One of the most common questions we have when talking to prospects is “how does ownCloud scale”. The answer? Quite well, thank you. Frank blogged about this recently to the community. From a commercial standpoint, I wanted to provide a few examples of how ownCloud scales for our customers.
ownCloud was designed from the beginning to operate in nearly every enterprise environment. This flexibility means our n-tier web application (LAMP stack) can be purely vanilla on a single server (such as our demo virtual appliance), or scaled in virtual and physical environments, made HA, and replicated to a second passive DR site. Frank already pointed to a great blog on how HA can work with ownCloud.
How do customers actually deploy ownCloud?
Typically customers install ownCloud as a load balanced n-tier app with a MySQL cluster, a clustered file system, and multiple ownCloud app servers. Most customers use F5 or NetScaler as their front load balancer, or even HA Proxy. ownCloud connects to virtually any storage option; from Red Hat Storage or Ceph, to DDN’s Web Object Store, IBM’s Spectrum Scale, or even NetApp. The flexibility of ownCloud allows our customers to choose the solution that works best in their environment, often something they already have.
How does it scale?
In this architecture, ownCloud scales quite well. A good practical example is Sciebo, which Holger discusses here. That architecture is based on our performance testing work with IBM, documented in our white paper. For other examples, you can find our white paper on how to size hardware in general with ownCloud that many find useful in discussing scale, and a reference architecture with performance results from an HP, Red Hat Storage and ownCloud here.
In short, ownCloud scales from the smaller business unit deployments, all the way up to the largest on-site enterprise deployments on industry standard hardware. If you would like to know more about how ownCloud could scale in your environment, please contact us at email@example.com or just fill out this form and we would be happy to answer your questions.