text stringlengths 0 2.67k |
|---|
**Justin Garrison:** The impacts of the things you do on top can definitely cause a problem below. |
**Autumn Nash:** Exactly. And you can't really fix something if you don't understand at least a few layers down, you know? |
**Justin Garrison:** So let's talk about CrowdStrike. Leading right into that... |
**Autumn Nash:** When it goes wrong... No, but honestly, as much as this is not the right way to do something, I deeply feel for the fact of like -- can you imagine the panic that you would feel if you took down airports, hospitals? We know that something went wrong, and there should have probably been more testing. An... |
**Justin Garrison:** So look, if you're sick of listening to CrowdStrike reviews from people that weren't actually involved, feel free to skip this part. This is just stuff that we've learned reading a couple different things. And the main places that I'm pulling information from are the CrowdStrike executive summary f... |
\[56:03\] In the executive summary, they say what they're doing to make sure this doesn't happen again. They say that they're enhanced software testing procedures. And they say "Improve rapid response content testing by using testing types such as local developer." That's not a thing you should be relying on. They wrot... |
**Autumn Nash:** I'm a little scared that Justin's face said... I knew it was going wrong. Y'all can't see it, but I knew it was going wrong when he -- okay, he has this Justin shade face that y'all don't get to see... And right before he's about to throw it... And I'm just so excited, and I'm just like "Oh, it's comin... |
**Justin Garrison:** I read local developer and I'm like "Okay, first of all, they're not talking to a developer." This is the executive summary. Okay, but local developer is not your testing framework for how these things -- they wrote the bad code. Don't rely. They could change the tests to pass. Do not rely on them ... |
**Autumn Nash:** That's the part that just -- it's almost like they don't understand what happened. And I know the engineers and everyone else understands, but when you read that, it's so tone deaf that you're just like "Bro, did you talk to the people with the technical background before you wrote this? Because..." |
**Justin Garrison:** And this is not a technical person that wrote. This is a legal team that is trying to limit the blast radius of what they're -- like "We're going to make improvements." |
**Autumn Nash:** Yes, but to gain trust of your customers wanting to use your product going forward, you should have discussed this with your technical team, because now - like, you've already had this huge incident, right? And people are already like "Well, do I trust you to roll out these updates?" And then you relea... |
**Justin Garrison:** Well, and I'll get to that last part. So this is just the testing improvements. And they have stress, fuzzing, fault injection, stability... What does stability mean? And content interface testing. I'm like "None of those things would catch this." |
**Autumn Nash:** Yeah. None of that addresses a configuration problem. |
**Justin Garrison:** Right, because this was not a business logic, quote-unquote, "business logic" change. |
**Autumn Nash:** It's not a code change. |
**Justin Garrison:** Yeah. Which usually roll out differently. |
**Autumn Nash:** Yeah. But what I'm saying is if you were a CrowdStrike customer, does this now make you feel any more trusting of them as a company? If you read something that they're like "We're going to make all these changes that would not have made this any better." You know what I mean? Why did you say it? You co... |
**Justin Garrison:** One of the other things they say here in the executive summary is the other enhancement is going to be adopting a staggered deployment strategy, which means they did not have one. This was a push go everywhere. And they even say in the reporting, if you were online during this hour and a half that ... |
**Autumn Nash:** Windows Office asks you if they want to update it. You know what I mean? I'm just like "What do you mean?" |
**Justin Garrison:** I would absolutely argue that Excel is more important than your antivirus for your company to do their work, and also, I would not be surprised at all if Excel had kernel hooks, at the kernel level, where they're like "This function, if you're calling that equals column", whatever, there's probably... |
**Autumn Nash:** Okay, we've just discussed that infrastructure has come so far, how we deploy things have come so far. Bro, you didn't have staggered deployments? At the same time, I feel bad for throwing shade at anything, because this job is hard, and there's always an edge case... But don't shoot yourself in the fo... |
**Justin Garrison:** \[01:00:20.07\] And the process isn't always uniform for everything. Because if this didn't have kernel hooks and wouldn't blue-screen the computer -- like, okay, if the CrowdStrike agent failed, not a big deal. Like, okay, we can reboot it. You can reboot the system and things will come back onlin... |
**Autumn Nash:** That's kind of interesting too though, because a lot of different articles are debating whether that makes it partially a Windows Microsoft problem, and if they're partially to blame. But like you said, in a way they did what they should be doing. You know what I mean? ...in one way of thinking. So it'... |
**Justin Garrison:** Well, Windows was protecting Windows, right? |
**Autumn Nash:** That's what I'm saying. That was technically an expected behavior that it should have been doing. You know what I mean? So that debate on like responsibility, and where does it go forward with SLAs... |
**Justin Garrison:** And I love the overview here from Pragmatic Engineer points to regulation as a problem... Because regulation is why these things still run in the kernel. And I remember I was at the university when we switched from XP to Vista, and during Vista timeframe, Microsoft was getting into the "We're going... |
**Autumn Nash:** This is also another one of those things of like "Talk to the technical people before you make regulations, and before you make business decisions." That's why it's really important to have those middle people who talk both languages, like your product managers, or your project managers, and to have so... |
**Justin Garrison:** Yeah. And the decision they could have made was "Okay, Windows Defender doesn't get kernel access either, so no one gets kernel access." But that's not the decision they made. They made the decision that Windows Defender needs kernel access, so everybody gets kernel access. And... |
**Autumn Nash:** Which is going the opposite direction, and not a good idea either. |
**Justin Garrison:** Well, I mean, at the time too -- we're looking back to early 2000s. That was not an uncommon thing, because there was less clear defined lines between user space and kernel. It's like "Well, a lot of things run in the kernel that shouldn't." |
The other interesting thing that Brendan Gregg, who's over at IBM now - he had really great insights into eBPF, the enhanced Berkeley Packet Filtering as a system would have made this safe. And eBPF runs at kernel space, but makes the code safe so it can't crash the kernel if the eBPF program goes down... Which I thoug... |
And so eBPF is going to be your one program to write for the kernel space, and it works on Linux and Windows. And it's also safe, where you get a little more elevated privileges to do things at a lower level, without making the whole system crash. |
**Autumn Nash:** \[01:04:09.24\] That's interesting. I feel that will be used in a lot of design reviews on a reason to use that going forward. They'll be citing that. My favorite part of all of this though is that in the programmatic programmer he talks about two extremes in shipping to production. It says YOLOs, and ... |
**Justin Garrison:** Yeah, we'll have the links in the show notes here. I just pulled up Brendan's blog too, because I thought it was really interesting, as as someone -- he's been doing eBPF for a very long time, when he was at Netflix, and beyond that. And so seeing these things shift to safer ways to run on the syst... |
**Autumn Nash:** I think that's the thing between security and usability. Like, you always want to make something secure, but you also want to make it usable. And you're constantly in that battle between "How do we make it secure, and also how do we make it where we're not making it so secure it's unusable?" So you're ... |
**Justin Garrison:** The Venn diagram of secure and usable are just two separate circles, right? There's no overlap there. |
**Autumn Nash:** That's what I'm saying. So something like this, what this product that is giving you -- that just seems like it's going to be a huge win, because if it's going to separate it in that manner and then stop you from ending up going into kernel panic, why wouldn't you include that? |
**Justin Garrison:** And I feel like the antivirus companies have had decades, literally two decades of training users that you want your antivirus definitions as fast as possible. Because people like me live through worms, and things that would spread through networks really fast, and we could not get definitions fast... |
**Autumn Nash:** I don't even know if that's just antivirus. That's like open source in a lot of places; we're constantly wanting to hurry up and fix things, but then you want to make sure that they're well tested. And you don't want to panic people. it just seems like it's always the cycle of software in general. |
**Justin Garrison:** Yeah, and a lot of that - you know, people moving from frameworks. It's a lot of work, and they're like "This is going to fix a bunch of bugs for me", or "This is going to solve some security vulnerabilities. When do I have time to do that sort of maintenance?" And that is stuff that you outsource ... |
**Autumn Nash:** But it's crazy, because I feel like the farther they automate security, there's more false positives, and then people just ignore it. We have to do some sort of automation, because if not, no one's going to sit there and just watch that all day. But people just keep -- it was so funny, I've seen people... |
**Justin Garrison:** And I absolutely have no problem throwing shade, because I know that Google and Microsoft both are like "AI is going to solve that problem for you." And I am so like "Oh, my gosh. You don't --" |
**Autumn Nash:** I love when you throw shade for me, so I don't have to do it. I'm always trying to not do these things... And Justin's like "Let me help you." |
**Justin Garrison:** And you could be thinking of someone else. I don't know. But those were the two that came to my mind as like "Oh, they went all-in on this new technology that no one understands is going to solve the problem that we all know very well." |
**Autumn Nash:** \[01:08:09.26\] Not just that, but how much compute power -- and you're throwing so much at a DoS attack when that's not the big picture. You know what I mean? Yeah, you're solving a very small section, but if you give so many false positives... And you're selling this as like this security suite, but ... |
**Justin Garrison:** And that goes right along with these AI claims, and Google and Microsoft both are no longer claiming to be carbon neutral. |
**Autumn Nash:** Did you see the new report that Microsoft made? It hurt my soul a little bit. They basically said if you just ran a ton of servers and laptops for years, it'd equate a couple days of what they're using right now. And I'm just like "How did we go from so concerned to YOLO?" We live on the same planet, y... |
**Justin Garrison:** I switched to DuckDuckGo. |
**Autumn Nash:** Oh, gosh. |
**Justin Garrison:** I literally switched to DuckDuckGo, because a) I don't want to contribute to AI generating anything for me that I'm not explicitly asking for and causing more stuff... DuckDuckGo has actually been really great. They have a setting in there, it's like "No ads. Just don't show me ads." It's amazing. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.