

53:49

kind of emulating that hey we're only

53:50

going to allow the OT Network to send to

53:52

it not the other way around right

53:54

unidirectional

53:56

but it's not as secure because it's

54:00

Hardware sof hardware and software can

54:01

be

54:02

hacked so unidirectional Gateway vendors

54:06

drive me crazy because they all talk

54:09

about their Appliance they all talk

54:11

about their appliances this will make

54:13

your network unhackable no it does not

54:16

make the network

54:18

unhackable right it uh allows me to send

54:21

traffic from the OT Network to it and

54:24

and not necessarily hopefully have it

54:26

return traffic to OT but I would not

54:29

consider your gateway unhackable and it

54:31

doesn't make the the network unhackable

54:33

there's lots of other ways to break into

54:35

the

54:36

environment you're addressing just one

54:39

of them one of the big ones sure but

54:42

just one of them so let's kind of calm

54:46

down the marketing rhetoric and and be a

54:49

little bit more

54:50

realistic it's a piece to the puzzle

54:53

right it's a layer of security we're

54:55

going to implement so if we're not going

54:57

to do data diodes let's do a

54:59

unidirectional Gateway yes but we're not

55:01

going to assume that it's going to make

55:04

the network unhackable because it

55:06

doesn't right if anything right somebody

55:08

brings in an infected USB drive and imp

55:10

plugs it into the the

55:12

network the network was just hacked the

55:15

network was just

55:16

compromised so saying a unidirectional

55:19

Gateway makes your network unhackable is

55:22

ridiculous that's that's me how I so so

55:25

I'll get off of it

55:29

now and this was just going back to

55:32

looking at that expanded Purdue model

55:34

and the idea that when we Implement data

55:36

diets let's say again we want to push

55:37

let's say process data from the data

55:40

historians in the it DMZ we want to push

55:43

that into the IT Network we want to make

55:45

sure it's that one-way communication so

55:48

we could do that with a data diode that

55:50

doesn't allow return traffic that's all

55:52

we're just trying to show there

55:56

remember and then boom it's blocked can

55:59

you do that with a firewall sure can the

56:01

firewall be hacked yeah just like a

56:03

unidirectional Gateway data diodes the

56:06

chance of it being com even if it's

56:07

compromise it's still physically the

56:09

path will not allow traffic from it to

56:12

OT period the end that's why it's that

56:15

much more

56:17

secure it's just going to be more

56:23

expensive so most of the rest of the

56:26

part of the section we're not going to

56:27

spend a ton of time on these but looking

56:31

at going back and looking at each of the

56:34

layers of the expanded Purdue model now

56:38

one thing I want you to keep in mind as

56:39

we're going through this and let's kind

56:41

of go back to this one just for a minute

56:43

is that there's a general rule when

56:46

we're talking about designing how

56:48

systems communicate within the Purdue

56:50

model or the expanded Purdue model I get

56:53

in trouble when I just say the Purdue

56:55

model all the all the the older folks in

56:58

OT cyber secur they jump on you because

57:00

you know there was no level 3.5 right

57:02

there was no it DMZ so if I call this

57:06

the Purdue model they literally will

57:08

jump on you that's not the Purdue model

57:09

you don't know what you're talking about

57:11

I've I've had people tell me tell me

57:14

that you know it's I I get it okay the

57:17

expanded Purdue model

57:21

right with the expanded Purdue model

57:25

we have kind of this General saying a

57:27

lot of people will use saying only allow

57:29

traffic to travel up one

57:32

layer and down one layer meaning if I

57:36

have an asset in the like the OT it

57:40

DMZ that data historian that patch

57:42

management server the antimalware server

57:44

I only want to allow it to communicate

57:47

with or have assets communicate to it

57:51

with with it from level four or level

57:54

three

57:55

that's it right it should not be able to

57:58

reach level five it should not be able

57:59

to reach level two or one or zero and

58:02

vice versa level 012 should not be able

58:05

to reach into level

58:07

3.5 remember one up one

58:12

down so if I'm at level two right those

58:15

level two assets they can talk with

58:17

level one assets and level three assets

58:20

but I don't want an engineering

58:21

workstation at level two to be able to

58:23

reach into the DMC right it's two levels

58:27

up to one level up one level down and

58:31

that's a pretty good general rule of

58:34

thumb I think for most people to get

58:39

started all right I'm now I'm going all

58:41

the way crazy so those upper levels and

58:45

I just I just for me I just group them

58:47

all together right the idea that okay we

58:50

have level four level four we have the

58:51

it back office or the Enterprise it

58:54

Network right again that's where

58:55

everybody does work they're sitting at

58:57

their desk they're browsing the internet

58:59

they're reading their emails they're

59:00

using their applications they're putting

59:02

data into

59:05

Excel again the most danger there is

59:08

that's the internet connection and

59:09

that's where we're receiving the emails

59:11

and bringing in the fishing attacks that

59:13

our users are going to be or our

59:14

employees are going to be

59:16

clicking but at least we have that an

59:18

enforcement boundary slf

59:20

firewall between it and the rest of the

59:23

OT Network including the the

59:28

DMZ and then we can

59:30

still it's I just grew four and five

59:33

together but if you get back into the

59:35

well one up and one down any resources

59:38

that are talking with the DMZ should be

59:40

in level

59:44

four so now the rest of the everything

59:47

else is the OT Network right the

59:50

DMZ and then the rest of the the OT

59:56

Network

59:58

itself and then we can break these down

60:01

one at a

60:03

time so again we have the the

60:06

DMZ so that way if we do want to allow

60:10

communication between OT and it and in

60:12

most environments there's some need for

60:14

some type of communication okay IDE

60:17

ideally we can push the data from OT

60:19

into it and we don't allow it to talk

60:22

excuse me with the OT

60:26

manufacturing is the one environment

60:27

where it's almost you have it talking

60:30

with with OT unfortunately because of

60:33

the way the manufacturing uh systems

60:36

work but most others thankfully that

60:39

I've been involved in I've been involved

60:41

in a lot of different projects some very

60:43

large environments that you can

60:45

configure it so only OT sends to

60:49

it there are some small exception we're

60:51

going to get to those in in a little bit

60:57

but we've talked about the DMC remember

60:58

we have those two layers of firewalls

61:00

ideally from different vendors to

61:02

protect those resources that are sitting

61:04

in the DMC and that we use those as

61:05

those those kind of middle pivot

61:08

Points or transfer points to get data in

61:12

and out of the OT Network so if I want

61:16

those patches or anti-malware updates I

61:18

go to the DMC to pull them into the OT

61:22

Network and they got on those servers

61:23

because they pulled them from the it

61:26

Network the it Network never initiated

61:28

any of those connections they were all

61:31

initiated from the OT

61:33

side to get data historian our process

61:36

data to it we push that right we push it

61:40

into the DMZ from the OT data historians

61:43

and then we push it into the IT Network

61:45

so the business gets its processed data

61:48

we give it right we pushed it there we

61:50

put it there they didn't come get it and

61:52

if we allow them to come get it that

61:55

while they're legitimate an attacker is

61:57

going to get into the environment and

61:58

use that same process against us so

62:01

that's why we don't want to allow it to

62:03

open up communication with us we open up

62:06

communication with

62:09

them that's basically what we're saying

62:14

here so now there's level three for site

62:17

operations and it get the Purdue model

62:19

also an expanded Purdue model talks

62:21

about not only what if I have a single

62:23

site but with managing managing multiple

62:26

sites how does that come into play so in

62:29

this case we're just talking about a

62:31

single site but so there's site

62:33

operations so right we're we're doing

62:36

what we do in OT and when we say like

62:38

scada right supervisory control and data

62:41

acquisition the whole point is be able

62:43

to acquire the data so we're monitoring

62:46

what's going on in the

62:47

environment and then we have the ability

62:49

to control to make changes to the

62:52

process if we ever see an alert or get

62:57

some other reason why we need to make a

62:59

make a change so we're monitoring and

63:01

control so ultimately a lot of what

63:03

we're talking about in security is

63:06

ensuring

63:08

visibility into the

63:11

process and the ability to control the

63:14

process fundamentally those are the two

63:17

things that we're trying to secure which

63:19

then gives us the ability to ensure

63:22

physical security environmental security

63:23

and operations or availability but we'll

63:26

come back and talk more about that

63:30

later at these higher levels and this is

63:32

what we've already been talking about

63:34

when I look at yeah the engineering

63:36

workstation hey that's running Windows

63:38

10 or Windows 11 that data historian

63:40

it's Windows Server active directory

63:43

definitely running Windows server that

63:45

operations management server more than

63:47

likely going to be running on Windows

63:49

Server all windows

63:52

servers which is great from an ease of

63:54

administration perspective which is why

63:56

they're there that's why we see Windows

63:58

everywhere in OT because we can transfer

64:01

it skills to make it easier to

64:04

manage but if we're making it easier to

64:06

manage we're also making it easier to

64:09

hack into right to break into and

64:12

compromise and that if we don't have

64:14

those boundaries remember in our example

64:16

without secure network architecture that

64:18

ransomware infection went right in from

64:20

it right into

64:23

OT

64:26

and it happens like that because all of

64:28

those higher layer assets that are

64:31

directly exposed essentially to the it

64:35

Network they're all running Windows if

64:38

we go down another level guess

64:42

what almost everything there is running

64:44

Windows as

64:46

well that engineering work another

64:49

engineering workstation whether it's a

64:50

laptop or a workstation it's running

64:52

Windows 10 or 11 or mayy

64:55

older that data historian is running

64:57

Windows server with SQL

64:59

Server that HMI it's probably running

65:03

Windows the ska now that might be some

65:06

type of proprietary system like with I

65:08

was just you know thinking back to the

65:11

the mandian announcement about the the

65:12

latest Ukrainian power outage where the

65:14

Russians had broken in and used live

65:17

living off the land techniques to move

65:18

through the window systems to then move

65:20

to the one micro skada server that they

65:22

used to turn the power off to flip all

65:25

the

65:26

breakers so almost everything at level

65:29

two once again it it's almost always

65:33

windows so it's going to become

65:35

important for a number of reasons

65:37

especially when we talk about Network

65:38

intrusion detection and things like

65:39

vulnerability management and and asset

65:43

registers which we're going to get into

65:45

in the the next part of the

65:48

course but again think almost all

65:51

windows almost all

65:54

then when you get to level one zero this

65:56

is where Windows drops off and it's

65:58

almost never going to be seen so this is

66:02

where we see our

66:03

plc's that are running typically their

66:06

own custom operating systems now some of

66:09

them I I have one over here from Phoenix

66:11

contact that it runs

66:14

Linux and it runs Linux just like any

66:16

other Linux so you can install all the

66:19

Linux based attack tools on it if you

66:21

wanted to so you could turn that PLC

66:23

into quite the little attack platform

66:26

which which which I've done for the home

66:29

lab of

66:32

course and then things like rtus and and

66:35

IEDs and in power environments but and

66:37

then the sis remember the safety systems

66:39

are fail

66:40

safes right those controllers that we

66:43

use to create the back the fail safe

66:44

backup to be able to shut down the

66:47

environment if we ever detect that fault

66:50

condition that's going to present any

66:53

risk to physical safety environmental

66:55

safety or you people or to the

66:59

operations of the

67:01

plant and then finally we oop sorry we

67:04

have things like our sensors and

67:06

actuators right the things that we

67:08

moving out in the real world the things

67:10

we're

67:11

controlling with our controllers like

67:14

the

67:15

plc's but no windows there right no no

67:19

windows at level

67:22

one but level two mostly Windows level

67:26

three all windows typically level 3.5

67:30

almost all windows maybe you have a

67:31

little Linux in there everyone it's

67:33

almost all

67:34

windows so when we go back and look at

67:37

this overview of the OT Network level

67:40

3.5 level three level two almost all of

67:44

that is going to be all windows space

67:47

and then when you get down to level one

67:48

two the true when we talk about Control

67:50

Systems that's where Windows

67:52

dramatically drops off

67:54

not that we don't have Windows based

67:56

plcs but that's where some people I

67:59

think they draw the

68:02

line but it's becomes important because

68:04

when we look at think of from an

68:05

intrusion detection perspective or the

68:07

attacker perspective it's windows I know

68:10

how to break into the IT environment and

68:12

move from Windows to Windows machine so

68:15

yeah I can just do that just as easily

68:16

in OT until I get down to things like

68:19

the plcs that's where the attackers slow

68:21

down and we'll talk about why that

68:22

happens coming

68:25

up what about things like vulnerability

68:27

scanning or penetration

68:30

testing there usually that General well

68:32

we don't do those things in the OT

68:35

environment because we don't want to

68:36

crash the systems like well we might

68:39

worry about level one and level zero

68:41

maybe that ska system at level two but

68:45

if everything else is running windows at

68:47

least maybe from a vulnerability

68:49

management perspective scan

68:52

away maybe maybe that's again that's a

68:55

conversation for the vulnerability

68:57

management unit so I'm I'm jumping ahead

69:00

again but just to kind of plant those

69:02

seeds so it's not just as cut and dry

69:05

because now we have most of the

69:08

environment is

69:13

Windows and so that's where we get a lot

69:15

of this crossover between it and OT or

69:17

we talk about it and OT conversion which

69:19

is the most you know risk that we have

69:22

in OT is from it OT

69:24

convergence all right but to get us back

69:26

on track let's kind of jump back ahead I

69:28

don't want to go too far down that

69:29

rabbit

69:31

hole all right so earlier we talked

69:34

about we kicked off the section talking

69:36

about the it kill chain from Lock Mar so

69:40

Micha lante which we again we really

69:42

think of as the the man who and there

69:46

were others as well but Michael Sante

69:48

really was the spearhead of really

69:50

creating this idea of this field of OT

69:53

cyber security I think they didn't even

69:55

have the term probably OT back then so

69:57

right or industrial Control Systems uh

70:01

cyber

70:02

security and uh of course

70:05

his um right-hand man was was Rob Lee so

70:11

they created the idea of the IC kill

70:14

cyber kill chain so what they're trying

70:16

to demonstrate here and you see the the

70:17

link in the paper is there if you tilt

70:19

your head and you can you can read it

70:22

but the idea is you

70:24

there's the normal proc process that the

70:27

attackers use to get into the IT

70:31

environment right and then okay they're

70:34

in the it

70:35

environment but if their ultimate goal

70:38

is the OT Network well then then they

70:41

have to get into the OT Network so they

70:43

basically have to start all over again

70:46

now remember one thing we were just

70:47

highlighting is that a lot of the OT

70:50

network is made up of those Windows

70:52

systems that well if an attacker got

70:55

into it Network because it was Windows

70:57

and they mooved probably pretty quickly

70:59

through the environment because it was

71:00

windows they're probably going to do the

71:02

same thing with the windows environment

71:05

on the OT side if we haven't done secure

71:10

network architecture appropriately so

71:12

that's where we talk about having

71:14

multiple layers of firewalls for the DMZ

71:16

different providers don't even allow

71:19

connections to originate from

71:22

it right only OT talks with it not the

71:25

other way around if we can do that we

71:28

vastly address the vast majority of the

71:31

risk in the environment it's as easy as

71:33

that again not all environments can

71:36

Implement that like we mentioned

71:38

manufacturing earlier just the way those

71:40

Manufacturing Systems

71:42

work

71:44

sadly but a lot of other Environ I've

71:47

been in environments where they have it

71:49

talking with OT when there's no reason

71:51

to and I see the these large or huge

71:55

holes in the firewalls and it's like

71:57

why and it's just because we don't know

72:00

better and I I completely understand

72:03

that's a big part of why I'm here right

72:04

so so hopefully we can get that

72:06

education and awareness out there so

72:08

people do know

72:11

better but take the time to go back and

72:13

and read the the IC cyber kill chain

72:17

anything Rob uh writes anything Michael

72:20

had written I definitely read um there's

72:23

some great videos if you go back to um

72:26

that not only Rob but micheel have you

72:28

know out on YouTube you know a lot of

72:29

the older stuff that Mike had done um

72:32

especially with his government work

72:33

there's a lot of his kind of lectures or

72:35

speeches and like to different groups in

72:37

government and Military um that I find

72:39

really really fascinating because again

72:41

that's really it's where it all started

72:44

and when we talk about things like

72:45

operation Aurora that that he led to

72:47

demonstrate right how computer code

72:50

could be used to blow things up in the

72:53

world in that case it was just a

72:54

generator out in the middle of nowhere

72:57

but they used computer code to blow

73:00

something up in the middle of nowhere it

73:02

can be

73:04

done so definitely check that out all

73:07

right now we also talked about in the

73:10

very first section or our history of at

73:14

least the annotated history of of events

73:17

in

73:18

IAS um over the years when we talk about

73:20

different incidents so tcis was the one

73:23

where the Russian nation state had come

73:26

in and into that petrochemical facility

73:30

that was run by saudio

73:34

rco and they gained access pretty much

73:37

the entire environment including they

73:38

had complete control over the

73:40

DCS and remember they had

73:43

99.99% control over the sis the the

73:47

safety system right the backup so those

73:50

trionics controllers that's where the

73:53

tcis comes from right try from trionics

73:56

and then the safety instrumented systems

73:58

or the fail safe

74:01

backup so with tcis one of the things

74:04

that that really sticks out well there

74:07

were a couple things there a lot of

74:09

lessons to be learned from tcis one was

74:12

the the sis those controllers they were

74:14

connected and exposed to the rest of the

74:15

network so there was no network

74:17

segmentation your your sis controllers

74:20

should never be exposed to the rest of

74:22

the network they need to be air gapped

74:24

and they need to be island off if you

74:26

want to send their processed data to

74:29

something like your sock that's great

74:30

put in a data diode and only allow it to

74:33

send traffic to your sock no one should

74:36

remotely be able to connect to your sis

74:38

ever period the end so that way if

74:41

somebody was going to compromise it they

74:43

would physically have to be on

74:46

site and then the other idea and this is

74:49

where my thesis comes in is where we

74:52

talk about

74:55

that most controllers have a key switch

74:59

I think we've already talked about this

75:00

right and then idea is if the key switch

75:02

was in Run mode then it would

75:06

prevent a remote attacker from being

75:08

able to make changes to code running on

75:11

the

75:13

system so the controller had its key

75:15

switch left in program mode which

75:17

allowed the attackers very easily to

75:19

make changes now you'll see this is

75:21

getting into my my m cees been talking

75:24

about where you there's this general

75:26

statement where we say oh if you put the

75:27

key you know key switch into run mode

75:30

then no changes can be made to the the

75:34

controller whether it's sis or PLC or

75:36

whatever type of controller and that's

75:38

not necessarily the case in some sure

75:43

but in a lot like the the plc's that I

75:46

tested for my master CA which are more

75:49

commonly used plcs in environment

75:51

especially in the United States or North

75:54

America there were a lot of bypasses

75:56

that existed so even though it was in

75:58

Run mode that if you went to update the

76:02

PLC programming code it would let you it

76:05

would actually say hey do you want to

76:06

bypass essentially the the safety

76:09

feature of run mode to upload your your

76:13

new programming code and you be like yes

76:15

sure you had to know what the password

76:17

was to log into the PLC but that was it

76:21

you say yes and your codes up running

76:24

now for most if you want to upgrade the

76:26

firmware that's where you have to take

76:28

it out of run mode so that that's

76:29

definitely a positive step in the right

76:30

direction so it's still important to

76:32

monitor for when plc's come out of run

76:35

mode or even more importantly to monitor

76:37

for when code changes are made on PLC so

76:40

you can investigate to make sure was

76:42

this an authorized legitimate change or

76:44

is this something more susp suspicious

76:47

or malicious in

76:49

nature but but that was remember tcis so

76:53

if anything with those controllers right

76:55

key switch is good right still not

76:58

bulletproof but still step in the right

77:00

direction right monitor for those

77:02

changes investigate when you need

77:06

to and then also make sure to isolate

77:09

the sis from the rest of the network

77:11

don't allow a remote attacker to be able

77:13

to connect to it if you're going to come

77:15

for it you have to be on sight and you

77:18

have to know where it is so come for it

77:22

and hopefully we'll have guards on guns

77:24

protect with guards or guards with guns

77:26

protecting it so and really big

77:29

dogs all right so the next part of the

77:32

section we're going to talk about

77:35

is how do we further break down or

77:38

segment or protect the OT Network itself

77:42

right so we've talked about the DMC that

77:44

we're going to put in between it and OT

77:46

maybe even a data diode or

77:48

unidirectional Gateway but now we're

77:51

looking at let's just focus on the OT

77:53

Network and this is where we go to

77:56

62443 so 62443 talks about the idea of

78:00

zones and conduits

78:03

which for me I admittedly it was just

78:06

kind of weird terms but I completely see

78:09

how Engineers came up with

78:12

this the idea is a Zone if you're coming

78:15

from the IT world think of the the zone

78:19

for now for now and it's not perfect but

78:22

for now as a

78:23

subnet and that in that

78:26

subnet we're going to put in all the

78:30

assets that share a common function like

78:35

we're just talking about the sis so let

78:37

say I have these these four different

78:39

controllers that make up the sis I'm

78:41

going to put them in this one zone and

78:45

actually we'll we'll add uh some data

78:47

diodes right also so they're in this

78:51

zone I'll have another zone for things

78:54

like maybe my control systems for uh how

78:57

about fire suppression right that's also

79:00

another very important Zone to

79:03

have so we have these different zones

79:06

that represent all these

79:09

different processes in the

79:13

environment and then we can allow

79:16

communication between those

79:19

zones again we might want to allow the s

79:22

has to send data to maybe a sock right

79:26

but that's it so it can push data to the

79:28

sock but that's it that's the only

79:30

communication that communication is what

79:33

they call a conduit so any communication

79:35

between zones is called a

79:37

conduit so with 62443 one of the things

79:40

we do to implement security is conduct a

79:43

risk assessment well that risk

79:46

assessment is based off of looking at

79:49

the zones right seeing how the assets

79:52

are grouped

79:53

and again usually those are grouped

79:55

within the same subnets which makes it

79:58

really easy uh to work with if they're

80:01

not still not impossible but it's just

80:03

not as nice and that between those

80:06

subnets any communication right if we're

80:08

allowing this case this engineering

80:10

workstation on the right hand side maybe

80:12

this is like the operations um

80:15

Zone that we're allowing that works that

80:19

uh engineering workstation to connect to

80:21

the plc in the the second Zone on the

80:24

left hand

80:26

side right that communication from the

80:29

engineering workstation to the PLC

80:31

that's a conduit so we do those risk

80:35

assessments we document all of the zones

80:37

all the assets and those zones to

80:39

understand everything that happens in

80:41

each of those zones and then also the

80:43

conduits between them which people

80:46

always hate because they typically they

80:47

might have the zones and the assets

80:50

documented but they probably don't have

80:53

the conduits documented but you need to

80:55

get them down to be able to review them

80:58

so usually you can pull those out of um

81:01

you firewall configs if you have

81:03

firewalls between them if it's maybe uh

81:05

just switch connectivity so you might be

81:07

able to pull that data out of the the

81:12

switches and things like net flow data

81:14

which can show us you know what systems

81:17

are talking with what other systems

81:19

without having to go onto each of those

81:21

systems and log in in and look at try to

81:23

see if they have something like an ARP

81:25

table or other way to be able to look at

81:27

network communication like netstat like

81:30

a lot of these devices have something

81:31

equivalent to to netstat like the I was

81:34

just playing with a a seaman PLC the

81:37

other day and it even exposed netstat

81:39

over SNMP just like in the windows world

81:42

it's like oh this is horrible not

81:47

secure that's the idea of zones and

81:49

condu we're going to come back and talk

81:51

about that more because we have a later

81:53

section talking more specifically to 62

81:56

443 and my next class that I'm going to

81:58

build it's all dedicated to

82:00

understanding 62443 because it's one of

82:03

those where it's just like this to me

82:05

and I think a lot of people at least

82:08

those from the IT background I I think

82:11

feel the same same way the ones I talk

82:13

with the people I talk with because it's

82:15

like you get this great standard or

82:17

Frameworks like here's all the

82:18

information you need to secure your OT

82:21

environment that's awesome it's like

82:23

this is what I've been looking for my

82:24

entire

82:25

life but then you start reading it and

82:29

the way they've structured it and the

82:31

parts and it's just to me there's like

82:32

no logical flow and there's no it's not

82:35

explained from a practical perspective

82:38

it's like I just want to know how to do

82:41

it and the way it's structured and laid

82:45

out just doesn't I again for me it just

82:48

does not work so so the the class I'm

82:52

writing now now is this is the Practical

82:55

approach to implementing a cyber

82:58

security program based off of 62443 I'm

83:00

really excited about that because I

83:02

think it's it's really key I think it'll

83:04

it'll help a lot of people

83:08

so so again we're going to come back

83:10

actually I think this is the wrong unit

83:11

number so don't quote me on this so but

83:14

it might it might actually be right uh

83:16

but we are going to come back and talk

83:17

more about zones and and conduits when

83:19

we get into 62443 when when we talk

83:22

about the risk assessments piece so so

83:24

we'll look at those so uh real quickly

83:27

let's talk about the the industrial

83:29

internet things and we'll come back and

83:30

talk about this later on as well but I

83:32

did want to introduce it here because we

83:34

are talking about secure network

83:37

architecture with I iot this is a term

83:40

that came from General Electric and the

83:43

idea what the way it was explained to me

83:45

was GE whether you know and actually

83:48

live not too far from the plant here in

83:49

Greenville South Carolina and one of the

83:51

gentlemen um that runs cyber security

83:54

for GE for many of the product lines is

83:57

is here in in Greenville he knows

83:58

everybody in the the OT cyber security

84:01

World hi Rob and

84:04

so the oh Gary actually but anyways long

84:10

story

84:11

there um the idea is that GE were and

84:17

two of the things that they built here

84:19

in Greenville one are the big turbines

84:21

for for wind farms and locomotives so if

84:27

you have a train locomotive and they say

84:30

you wrap it in thousands of sensors to

84:33

do things like Predictive Analytics to

84:35

understand you know when parts are going

84:37

to break down where they need to be

84:39

replaced ahead of time so we don't cause

84:41

any accidents right you can have

84:44

predicted analytics even on you you

84:47

understanding how fast the the train is

84:49

traveling is it traveling too fast

84:51

there's a curve coming up right does the

84:52

train need to slow down what if the

84:54

conductor had a heart attack and they're

84:55

flying they're laying on the floor of

84:57

the the

85:00

the the

85:02

train right so being able to take all of

85:05

this data into

85:06

consideration the idea is you need a a

85:08

powerful computer to be able to Crunch

85:10

those numbers and you're not going to be

85:11

able to fit that computer on the

85:15

train but you can take that data and you

85:18

can push it up to the

85:20

cloud and you can crunch the numbers

85:22

there and make that information

85:23

available in the

85:26

cloud so it's awesome that you can get

85:28

all this data that we can have all these

85:29

sensors we can get all this data up to

85:31

the cloud but in that case we're

85:33

connecting our OT environment in this

85:35

case the train to the internet to get to

85:38

those cloud-based

85:39

servers so it's another one of those

85:42

where we don't want to tell the business

85:43

no you can't do this because they're

85:45

going to do it anyways but we want to

85:47

make sure we say okay let's do this as

85:49

securely as

85:51

possible

85:52

so that's where we started thinking of

85:54

things like well data diodes or

85:56

unidirectional gateways or a firewall to

85:59

be able to allow the OT Network to send

86:01

to the cloud in oneway fashion but not

86:05

allow return

86:07

traffic that can make it as secure as

86:09

possible so it's a win-win for everybody

86:12

it's secure right we don't allow

86:14

attackers coming in from the internet to

86:16

talk directly to the OT

86:18

environment whether it's the train or a

86:20

wind farm or power plant

86:23

but we get that data out in the cloud

86:26

and then someone can pull up a dashboard

86:28

and realize when they need to do

86:29

maintenance and replace certain parts on

86:31

the train because we don't need any more

86:33

derailments we have way too many

86:35

derailments as it is that could have

86:37

been prevented by things like

86:39

maintenance so that's the quick

86:42

introduction to Industrial internet of

86:44

things which again we're going to come

86:45

back and talk about later

86:47

on oh and this is just the the idea

86:50

right if we're sending data out to the

86:52

cloud we want to make sure that

86:54

attackers aren't able to come

86:57

back so remember we make sure that it's

87:00

an outbound only connection we do not

87:03

allow the data to come back so we're

87:05

just providing the data for readon

87:10

capabilities in most environments and

87:12

then that's changing as well so more to

87:16

come and then the last thing to talk

87:17

about secure remote access okay we're

87:19

going to come back and talk about this

87:21

more in detail later on but again just

87:25

to wrap up the secure network

87:26

architecture because it is part of that

87:28

conversation the idea that we want to

87:32

allow remote parties to come in even

87:35

before covid but when Co hit it was hey

87:38

everybody is doing everything remote

87:39

almost

87:41

always so we want to be able to I

87:43

remember my first site right I was like

87:45

yeah we want to allow remote connection

87:47

because Sam's the only person that can

87:49

come and do maintenance on these systems

87:51

and Sam lives on the other side of the

87:52

country and we don't want to spend

87:54

$4,000 in travel cost every time we

87:56

bring Sam out we'd rather put in a VPN

88:00

Appliance and give him access

88:02

remotely can we do that yes now we're

88:07

going to add some more security features

88:09

to that things like jump boxes and

88:11

OnDemand access and multiactor

88:12

authentication to make it more

88:15

secure because again you're not going to

88:17

have a choice the business is going to

88:19

do it but so we want to take the

88:22

opportunity to work with them and do it

88:27

securely the other reason why we like

88:30

people doing work remote is they don't

88:32

have to be on site where they're in

88:34

danger

88:36

potentially so it's much safer to be

88:38

able yeah to make those changes from the

88:41

other side of the

88:44

country again we just have to remember

88:46

and we have to always assume no matter

88:48

how secure we make it it is going to be

88:51

compromised and it is going to be used

88:52

against us so we have to prepare for

88:54

that and understand how are we going to

88:56

detect it and how are we going to jump

88:59

into action to respond when it does

89:02

happen because it will

89:04

happen one way or another one

89:09

day and with that that's the end of the

89:12

section so I appreciate everybody for

89:15

connecting in if you like the video if

89:17

you can like it on YouTube if you

89:19

haven't subscribed to the channel if you

89:20

subscribe I appreciate apprciate that uh

89:23

and uh if you haven't already you find

89:25

me on LinkedIn uh and reach out to me

89:27

there so again thanks and I'll see you

89:29

soon for uh part

89:31

five

UNLOCK MORE

Sign up free to access premium features

INTERACTIVE VIEWER

Watch the video with synced subtitles, adjustable overlay, and full playback control.

SIGN UP FREE TO UNLOCK

AI SUMMARY

Get an instant AI-generated summary of the video content, key points, and takeaways.

SIGN UP FREE TO UNLOCK

TRANSLATE

Translate the transcript to 100+ languages with one click. Download in any format.

SIGN UP FREE TO UNLOCK

MIND MAP

Visualize the transcript as an interactive mind map. Understand structure at a glance.

SIGN UP FREE TO UNLOCK

CHAT WITH TRANSCRIPT

Ask questions about the video content. Get answers powered by AI directly from the transcript.

SIGN UP FREE TO UNLOCK

GET MORE FROM YOUR TRANSCRIPTS

Sign up for free and unlock interactive viewer, AI summaries, translations, mind maps, and more. No credit card required.