While an IT team might view OpenStack and its open source model as liberating, the platform's backup and disaster recovery capabilities still fall short for some. Fortunately, several vendors and the OpenStack community at large continue to work toward a more robust and user-friendly experience in these areas.
OpenStack backup itself is a separate issue from using the platform to back up application workloads, said Carl Brooks, analyst at 451 Research. Admins, for instance, have to manually put cloud backup and recovery processes in place for each OpenStack component they use or look for a software tool that does that.
"It's all part of the grunt work of setting up an IaaS environment properly," Brooks said.
Depending on the details of the implementation and tool sets involved, OpenStack backup challenges vary. OpenStack Cinder natively has a command-line interface and an API for cloud backup and disaster recovery (DR), said Laura DuBois, analyst at IDC. And there are numerous backup drivers from different commercial and open source products, including:
- Google Cloud Storage;
- POSIX (Portable Operating System Interface) file systems;
- OpenStack Swift; and
- IBM Tivoli Storage Manager.
"These drivers just provide basic functionality to do incremental, differential and full backups of a volume," DuBois said. "How effective these drivers are will vary on the implementation."
When to turn to third-party tools
Thomas LeeCEO, Wingu
Thomas Lee, CEO of Wingu, a cloud infrastructure company based in South Africa, built his company's platform on OpenStack and knows the challenges of backup and recovery firsthand. The company evaluated VMware and other commercial offerings but selected OpenStack due to its open source foundation. While the OpenStack Horizon management dashboard is fairly straightforward, the implementation of backups can be challenge, according to Lee.
"Unless you are able to write custom code, you won't get to an easy solution for conducting regular backups and managing the versions," he said.
After evaluating some cloud backup products from Symantec and Veeam, which didn't have a strong OpenStack focus, Wingu looked at Trilio. The vendor's TrilioVault offering was a good fit for Wingu and its customers, enabling them to still use the familiar Horizon dashboard, while eliminating the labor-intensive steps in the back end normally required to implement an OpenStack backup strategy, Lee said.
Roughly 50 of Wingu's customers now use Trilio to manage their backups. To ensure maximum redundancy on the Wingu network, the company supports traditional block storage, using Cinder, and then physically separates object storage with Swift.
"We back up from block storage onto the object storage because it is reachable from the same network," Lee said. "But because it is different technology, if there were ever a catastrophic failure, one won't take down the other."
Another company focused on the OpenStack backup challenge is Awnix, which offers ARChive, a tool that provides automatic, policy-based backups and DR in conjunction with OpenStack APIs.
In general, though, OpenStack deployment and management are still a struggle for many enterprise customers, both because of a lack of skills in the market and the platform's software-defined model. However, as OpenStack continues to evolve and the ecosystem around it expands, the business case for the technology could grow.