The SolarWinds hack continues to make news and every day some more details reach the surface. Fortune 500 companies, large U.S. federal government agencies, security vendors and operating system vendors are affected and its putting the spotlight on how fragile and sensitive the software development cycle is. Using best practices and digital certificates make you check some boxes on your checklist, but this hack shows clearly the need of additional protection. The bottom line is – if your CI/CD pipeline and your code release is unsafe, the door is wide open and invites people with bad intentions. Furthermore, all the money and effort you put into automation, vulnerability scanning and DevOps training is wasted.
This blog post covers the hack as well as the trusted CI/CD pipeline on a high level. If you’re interested in the technical details please check our other blog post as well.
SolarWinds, a popular IT monitoring and management vendor, made clear that hackers breached its systems and inserted malware into the software build process of its Orion platform. As this was undetected for a long time, product updates shipped with the vulnerability for several months. That malware introduced a backdoor to access running Orion servers on customers infrastructures.
As many customers are using the Orion platform and do their job in having the software always updated, it’s estimated that roughly 18.000 customers installed the backdoor and many got breached.
Those affected include Microsoft, the U.S. Department of Homeland Security (DHS), and FireEye, a cyber security vendor which first detected the attack this month after the hackers stole proprietary security tools it offers to its clients.
The real problem of this hack is its scope.
The attackers found a way to legitimize the malware by injecting it into the build pipeline. That build pipeline started to produce digitally signed and trusted software for over 18.000 customers worldwide. Currently it’s not known, how many customers really installed the malware, but none of them would have questioned software signed and introduced by SolarWinds. And that’s the real problem, the build pipeline produces builds and they are digitally signed with the SolarWinds certificate trusted by the Certificate authorities in Microsoft Windows, macOS and your browser. Revoking the digital certificate revokes the good and the bad software, as you can’t differentiate between these.
This given trust in digital certificates and company brand is the real issue.
Let’s make it clear! SolarWinds didn’t run a security lacking, lousy maintained development lifecycle – they already did a lot to secure the build process and delivery by having architectural reviews, static and dynamic code analysis, open-source analysis, vulnerability checking and more. After the hack was realized, they added more security related steps:
It wasn’t a big open hole, the attackers used – it was a very sophisticated attack. Not entirely true, as there were some FTP credentials in a public Git repository. Nevertheless, it’s unclear if that FTP account was even part of that attack.
Of course, the big question to answer is, what can be done to avoid these widespread hacks in the future? Every vendor and security expert has its own favorite answer to it, but it all comes down to: There is a need for a comprehensive, holistic DevSecOps strategy, that aims to provide full observability:
There is no doubt that highly sophisticated hacker can use the tiniest flaw to get into systems and won’t stop to be very innovative. But that’s not a reason to give up, only to make it harder for them every single day.
A trusted CI/CD pipeline where every used open source project, artifact, or module (hopefully) is known and compatibility tested before being integrated into a release that is going to be shipped or deployed is the most important start.
Therefore, our customers use Codenotary to notarize every step and artifact within the pipeline by people (Developers, QA) or systems (vulnerability checker, compatibility testing software aso.).
You can think of having the unique identity of every single artifact and the bill of material of your build cryptographically verifiably stored on an immutable ledger. Using Codenotary Immutable Ledger you can choose to use a fast (up to 60.000 TX/min) worldwide distributed consortium blockchain (Codenotary Immutable Ledger CI/CD) or an ultra-fast (up to 10.000.000 TX/sec) on-premise deployment (Codenotary Immutable Ledger) or both.
When integrating the Codenotary solution into your CI/CD pipeline, you gain 4 main benefits:
The CI/CD protection can uniquely identify:
One benefit that should not be underestimated is the possibility to change the trust level of any digital object at any time and that everybody that should have access to that information can verify it. When using digital certificates you can only revoke that certificate for everything you ever signed, using Codenotary you can precisely change the status of a certain release or artifact.
Unlike the digital certificate approach using Codenotary you can untrust your release the moment you realize there is a malware or even a non-approved component in it without giving up your whole trust chain.
That allows fine grained observability of your build and release lifecycle as well as fine grained revocation of artifacts that have gone bad.
The combination of machine and user identity (separated from your SSO or inhouse user management), unique digital artifact identities and trust level (including a timestamp) is already powerful but worthless if stored in a normal database.
Key is to store that information in an immutable ledger, that is always accessible and maintains the full history of changes.
That is exactly what Codenotary provides – tamperproof trust at every stage of your pipeline from source to deployment and beyond.
Last but not least , we need to cover the physical access (SSH, Console, KVM) to the systems that run and distribute your code and builds.
Therefore, we’re always looking to integrate security technology that allows protecting physical access as well. Crypto cards to secure the cryptographic root signing by the server or integrating Intel SGX and its protected enclave mechanism are just some examples how we make sure, that even a local physical access to manipulate artifacts can be observed and alerted on or of course protected against.