WEBVTT

00:00:00.752 --> 00:00:04.586
So software security and security in the software development lifecycle.

00:00:04.586 --> 00:00:06.967
The end of requirement 6.

00:00:06.967 --> 00:00:12.670
By now, you know, many people and thinking, gosh, there's a lot in DSS.

00:00:12.670 --> 00:00:15.855
The PCI SSE has been talking about this for an awful long time.

00:00:15.855 --> 00:00:16.306
In fact,

00:00:16.306 --> 00:00:18.947
they're developing a whole new set of standards about

00:00:18.947 --> 00:00:20.889
software development for payment cards,

00:00:20.889 --> 00:00:25.843
which we'll expect to see coming in the future,

00:00:25.843 --> 00:00:28.334
but you know,

00:00:28.334 --> 00:00:31.605
if vulnerabilities of packaged applications and operating

00:00:31.605 --> 00:00:35.344
systems is the number one cause of compromises,

00:00:35.344 --> 00:00:38.156
poor software development is the second cause of things.

00:00:38.156 --> 00:00:39.116
Easily.

00:00:39.116 --> 00:00:41.442
I still can't believe I still read of SQL injection attacks happening to people.

00:00:41.442 --> 00:00:41.770
Indeed.

00:00:41.770 --> 00:00:43.562
So this is still really important.

00:00:43.562 --> 00:00:45.026
We haven't fixed this as an industry yet.

00:00:45.026 --> 00:00:45.943
No, we have not at all.

00:00:45.943 --> 00:00:49.509
I’m going to take 6.3, which is software development generally,

00:00:49.509 --> 00:00:53.080
and then 6.5, which is web-facing software development together,

00:00:53.080 --> 00:00:57.712
and let's talk about the stuff that goes wrong in organizations,

00:00:57.712 --> 00:00:59.938
the stuff that you're looking for as an assessor.

00:00:59.938 --> 00:01:01.542
So firstly, you know,

00:01:01.542 --> 00:01:05.970
we know how to do this stuff now; this is not rocket science anymore.

00:01:05.970 --> 00:01:06.959
It's not.

00:01:06.959 --> 00:01:08.881
There's lots of great courses on Pluralsight about

00:01:08.881 --> 00:01:09.947
doing secure software development.

00:01:09.947 --> 00:01:12.867
Microsoft really worked at how to do this a while ago.

00:01:12.867 --> 00:01:16.124
So what are the barriers to organizations doing this well?

00:01:16.124 --> 00:01:16.635
It's hard.

00:01:16.635 --> 00:01:19.345
I mean, I'd say it's known; we know how to do this,

00:01:19.345 --> 00:01:20.742
we have a lot of training,

00:01:20.742 --> 00:01:24.490
there's a lot of sort of principles of good software practice,

00:01:24.490 --> 00:01:25.734
but it's hard.

00:01:25.734 --> 00:01:26.786
It's still a lot of effort.

00:01:26.786 --> 00:01:29.558
You have to train developers.

00:01:29.558 --> 00:01:32.915
You have to essentially focus the organization's software

00:01:32.915 --> 00:01:35.215
development effort on software security.

00:01:35.215 --> 00:01:36.898
It doesn't arise organically.

00:01:36.898 --> 00:01:39.347
It doesn't matter if you've got a bunch of really smart programmers,

00:01:39.347 --> 00:01:41.640
if you've got a really good software engineering team,

00:01:41.640 --> 00:01:46.680
if they are not making software security a major goal,

00:01:46.680 --> 00:01:48.287
it will not happen.

00:01:48.287 --> 00:01:50.550
Okay, so it's not going to happen organically is what you're saying?

00:01:50.550 --> 00:01:51.487
No, not at all.

00:01:51.487 --> 00:01:55.686
So how do you make it, but it's organic, inorganic, how do you make it?

00:01:55.686 --> 00:01:57.971
Well, structured, intended, right?

00:01:57.971 --> 00:02:01.308
It's got to be a major focus for the organization, right?

00:02:01.308 --> 00:02:05.777
And that means it has to win against, you know,

00:02:05.777 --> 00:02:10.612
other factors like time to market or you know,

00:02:10.612 --> 00:02:13.485
ease of development,

00:02:13.485 --> 00:02:17.527
or just sort of the natural inclination to do kind of the lazy thing.

00:02:17.527 --> 00:02:18.760
So you're going on site,

00:02:18.760 --> 00:02:21.778
you're going to an organization that writes their own software,

00:02:21.778 --> 00:02:24.762
that stores, processes, or transmits payment card data,

00:02:24.762 --> 00:02:29.146
often an e-commerce website nowadays, or point-of-sale software.

00:02:29.146 --> 00:02:30.056
How do you start that assessment?

00:02:30.056 --> 00:02:31.427
What are you looking for?

00:02:31.427 --> 00:02:33.536
So ideally, by the time I've gone on site,

00:02:33.536 --> 00:02:35.959
I've gotten a hold of whatever their SDLC is,

00:02:35.959 --> 00:02:38.227
the actual policy documents, right?

00:02:38.227 --> 00:02:40.863
I like reading the policies, you know, on my tablet,

00:02:40.863 --> 00:02:45.498
on my couch, watching Game of Thrones, rather than in their conference room,

00:02:45.498 --> 00:02:45.966
right?

00:02:45.966 --> 00:02:47.886
So ideally we have that, we've read it,

00:02:47.886 --> 00:02:50.678
and we can immediately start asking intelligent

00:02:50.678 --> 00:02:53.126
questions of their development team.

00:02:53.126 --> 00:02:56.237
Tell us about your software development practice,

00:02:56.237 --> 00:03:01.824
tell us about your software security practices,

00:03:01.824 --> 00:03:03.267
right?

00:03:03.267 --> 00:03:06.728
And you know, this is the one where nobody's a very good liar about it,

00:03:06.728 --> 00:03:11.125
and so if they're not very good at it, it tends to show up very quickly,

00:03:11.125 --> 00:03:11.448
right?

00:03:11.448 --> 00:03:18.730
And frankly, the easiest tell is tell me about your code reviews, right?

00:03:18.730 --> 00:03:22.001
The 6.3, 6.3 .2 talks about doing code reviews,

00:03:22.001 --> 00:03:25.909
doing security reviews for code that's developed,

00:03:25.909 --> 00:03:26.277
right?

00:03:26.277 --> 00:03:29.271
Making sure that it gets reviewed for vulnerabilities and corrected

00:03:29.271 --> 00:03:32.516
before it moves on in the process for release.

00:03:32.516 --> 00:03:35.938
And very often organizations that are new to this or have a

00:03:35.938 --> 00:03:37.352
very immature software security practice,

00:03:37.352 --> 00:03:43.007
their code review is we gave it to a smart guy and he said it looked okay.

00:03:43.007 --> 00:03:43.208
Okay,

00:03:43.208 --> 00:03:45.479
but I'm just looking at the standard here to see

00:03:45.479 --> 00:03:46.927
what the testing requirement is.

00:03:46.927 --> 00:03:49.398
It doesn't say, give it to a smart guy.

00:03:49.398 --> 00:03:50.280
No, it doesn't.

00:03:50.280 --> 00:03:51.358
No.

00:03:51.358 --> 00:03:52.919
It says, you know,

00:03:52.919 --> 00:03:57.726
have somebody who understands application security and secure coding techniques

00:03:57.726 --> 00:04:00.858
do the review and make sure it's not the originating author,

00:04:00.858 --> 00:04:02.510
make sure that management approves,

00:04:02.510 --> 00:04:04.931
make sure that appropriate corrections are implemented,

00:04:04.931 --> 00:04:05.595
right?

00:04:05.595 --> 00:04:06.459
But all of that,

00:04:06.459 --> 00:04:12.722
I mean all of that can fall down if whoever's doing the review says,

00:04:12.722 --> 00:04:14.498
I didn't find anything.

00:04:14.498 --> 00:04:17.941
Does this have to be a human reviewing it or can it be a computer reviewing it?

00:04:17.941 --> 00:04:20.246
No, it can absolutely be an automated tool,

00:04:20.246 --> 00:04:20.432
right,

00:04:20.432 --> 00:04:23.029
and there's lot of value in considering either a static

00:04:23.029 --> 00:04:25.751
analysis tool or a dynamic analysis tool.

00:04:25.751 --> 00:04:27.818
It can also be a manual tool.

00:04:27.818 --> 00:04:29.168
A lot of organizations,

00:04:29.168 --> 00:04:32.251
particularly small shops don't always have the capital to buy,

00:04:32.251 --> 00:04:35.105
you know, the licenses for code review.

00:04:35.105 --> 00:04:40.017
But that's cloud based, now, that's not expensive like it used to be.

00:04:40.017 --> 00:04:40.953
It is increasingly inexpensive.

00:04:40.953 --> 00:04:46.399
It is absolutely increasingly inexpensive and compared to labor costs,

00:04:46.399 --> 00:04:51.179
you know, should be a pretty obvious win for this sort of thing.

00:04:51.179 --> 00:04:52.280
Because a small shop, you know,

00:04:52.280 --> 00:04:53.896
to find an individual with knowledge in code review

00:04:53.896 --> 00:04:57.280
techniques and secure coding practice, if you only have two developers,

00:04:57.280 --> 00:04:59.147
that's tough, isn't it?

00:04:59.147 --> 00:05:00.200
It is very tough,

00:05:00.200 --> 00:05:05.232
but a lot of the small shops are brim-full with confidence

00:05:05.232 --> 00:05:08.083
that they know exactly what they're doing.

00:05:08.083 --> 00:05:11.951
Oh, is this scenario where we have what I call the Dunning-Kruger effect,

00:05:11.951 --> 00:05:14.507
you know, where people think they really know what they're doing,

00:05:14.507 --> 00:05:15.882
but lack the knowledge?

00:05:15.882 --> 00:05:17.713
But really, really don’t?

00:05:17.713 --> 00:05:20.006
Yes, yes, that's very, very common.

00:05:20.006 --> 00:05:24.404
It's very common in at least in the U.S., in payment application development,

00:05:24.404 --> 00:05:24.897
right?

00:05:24.897 --> 00:05:26.117
There are a few big players that generate,

00:05:26.117 --> 00:05:28.858
that develop payment applications that are seen everywhere,

00:05:28.858 --> 00:05:32.969
and then there are a lot of really small,

00:05:32.969 --> 00:05:36.193
very niche developers that have you know,

00:05:36.193 --> 00:05:38.612
a particular segment, pizza places or laundromats,

00:05:38.612 --> 00:05:39.414
or whatever,

00:05:39.414 --> 00:05:43.584
for a particular kind of region or for a particular network of clients,

00:05:43.584 --> 00:05:44.411
and they're often very,

00:05:44.411 --> 00:05:47.354
very small shops with just a handful of developers and the

00:05:47.354 --> 00:05:49.884
kind of owner who wrote the first version way back when and

00:05:49.884 --> 00:05:50.767
just don't have the expertise.

00:05:50.767 --> 00:05:54.771
So they've got business knowledge, but not great secure coding knowledge?

00:05:54.771 --> 00:05:58.877
Right, right, and many of them are very resistant to some of these ideas.

00:05:58.877 --> 00:06:03.880
But as an assessor, I mean, you know, you're trying to not,

00:06:03.880 --> 00:06:05.144
because you're dealing with two folders,

00:06:05.144 --> 00:06:07.190
are they compliant and how can I get into a better state?

00:06:07.190 --> 00:06:07.987
Right.

00:06:07.987 --> 00:06:09.647
So how do you get people to a better state?

00:06:09.647 --> 00:06:10.668
Well, ideally,

00:06:10.668 --> 00:06:15.244
ideally you can show them this is what secure software practices are about,

00:06:15.244 --> 00:06:15.541
right?

00:06:15.541 --> 00:06:17.879
This is how we develop secure code.

00:06:17.879 --> 00:06:21.447
These are the kinds of attacks that happen against insecure

00:06:21.447 --> 00:06:24.771
applications and why you need to protect against them,

00:06:24.771 --> 00:06:25.292
right?

00:06:25.292 --> 00:06:27.656
And if you can reason with them, if you can show them,

00:06:27.656 --> 00:06:29.633
this is how you do it, right, and yeah,

00:06:29.633 --> 00:06:33.101
some of it is maybe a reach if you have very

00:06:33.101 --> 00:06:34.241
immature software security practices,

00:06:34.241 --> 00:06:38.364
but some of it is very attainable and you sort of ladder your way up,

00:06:38.364 --> 00:06:41.439
right, then you can at least point them in the right direction,

00:06:41.439 --> 00:06:41.671
right?

00:06:41.671 --> 00:06:45.768
And get them on sort of a cycle of increasing maturity over a year.

00:06:45.768 --> 00:06:49.082
Right, so your assessment is read the SDLC,

00:06:49.082 --> 00:06:50.653
the SSDLC, see if it works,

00:06:50.653 --> 00:06:53.466
talk to the developers and see if they do what's in the SSDLC,

00:06:53.466 --> 00:06:55.815
and look at their code reviews.

00:06:55.815 --> 00:06:56.662
Yeah, look at their output,

00:06:56.662 --> 00:06:59.318
and the code review is the most sort of tangible

00:06:59.318 --> 00:07:02.889
kind of software security output, and like I said,

00:07:02.889 --> 00:07:04.517
it's the biggest tell, right?

00:07:04.517 --> 00:07:06.812
If I looked at it and it was okay, right,

00:07:06.812 --> 00:07:11.270
or if it's a checklist of the OWASP Top 10,

00:07:11.270 --> 00:07:13.797
I checked for SQL injection, check!

00:07:13.797 --> 00:07:15.704
Like oh, did you?

00:07:15.704 --> 00:07:16.180
Right?

00:07:16.180 --> 00:07:17.236
I'll come on to that in a second.

00:07:17.236 --> 00:07:22.167
So do you actually look at code is the question?

00:07:22.167 --> 00:07:22.827
Sometimes.

00:07:22.827 --> 00:07:22.959
Right.

00:07:22.959 --> 00:07:26.446
Okay, so you'll look at code and go well, that's not very good.

00:07:26.446 --> 00:07:28.802
Well, or if it's, you know,

00:07:28.802 --> 00:07:33.818
if there's a relevant section of code in like a code review.

00:07:33.818 --> 00:07:34.056
Right?

00:07:34.056 --> 00:07:35.816
A lot of organizations, when they do the code review,

00:07:35.816 --> 00:07:40.280
they're doing a diff of you know, the current code versus the last version,

00:07:40.280 --> 00:07:45.919
and so yeah, show me the diff; let's take a look.
