We are constantly searching for ways to increase the reliability of our business voip phone service and our business internet service. Today's article focuses on recent changes to enhance the reliability of our business voip phone service. This infrastructure plays a role in providing our Virtual PBX service as well as our Integrated Voice and Data Solutions. We maintain a cluster of geographically disparate VoIP servers, each of which has its own local storage responsibilities. This storage has been off-loaded to a separate cluster, designated solely for providing highly available file storage. This storage cluster does not have a single point of failure, and in fact it is designed such that three distinct nodes need to fail before any service is affected.
This was accomplished with the help of the folks at Gluster who have developed:
an open source, distributed file system capable of scaling to several petabytes (actually, 72 brontobytes!) and handling thousands of clients.
Gluster was recently acquired by Red Hat which means that we can expect continued support and development. Scalability was a crucial component of the decision to choose this architecture. Call Recording, a feature of our Hosted PBX service, is growing rapidly. Part of the call recording offering is storing the call recordings indefinitely. This means our storage solution needs to be Highly Reliable as well as Scalable. A distributed file system consisting of commodity servers will allow our storage offering to expand to meet the increased demand we are seeing for Call Recording. Some other companies who have migrated to Gluster include Deutsche Bank, Samsung, Autodesk, BAE Systems, Barnes & Noble, Dana Farber Cancer Institute and Pandora.