Research Saturday 2.9.19
Ep 72 | 2.9.19

Trends and tips for cloud security.


Dave Bittner: [00:00:02] Hello everyone, and welcome to the CyberWire's Research Saturday, presented by Juniper Networks. I'm Dave Bittner, and this is our weekly conversation with researchers and analysts tracking down threats and vulnerabilities, and solving some of the hard problems of protecting ourselves in a rapidly evolving cyberspace. Thanks for joining us.

Dave Bittner: [00:00:25] And now a quick word about our sponsor, Juniper Networks. They're empowering you to automate your security, see your networks, and protect your clouds. Juniper Networks has you covered so your security teams can finally get back to fortifying your security posture. Learn more at, or connect with Juniper on Twitter or Facebook. That's And we thank Juniper for making it possible to bring you Research Saturday

Dave Bittner: [00:00:58] And thanks also to our sponsor, Enveil, whose revolutionary ZeroReveal solution closes the last gap in data security: protecting data in use. It's the industry's first and only scalable commercial solution enabling data to remain encrypted throughout the entire processing lifecycle. Imagine being able to analyze, search, and perform calculations on sensitive data - all without ever decrypting anything. All without the risks of theft or inadvertent exposure. What was once only theoretical is now possible with Enveil. Learn more at

Ryan Olson: [00:01:33] So, this one's a little bit different from the kind of stuff that we normally publish.

Dave Bittner: [00:01:42] That's Ryan Olsen. He's Vice President of Threat Intelligence for Palo Alto Networks. The research we're discussing today is titled, "Cloud Security Trends and Tips: Key Learning to Secure Your AWS, Azure, and Google Cloud Environments."

Ryan Olson: [00:01:57] Something that we think is important, as more organizations move into the cloud - they start moving more of their infrastructure there, especially into public clouds - is just sort of understanding that there's this shared responsibility model between the cloud provider - the infrastructure provider - and the organization that's actually deploying their software out there. They do not take over all of the security for everything that's happening just because it's on their systems. You know, you still have to patch your systems. You have to update their software. You still have to manage credentials, and do all of these other things that you would do if it was in your data center as well. It's just - you don't need to worry about, you know, the physical devices themselves.

Ryan Olson: [00:02:33] So, we published this report to help people understand that these threats, you know, they do exist. A lot of people are making mistakes out in the cloud, how they configure their systems. I want to shine a bit of a light on that, so there's a better understanding that there's real threats out there.

Dave Bittner: [00:02:48] And do you think there's a misperception out there, that folks feel as though - I guess, is there a false sense of security when folks move their stuff to the cloud?

Ryan Olson: [00:02:57] I think there is a false sense of security that people feel because, you know, just the fact that people don't - they can't necessarily just sort of quit your systems the same way. You know, you don't feel as exposed because it's in the cloud, and you sort of have this trust in your provider - whoever they might be - that they're doing all these good things for you to keep your systems safe.

Ryan Olson: [00:03:15] But no matter what they do, it still comes down to - you're running software. That software has possible vulnerabilities. Your data is there, and anyone can access it under the right conditions, which is why we've seen lots of breaches that have occurred in public cloud environments, because people have simply left their, you know, their buckets exposed. So anyone can go and download a whole bunch of files, because the person who deployed it didn't necessarily understand what the new security model really was.

Dave Bittner: [00:03:40] Right. Right. Well, so, before we dig into some of the specifics here - I mean, you all basically highlighted five broad categories of things to be concerned with, the things that require your attention when moving to the cloud. You want to just start off by taking us through that list?

Ryan Olson: [00:03:56] Sure. So, the first one is account compromises. And based on the data that we're collecting through our RedLock platform, we were able to identify that probably around 29 percent of organizations have some accounts that are actually compromised, that they haven't changed passwords for - which is a lot. That's problematic. Because if you think about how the public cloud works, in many cases, the only thing stopping someone from accessing a host is a login and password accessing some sort of data.

Ryan Olson: [00:04:22] So, having excellent credentials - credentials that are, you know, regularly-changed, that haven't been breached, and, ideally, using multi-factor authentication for particular logins - becomes even more important in the public cloud. Because if you compare to your own data center, if - to access, you know, a sensitive application inside your network - you generally need to be either inside the network, or connected to a VPN of some kind to allow you to get to that host in the first place. But with a lot of public cloud deployments, those hosts are just completely exposed to the Internet, so anyone can get to it. So the only thing stopping an attacker from getting to the data is just having the right password - the right sort of key to the door.

Dave Bittner: [00:05:00] One of the things that I noticed in your research is this whole notion of there being dormant accounts. You know, folks who've been given access who might not need it anymore.

Ryan Olson: [00:05:10] Yup. And that's actually pretty common. Someone will create an account for an individual somewhere in some public cloud infrastructure. They set it up because they need to make some changes, and they never sort of go and clean that account up. Maybe the person has left the company. Maybe they just never had another reason to go and log into it, and the account just sits there dormant. And that's an exposure, because that's a user with a password who could be compromised. If someone compromises your, you know, your laptop, and they get access, they install a keylogger, and they get those credentials, if there isn't some sort of multi-factor authentication, all those accounts that are sitting out there unused - they're really just extra exposure that's unnecessary.

Dave Bittner: [00:05:48] And you make the point that you need to operate under the assumption that credentials will be compromised. Take us through - what do you mean there?

Ryan Olson: [00:05:56] Yeah, so this is a little different than how most people - a lot of people have these policies inside their companies where they have to rotate their credentials every, you know, sixty or ninety days. And I think those are - sometimes those policies can be detrimental, because people then create these password schemes that, you know, are easy to crack by someone who actually has seen one of the passwords.

Ryan Olson: [00:06:15] But in public cloud infrastructure, I think it's more important to take those steps, because credentials do get compromised, especially when people have bad sort of password systems. And because it's exposed out to the Internet, you wouldn't even necessarily know - unless you're monitoring - that those credentials were used at some point.

Ryan Olson: [00:06:33] So, it isn't a guarantee that every single login and password combination is going to be breached at some point but making these changes - and especially because you can do this pretty easily in public cloud infrastructure, rotate credentials - it just becomes an even more important component of keeping that data more secure.

Dave Bittner: [00:06:48] So, when it comes to preventing account compromises, you all have some tips here. You want to walk us through those?

Ryan Olson: [00:06:54] Yeah, so I think the most important one that I just mentioned was enforcing multi-factor authentication, especially on privileged accounts. Most public cloud infrastructure have this concept of a root account that has access to all of the different services - they can, you know, deploy additional virtual machines. They can do all these things and they're not really restricted by the same kind of policies that other users are. That one, in particular, is most important to have multi-factor authentication on, either through a token or through some sort of SMS.

Ryan Olson: [00:07:22] And then the other aspect of this is - credentials aren't just passwords. Credentials, especially in public cloud, often are access keys. Sort of these long tokens that are generated because one service is talking to another service. It's not necessarily a user - it's just, you need to provide this API key to be able to access the data. Those can be exposed through ways that people just aren't used to. Oftentimes, they get bundled into code - code, you know, inside a script that's going to access a system - and that code can get exposed itself. People make mistakes by accidentally uploading it to, you know, to GitHub or to somewhere else, and maybe making a repository public that wasn't supposed to be public. Those are the same issue of credentials that have been exposed and someone can take control over that you just have to think more about when it comes to the public cloud.

Ryan Olson: [00:08:07] So, I think those are the two main ones, of the ones that we recommended, that I really want to note for the listeners.

Dave Bittner: [00:08:13] Yeah. Well, let's move on to the next topic and this was cryptojacking.

Ryan Olson: [00:08:16] Cryptojacking has been a huge topic for the last, I'd say, almost eighteen months now. Because we were seeing, on the desktop, on people's computers, we were seeing lots of attackers move to compromising hosts so that they could go and mine specific cryptocurrencies. Monero was the one that we were seeing most.

Ryan Olson: [00:08:34] But we also saw these attacks happening in the public cloud. Basically, if the attacker can get access to those credentials - whatever they are, API keys or otherwise - if they can spin up a virtual machine in your cloud, they can spend a lot of your money going and mining cryptocurrency. And what we've seen toward the end of the year is, as the currency prices have been dropping significantly - we think the price of Bitcoin is around $3,500 US right now, and last year at this time it was closer to $18,000 - we've seen this trend cooling off a little bit.

Ryan Olson: [00:09:06] This is one that I don't expect to just disappear. I don't think we're going to see all attackers who are using the cryptomining business model stop performing it. But I do think we're gong to see people shift back more toward, you know, launching ransomware attacks and other kinds of attacks that they have, that they can make more money off of, as the prices drop.

Dave Bittner: [00:09:23] And so, what are your recommendations for best practices here?

Ryan Olson: [00:09:25] Really understanding what's getting deployed inside of all of your virtual machines is really important. And one thing that you can do is, in these cryptomining attacks, something that's required is that the virtual machine that is deployed has to be able to go and talk to a mining pool, and the best thing you can do is make sure you're actually monitoring the traffic out of all the virtual machines inside of your public cloud, so that you can see, are we seeing traffic to mining pools? And ideally, just implement a deny-all policy to say, unless you have specifically allowed traffic from these VMs, don't allow anything out. It's just unnecessary traffic going out, and it'll prevent you from being impacted. Because nobody's going to deploy these in your network, even if you are compromised, if they're not going to make any money off of it.

Dave Bittner: [00:10:10] So, the next category you looked at was compliance. Take us through what you found here.

Ryan Olson: [00:10:15] Yeah, so, compliance isn't something that, in Unit 42, we normally have to talk a lot about. But in the public cloud, it becomes more important, because you're effectively saying, hey, you have a whole bunch of new infrastructure that you've gone out and deployed, and there are a lot of regulations around how you actually defend this data.

Ryan Olson: [00:10:31] So, one of the things that people like us do with products like RedLock is we look at, how are you doing from a configuration perspective compared to various standards? You know, HIPAA standards, GDPR, PCI - all of these things where there are regulations around you have to encrypt certain kinds of data, and you have to maintain certain kinds of information to actually keep it secure. And what we found were, a lot of organizations, as they deploy, are just not meeting those requirements. You know, there are certain best practices and requirements that are are laid out, that just - they're not configured properly to hit all of them.

Ryan Olson: [00:11:04] So, one example was we found about 32 percent of organizations weren't fully hitting their requirements for GDPR, the general data privacy requirement that was passed in the EU recently. And that's a big percentage. That's a lot of organizations who will all - what they really need to do is start making some changes in how they're managing that infrastructure, but also have the data - understand, are you actually hitting these compliance requirements or not, so that you have the ability to say we need to go and make some changes?

Dave Bittner: [00:11:30] And why specifically is this more of an issue in a cloud environment than if you're hosting the stuff yourself?

Ryan Olson: [00:11:36] I think it's a similar issue between the cloud as well as keeping this inside of your data center. But I think something that happens in public cloud environments, which doesn't happen as much in data centers, is people inside an organization deploying virtual machines or deploying other kinds of infrastructure, without having to go through their IT org - the IT org who generally has, you know, things in place to make sure they're complying with all those regulations.

Ryan Olson: [00:12:01] If you have, you know, a separate part of the business who can suddenly spin up virtual machines, and deploy code, and set up databases without ever talking to their IT team, they end up in a situation where they might be falling out of compliance with these regulations. So, making sure that the IT and the security teams have the ability to perform these kind of audits and get this data becomes even more important in those public cloud situations.

Dave Bittner: [00:12:26] Well, let's move on to the next one, and this was vulnerability management. Take us through this one.

Ryan Olson: [00:12:31] So, when you deploy virtual systems out into the public cloud, unless you're using, you know, one of the services where it's fully provided to you by the cloud vendor - so, if they're hosting the entire SQL database or something like that - you're still responsible for making sure the systems are getting patched. So you still have to update all of your software, make sure you're running the proper kernels on your operating system, making sure you're updating the PHP libraries, whatever they might be for what you're actually running.

Ryan Olson: [00:12:59] And what we found was about 23 percent of organizations have a host that's missing some sort of critical patch in the cloud. Which isn't really that terrible. You know, if you look at a data center, oftentimes, you do find a lot of hosts that don't have everything patched. That's why patching has been such a big challenge. But what's true in a data center is you can deploy a vulnerability scanner, scan across all your hosts, and sort of know what the situation is. That's not quite as easy to do in the public cloud. You've got to have different ways to actually go and gather that inventory of all the hosts that are actually running inside your public cloud and what software is on them. And that's a little bit different. It just requires some different kinds of visibility.

Dave Bittner: [00:13:38] And so, what do you recommend? How can folks get on top of this?

Ryan Olson: [00:13:42] I really think the most important thing here is, again, having that visibility. So, being able to sort of correlate all the vulnerability data that you have inside your network in the public cloud, and then look at the controls that you have in place. If you don't - if you have host that aren't patched, understand what kind of controls exist for you in the public cloud, either as a virtual firewall or something else, that can prevent those vulnerabilities from being exploited from the outside.

Ryan Olson: [00:14:05] I don't think we're ever going to be in a situation where we can tell everyone, just make sure you've got everything a hundred percent patched. It's a great world of the future, a great thing to think about. But if we're never gonna get there, it's about mitigating what the risk is. And the way that you approach that mitigation is different in public cloud, because you're - you don't have the same kind of infrastructure, necessarily.

Dave Bittner: [00:14:24] Hmm. Now, the last category that you explored was managed container services. What were we talking about here?

Ryan Olson: [00:14:30] Yeah, so, containers are really growing in popularity, especially inside the public cloud. I believe that our number was about 25 percent of organizations were using some sort of container service in their public cloud - either Amazon's Elastic Container Service or Kubernetes, a different version of Kubernetes from Google or from Azure. And that means that's a quarter of them who have deployed this container management system.

Ryan Olson: [00:14:55] So, in a normal situation, if you were doing this in your data center, those Kubernetes instances - the hosts that are actually allowing you to sort of orchestrate the creation of new containers - it would be walled off inside your data center. No one's going to be able to go and touch it. But in the public cloud, it's possible that you leave that fully exposed to the Internet.

Ryan Olson: [00:15:13] So, we found 46 percent of organizations - their Kubernetes pods, where they could accept these instructions - they were allowed to accept traffic from anywhere. Which is just unnecessary. It's unnecessary for those to be left exposed, and it's an important thing to actually control. Because being able to deploy, basically, a virtual machine - you know, a docker image that contains a bunch of software - inside somebody's network gives you a lot of access, both for an abuse perspective, for taking up their resources, but also to potentially launch some sort of more malicious attack, where they go and try to access more data.

Dave Bittner: [00:15:45] Hmm. So, what do you suggest here?

Ryan Olson: [00:15:48] In this case, there's there's some good best practices around just sort of blocking off access to these systems, using the user management system within your public cloud provider to prevent people who don't need access to your Kubernetes managed instance from accessing it. And then also, where possible, try to find good solutions for monitoring your container deployments. You want to be able to understand, what containers are you launching? What is the source of the images for those? So that you have more confidence that there isn't something malicious going on in one of these sort of things that are foreign to a lot of people in the world of IT.

Dave Bittner: [00:16:20] So, I mean, looking at these as a whole, sort of a high-level look, what are the overall take-homes from the examination that you did here?

Ryan Olson: [00:16:29] I think the best takeaway is that a lot of people who are moving their infrastructure the public cloud - and nearly everybody is doing in one way or another, they're at least sort of dipping their toe in the water - don't necessarily know how to keep that secure, and they don't have the right tools to make sure they're keeping it secure.

Ryan Olson: [00:16:47] And from a security perspective, in particular, my team - generally, what we're talking about is adversaries who have compromised networks, and they're installing malware, and they're trying to steal people's data. What you see in the public cloud is a different kind of attack. You know, you might have seen somebody attack an organization and steal their credentials, but then they use those credentials just to log into a database and extract a whole bunch of data, or download a whole bunch of files. The actual evidence that occurs out in the public cloud is really different.

Ryan Olson: [00:17:14] And you introduce these new capabilities, like containers, and all these different services that the cloud providers have deployed. Just to give you one quick sort of anecdote - I was talking with someone about why public cloud is different from a from a threat research perspective. And I've got a lot of really smart people on my team, but I told them, I opened up the Amazon page, where Amazon shows all their services, and there's so many of them, they've all got interesting names. And I said, if you asked anyone on my team, "what role does Amazon..." - and I just sort of scrolled down the list and said, "Amazon Lightsail. What role does it play in an attacker's playbook, and how they're going to launch their attack?" They would probably say, "What is Amazon Lightsail?".

Dave Bittner: [00:17:50] (Laughs)

Ryan Olson: [00:17:51] Because it's just - and there's new services all the time, where just - this is - the world of public cloud moves really, really quickly. There's a lot of new services that are popping up, and it really does require someone to sit down and educate themselves on, what are these things and what do they mean to my organization from a defensive perspective? Do they open a new hole? Or do they, quite frankly, make it easier for me to manage my infrastructure?

Ryan Olson: [00:18:12] A lot of this automation, a lot of the things that we've learned through the development of DevOps, can help a lot with security. Because you can suddenly, you know, destroy a virtual machine and restart a new one, and you don't really have to worry so much about, was there malware involved on that machine? Because you can kill it anytime you want to, and maybe you do. Maybe you kill those machines every day. But now you have to think, is the thing that's generating that virtual machine itself compromised? And how does that work, and how do we control access to it? So, it's just new concepts that people need to wrap their heads around in security

Dave Bittner: [00:18:45] Our thanks to Ryan Olson from Palo Alto Networks Unit 42 for joining us. The research is titled, "Cloud Security Trends and Tips: Key Learning to Secure Your AWS, Azure, and Google Cloud Environments." We'll have a link in the show notes.

Dave Bittner: [00:19:01] Thanks to Juniper Networks for sponsoring our show. You can learn more at, or connect with them on Twitter or Facebook.

Dave Bittner: [00:19:10] And thanks to Enveil for their sponsorship. You can find out how they're closing the last gap in data security at

Dave Bittner: [00:19:18] The CyberWire Research Saturday is proudly produced in Maryland out of the startup studios of DataTribe, where they're co-building the next generation of cybersecurity teams and technology. It's produced by Pratt Street Media. The coordinating producer is Jennifer Eiben, editor is John Petrik, technical editor is Chris Russell, executive editor is Peter Kilpe, and I'm Dave Bittner. Thanks for listening.