WEBVTT

00:00:00.907 --> 00:00:05.403
Encryption of transmitted cardholder data in requirement 4 was actually one of

00:00:05.403 --> 00:00:07.742
those ones that there was not really much to worry about.

00:00:07.742 --> 00:00:10.220
I mean, you turned stuff on and it encrypted data,

00:00:10.220 --> 00:00:12.954
and then we have the SSL to TLS thing.

00:00:12.954 --> 00:00:14.520
So I guess that's what we need to talk about most.

00:00:14.520 --> 00:00:17.786
Is that what your clients are having most problems with?

00:00:17.786 --> 00:00:20.607
That is the biggest trouble, right,

00:00:20.607 --> 00:00:26.710
is using SSL or using TLS correctly and stopping using SSL and

00:00:26.710 --> 00:00:30.814
earlier versions of TLS and some of the other kind of practical

00:00:30.814 --> 00:00:34.206
implementation questions for TLS communications.

00:00:34.206 --> 00:00:40.472
Okay, but from a timing perspective, we are in September 2018?

00:00:40.472 --> 00:00:40.492
Yes.

00:00:40.492 --> 00:00:44.811
And the deadline for moving from SSL to TLS has past?

00:00:44.811 --> 00:00:45.991
It has.

00:00:45.991 --> 00:00:48.719
I mean, it's passed about three times now,

00:00:48.719 --> 00:00:51.233
but it's now all dead as opposed to just mostly dead.

00:00:51.233 --> 00:00:51.776
Yeah.

00:00:51.776 --> 00:00:57.165
Right, and amusingly, in July, in August,

00:00:57.165 --> 00:01:02.204
I had some clients ask me, well, what if we need some time to transition this?

00:01:02.204 --> 00:01:04.143
We have clients that we need to work with that are

00:01:04.143 --> 00:01:08.791
going to take some time to do this, and I was sort of puzzled by that,

00:01:08.791 --> 00:01:12.940
and then you know, only momentarily before that I was entirely not surprised,

00:01:12.940 --> 00:01:13.752
because of course,

00:01:13.752 --> 00:01:19.005
there's been this long transition period where the PCI standard

00:01:19.005 --> 00:01:20.571
has talked about the need to get rid of this,

00:01:20.571 --> 00:01:25.321
the need to have a transition plan, a risk mitigation.

00:01:25.321 --> 00:01:27.541
That has gone on now for several years, right,

00:01:27.541 --> 00:01:30.151
with what was not even implicit,

00:01:30.151 --> 00:01:35.093
but what was explicit that you need to figure out what your business

00:01:35.093 --> 00:01:39.117
requirements are for getting off of these problematic protocols.

00:01:39.117 --> 00:01:41.530
And so then to have somebody say, but wait,

00:01:41.530 --> 00:01:46.086
we can't just do this all at once, really shouldn't surprise anybody.

00:01:46.086 --> 00:01:48.254
I think it's been three years in the coming, in happening?

00:01:48.254 --> 00:01:49.197
It has.

00:01:49.197 --> 00:01:49.393
Okay,

00:01:49.393 --> 00:01:53.134
so let's take it where we're in a position now where what should we be using?

00:01:53.134 --> 00:01:55.397
We shouldn't be, well, tell me what we should be using,

00:01:55.397 --> 00:01:56.659
not what we shouldn't be using.

00:01:56.659 --> 00:01:58.758
Do both, do both.

00:01:58.758 --> 00:02:00.769
Sure, we'll start with the good stuff, right?

00:02:00.769 --> 00:02:04.506
We should be using TLS 1.2, right?

00:02:04.506 --> 00:02:10.877
You can use TLS 1.1.

00:02:10.877 --> 00:02:18.056
From a protocol standpoint, it is not broken the way SSL v3 or TLS 1.0 are,

00:02:18.056 --> 00:02:18.405
right?

00:02:18.405 --> 00:02:22.491
It's maybe on the bubble a little bit, and there are,

00:02:22.491 --> 00:02:27.186
there are potentially some troubles because TLS 1.1 supports a lot of

00:02:27.186 --> 00:02:33.787
crummy ciphers and relatively few kind of good ones.

00:02:33.787 --> 00:02:36.883
The standard has got this great big euphemism in it, hasn't it?

00:02:36.883 --> 00:02:38.071
Right.

00:02:38.071 --> 00:02:39.325
The euphemism is early TLS.

00:02:39.325 --> 00:02:40.269
Early TLS.

00:02:40.269 --> 00:02:43.326
Yes, it doesn't say TLS 1.0, it doesn’t say TLS 1.1.

00:02:43.326 --> 00:02:44.444
Right.

00:02:44.444 --> 00:02:49.072
Believe me, it means TLS 1.0 and TLS 1.1, right?

00:02:49.072 --> 00:02:51.927
That was certainly the intent between the drafters of that clause.

00:02:51.927 --> 00:02:55.406
That was the intent, right, and you know,

00:02:55.406 --> 00:03:01.068
is 1.1 vulnerable to the same kind of weaknesses as 1.0 or SSL 3?

00:03:01.068 --> 00:03:01.498
Not exactly, right?

00:03:01.498 --> 00:03:03.397
But it's not a great idea, right?

00:03:03.397 --> 00:03:05.341
And you should be moving towards TLS 1.2,

00:03:05.341 --> 00:03:06.856
and conceivably, right,

00:03:06.856 --> 00:03:11.853
when TLS 1.3 stops being something that exists entirely on cocktail napkins

00:03:11.853 --> 00:03:16.480
and maybe start happening in actual browsers or web servers,

00:03:16.480 --> 00:03:19.158
that would be the new ideal.

00:03:19.158 --> 00:03:20.349
And we're close to that now.

00:03:20.349 --> 00:03:21.085
We are.

00:03:21.085 --> 00:03:21.588
We're really close to that.

00:03:21.588 --> 00:03:23.147
We are close to that.

00:03:23.147 --> 00:03:27.636
Okay, so the first thing is getting the right protocol, CLS 1.2.

00:03:27.636 --> 00:03:30.458
The second thing is the right suite.

00:03:30.458 --> 00:03:32.976
The right ciphers, right.

00:03:32.976 --> 00:03:34.384
And you know,

00:03:34.384 --> 00:03:40.870
here I will actually just endorse something from Qualys called SSL Labs.

00:03:40.870 --> 00:03:48.638
Go to www.ssllabs .com and you can point any sort of

00:03:48.638 --> 00:03:53.874
internet-facing site to the test available from SSL Labs,

00:03:53.874 --> 00:04:01.056
and it will give you a load of information about how your website is configured.

00:04:01.056 --> 00:04:02.667
Yeah, go to the Test Your Server.

00:04:02.667 --> 00:04:03.855
So let me do that, let me go Test Your Server,

00:04:03.855 --> 00:04:06.030
and should we test the PCI Council?

00:04:06.030 --> 00:04:07.318
That would be?

00:04:07.318 --> 00:04:09.595
That would be fun.

00:04:09.595 --> 00:04:10.082
Is that okay?

00:04:10.082 --> 00:04:10.697
Yeah.

00:04:10.697 --> 00:04:11.778
Do you not show these results just in case?

00:04:11.778 --> 00:04:14.399
They're bound to be an A+ anyway.

00:04:14.399 --> 00:04:20.007
Well, let's find out.

00:04:20.007 --> 00:04:21.176
And it is actually pretty quick.

00:04:21.176 --> 00:04:25.186
You see the little spinner going round and round, but it moves along.

00:04:25.186 --> 00:04:28.413
So now it's just pulling out all the different certs that were in use.

00:04:28.413 --> 00:04:29.488
Right.

00:04:29.488 --> 00:04:30.741
I think this is a brilliant tool.

00:04:30.741 --> 00:04:31.973
I use it a lot.

00:04:31.973 --> 00:04:34.343
It's really useful; it's really useful as long as

00:04:34.343 --> 00:04:35.699
that site is internet accessible.

00:04:35.699 --> 00:04:39.049
If you've got an internal site that isn't exposed to the internet,

00:04:39.049 --> 00:04:41.920
obviously you can't get SSL Labs to find it,

00:04:41.920 --> 00:04:45.520
you can conceivably replicate that with some of like

00:04:45.520 --> 00:04:49.625
the open SSL command line utilities, but that's more work.

00:04:49.625 --> 00:04:51.989
Now one of the things that I always used to stumble with

00:04:51.989 --> 00:04:55.499
clients was about sending and receiving data.

00:04:55.499 --> 00:04:58.166
So some clients would say, well, you know,

00:04:58.166 --> 00:04:59.996
requirement 4, it's something about me sending data,

00:04:59.996 --> 00:05:02.958
it's not about my web server accepting data.

00:05:02.958 --> 00:05:03.508
Well, of course it is.

00:05:03.508 --> 00:05:05.843
I know of course it is, but I just wanted to make it,

00:05:05.843 --> 00:05:07.802
you know, it's really clearly in the standard.

00:05:07.802 --> 00:05:11.412
If you're sending data, you need to send it over strong cryptography,

00:05:11.412 --> 00:05:13.584
and if you're receiving data,

00:05:13.584 --> 00:05:18.256
you need to configure your system to receive it over strong.

00:05:18.256 --> 00:05:21.094
There's no cop-out to say well, that browser over there is really poor.

00:05:21.094 --> 00:05:24.875
It's like, but no, your server will accept that, that's not accept that.

00:05:24.875 --> 00:05:27.297
The other fun cop-out that we should probably sort of quash

00:05:27.297 --> 00:05:32.489
while we're at it is the idea that oh, I establish a VPN connection,

00:05:32.489 --> 00:05:35.161
you know, to my payment processor, to my business partner,

00:05:35.161 --> 00:05:35.555
right?

00:05:35.555 --> 00:05:38.176
That's not a public network.

00:05:38.176 --> 00:05:40.839
Well, it's a VPN connection riding over the public internet,

00:05:40.839 --> 00:05:45.396
and the cryptographic properties of that VPN matter so yeah,

00:05:45.396 --> 00:05:48.525
it is and your assessor should look at it.

00:05:48.525 --> 00:05:50.313
Okay, so it looks like our results have finished.

00:05:50.313 --> 00:05:54.547
Yeah, and it's got an A, and so funny enough,

00:05:54.547 --> 00:05:58.213
I feel like you need to pull the sort of Spaceballs bit

00:05:58.213 --> 00:06:01.031
where what does the letter grade mean and you show Rick

00:06:01.031 --> 00:06:04.016
Moranis in the outfit saying, absolutely nothing,

00:06:04.016 --> 00:06:05.778
because that's really appropriate.

00:06:05.778 --> 00:06:08.695
The letter grade isn't all that interesting,

00:06:08.695 --> 00:06:13.194
because you can get an A with some potentially problematic things,

00:06:13.194 --> 00:06:15.986
particularly if you have PCI DSS obligations.

00:06:15.986 --> 00:06:19.680
Okay, so if you scroll down to the results like you're doing now,

00:06:19.680 --> 00:06:22.415
you can see some of the certificates, you can see the protocols supported,

00:06:22.415 --> 00:06:24.694
and then here's where it gets interesting.

00:06:24.694 --> 00:06:27.468
If you go look at that Cipher Suites list.

00:06:27.468 --> 00:06:30.976
But the first thing is that it's showing me that like it's just using TLS 1.2.

00:06:30.976 --> 00:06:32.330
It is.

00:06:32.330 --> 00:06:34.522
So am I using deprecated protocols?

00:06:34.522 --> 00:06:34.887
No.

00:06:34.887 --> 00:06:35.963
No, well done.

00:06:35.963 --> 00:06:37.768
Alright, but I am using some weak ciphers.

00:06:37.768 --> 00:06:38.021
Yes.

00:06:38.021 --> 00:06:38.274
Right,

00:06:38.274 --> 00:06:43.110
and so you can of course still get an A and have kind of a load of weak ciphers.

00:06:43.110 --> 00:06:45.366
Is that a DSS fail?

00:06:45.366 --> 00:06:50.159
I would contend, I would contend that weak ciphers,

00:06:50.159 --> 00:06:55.335
right, are not strong cryptography, right?

00:06:55.335 --> 00:06:57.554
And generally this is just about bad configuration, isn't it?

00:06:57.554 --> 00:06:59.864
Someone not thinking what cipher suites to use?

00:06:59.864 --> 00:07:06.691
Right, or more precisely, not actively disabling the problematic ones.

00:07:06.691 --> 00:07:06.824
Right?

00:07:06.824 --> 00:07:08.047
When you see this set of cipher suites,

00:07:08.047 --> 00:07:12.003
this is the set of cipher suites that that server will allow

00:07:12.003 --> 00:07:15.019
in the negotiation between the client, the browser usually,

00:07:15.019 --> 00:07:16.117
and the web server.

00:07:16.117 --> 00:07:20.675
Typically an up-to-date browser will recommend you know,

00:07:20.675 --> 00:07:22.664
one of the stronger cipher suites.

00:07:22.664 --> 00:07:26.082
They'll pick the top one off the stack is typically what'll happen,

00:07:26.082 --> 00:07:31.900
and so it's only if you've got an older browser or if you've got a man in

00:07:31.900 --> 00:07:35.613
the middle forcing a downgrade to do something spurious,

00:07:35.613 --> 00:07:38.053
then you get the weaker cipher suites.

00:07:38.053 --> 00:07:39.597
I have a terrible confession to make.

00:07:39.597 --> 00:07:40.593
Okay.

00:07:40.593 --> 00:07:42.706
But last week I did some of this cleaning up for

00:07:42.706 --> 00:07:44.777
somewhere and I broke a whole lot of stuff,

00:07:44.777 --> 00:07:50.108
because somewhere out there on the internet were browsers that I

00:07:50.108 --> 00:07:51.686
think were about a year and a half out of date.

00:07:51.686 --> 00:07:52.610
Right.

00:07:52.610 --> 00:07:56.998
So no one had updated, and they stopped working, and I got into trouble.

00:07:56.998 --> 00:08:03.275
Well, I will say, you know, for this, as well as really anything else,

00:08:03.275 --> 00:08:06.825
maybe test it before you sort of deploy.

00:08:06.825 --> 00:08:08.558
You know, that theoretically had happened,

00:08:08.558 --> 00:08:10.279
you know, but you know,

00:08:10.279 --> 00:08:11.724
the practice was the telephone rang and somebody rang

00:08:11.724 --> 00:08:14.172
and it was we can't make transactions, that's quite poor,

00:08:14.172 --> 00:08:15.407
can you fix it?

00:08:15.407 --> 00:08:17.142
Right, right, and well,

00:08:17.142 --> 00:08:20.380
you know you've got a fair point where there is some concern,

00:08:20.380 --> 00:08:22.251
right, if you've got an older browser,

00:08:22.251 --> 00:08:27.086
if you are supporting folks that are still running Windows XP and its

00:08:27.086 --> 00:08:30.658
attendant version of IE or very old Android devices.

00:08:30.658 --> 00:08:32.785
Android devices are what my marketing department always say,

00:08:32.785 --> 00:08:36.918
look, we get 1% of our sales through these old Android devices.

00:08:36.918 --> 00:08:37.824
Right, well,

00:08:37.824 --> 00:08:42.284
and then you get into the realm of what's a viable compensating control,

00:08:42.284 --> 00:08:45.829
right, and what is the valid business constraint?

00:08:45.829 --> 00:08:47.408
If your marketing department can actually show you,

00:08:47.408 --> 00:08:50.680
no, we really do get real money this way,

00:08:50.680 --> 00:08:53.896
you know, okay, now we're talking about a valid constraint.

00:08:53.896 --> 00:08:58.716
There's a QSA, that's a valid business reason; that's what you're looking for?

00:08:58.716 --> 00:09:00.903
That valid means it's not someone making it up.

00:09:00.903 --> 00:09:02.871
Right, someone can substantiate that.

00:09:02.871 --> 00:09:04.181
If your marketing people are saying, oh well,

00:09:04.181 --> 00:09:07.696
we might possibly get some sales from crummy old Android devices,

00:09:07.696 --> 00:09:10.040
that doesn't seem like a valid business constraint,

00:09:10.040 --> 00:09:12.404
but if your marketing department can say,

00:09:12.404 --> 00:09:18.243
here is data that shows that we do in fact actually get real money,

00:09:18.243 --> 00:09:20.197
now we're talking.

00:09:20.197 --> 00:09:25.847
Okay, so if we look at the requirements the way we've been looking at them,

00:09:25.847 --> 00:09:29.129
for requirement 4, I guess your main fails have been on TLS?

00:09:29.129 --> 00:09:30.693
Typically.

00:09:30.693 --> 00:09:33.159
Not having the transition program in place?

00:09:33.159 --> 00:09:33.815
Right.

00:09:33.815 --> 00:09:37.476
Anything else in 4.1 or 4.2 that are real gotchas for

00:09:37.476 --> 00:09:40.328
merchants or service providers?

00:09:40.328 --> 00:09:43.386
Well, so you could argue that the ciphers are not about the transition, right?

00:09:43.386 --> 00:09:45.243
They're just about the protocol,

00:09:45.243 --> 00:09:47.204
and so you get a lot of people complaining like oh,

00:09:47.204 --> 00:09:47.829
I fixed this,

00:09:47.829 --> 00:09:51.283
I fixed this; why do I still have vulnerabilities show up in my scan?

00:09:51.283 --> 00:09:54.028
Why is my assessor still crowding at me about TLS?

00:09:54.028 --> 00:09:57.323
Well, you fix the protocol, you then fix the ciphers,

00:09:57.323 --> 00:10:03.323
right, or you've got bum certificates.
