SIOS High Availability Software
SIOS makes old-school node-based storage replication software for clustering and failover. SIOS has some other products, but this software is the core of what SIOS is good at.
I hadn’t heard of SIOS before #SFD23, which feels weird, but it seems to be because it’s mostly used for smallish SQLServer or SAP databases. It’s apparently very popular in SQLServer land. During my DBA days I mostly dealt with Ingres, Oracle, and PostgreSQL so I didn’t encounter it. I was also mostly in large enterprise, which tends to use other approaches like SAN-based replication or more complex and expensive options.
Still, it’s apparently well regarded, and seems to function a lot like Veritas Cluster Server and Veritas Volume Replication (VCS and VVR for those old enough to remember it) which I did use a lot about 20 years ago. SIOS does synchronous and asynchronous replication, with store-and-forward of changes if there’s a communications interruption (same as VVR used to do).
SIOS may seem simplistic compared to all the fancy features of some high-end storage options that I’m more familiar with, but there’s a lot to be said for something simple and reliable. Particularly when you’re talking about important data.
Options that are solid, well-tested, and reliable are great. Options that are also easy to reason about are even better, because it helps you be sure that you haven’t overlooked something important that will come back to bite you hard when things go awry. All steely-eyed storage admins know that things will go wrong, usually at the worst possible time, often in the worst possible way. Keeping data safe and sound at these times is what you need, not fancy blinkenlights and overwrought marketing claims.
This is particularly important when you’re in smaller environments that don’t have armies of specialised storage people to water and weed the complex storage behemoths.
SIOS is proving quite handy in the cloud as well because people have lots of nodes (aka instances) but don’t tend to use storage arrays as much. Using a software-based clustering and replication method makes a lot of sense when everything else you’re doing is software anyway. There are no SANs to integrate with in a lot of systems that people are setting up in the cloud.
SIOS seems a pretty good option for people doing 2- or 3-node clusters of SQLServer and don’t need all the gee-whiz fancy scale-out distributed systems theory things that occupy a lot of marketing space. Worth checking out if this is what you’re looking to set up.