Author: Tom Geller
Trilio seeks to enable updates, migration, and anomaly detection.
“We’re looking to have the term ‘data protection’ itself evolve. And I think we’re leading that charge,” said Trilio CEO David Safaii in an exclusive interview with Swapnil Bhartiya, founder and editor-in-chief or TFIR, at the OpenStack Summit in Vancouver last May.
Safaii’s vision for data protection encompasses both security and system administration, going well beyond typical backup-and-restore procedures. And he believes that OpenStack — which his company’s product TrilioVault targets, along with RHV — has a community prepared to appreciate these changes.
“OpenStack has come so far, with production workloads in a lot of environments,” he said. “As that happens, there’s this quick need for data protection from the backup and DR [disaster recovery] side. That’s sort of just ‘blocking and tackling’. But now that you have that box checked, you can take it one step further.”
Moving (and updating) made easy
For Safaii, easy migration is a natural next step. “[Trilio] lets a tenant capture the full application at an exact moment in time: The application blueprint, the metadata, and the data as a whole. We hold that in a qcow2 format you can store wherever you’d like. That gives you workload mobility. Not only are you backing up to a repository, but I’m giving you, the tenant, the ability to take that moment in time and recover it into a different cloud.”
Once your data-protection solution can do that, Safaii says, other opportunities open up. “A big topic within this community today is fast-forward upgrade cycles. How do I jump from Red Hat 10 to 13? You want the ability to capture all the tenants, back them up, fast forward, fast upgrade your cloud, and reinstate those tenants and those workloads. If the process isn’t clean, you have the ability to downshift again because you’ve already captured everything.”
Security, security, security
But the ability to roll back does little good when the archives themselves are corrupted, for example with ransomware. Safaii says Trilio is attacking that problem with tools to look inside the archives. “We have yet to announce this, but we’re talking about the ability to scan your backups,” he said. “So you’ll be able to check configurations, to check antivirus, to make sure that you have to get the best last-known state.”
Part of the challenge is to pinpoint exactly when a corruption first appeared. For that, Safaii is seeking to apply artificial intelligence to these scans. “With AI capabilities, with anomaly detection, we can look at these various moments in time and say, ‘They don’t look exactly alike. What’s changed? Should I rewind the clock? Should I start over again? Should I push things to test/dev environments sooner before pushing them out to production?'”
The ability to answer these questions will become more important as the “Internet of things” (IOT) grows, said Safaii. “You extend the cloud to the edge. Now you’ve got all these IOT devices capturing information. These are new open end points. You’re opening up a door because they’re sucking in all this information. So the evolution of security, the evolution of data protection, the need for backup is going to be critical.”
The future is containers
Safaii believes the trend toward containerization also increases the need for advanced data protection. “The container market feels a lot like where OpenStack was a couple of years ago,” he said. “It will need backup. When people first started down the OpenStack journey, they were saying ‘Oh, only stateless workloads will live here.’ Stateful workloads do find their way in. The same thing’s going to happen in the container world.
As for Trilio, Safaii is happy with his company’s place in the OpenStack ecosystem. “You know, backup is never a sexy thing,” he said. “But any time that there’s data generated in a point in time needs to be captured, data protection needs to exist. It’s fantastic to be the only native data protection solution in this market. But it’s a lot of work on our part. Every release we have to qualify against Red Hat, Canonical, Mirantis, SUSE, and upstream projects. We qualify from Kilo all the way forward because everyone’s in different parts of their journey. So it’s a lot of work on our part. But that’s what it takes to be the standard.”