Scaling the Number of Storages

By Patrik - August 18, 2015 (Last updated: April 3, 2017)

So, you’re building a large system based on Vidispine, with auto-import from many different storages. What can you expect from Vidispine in terms of scalability and what kind of issues could you run into?

We say that you should go for it. The database have no issues handling a large number of storages. What you have to watch out for is the load on the file system.

Every storage is monitored by its own StorageSupervisorWorker thread. A lot of storages will create a lot of threads, which is both good and bad. Good in the sense that you can throw cores to it, and the threads will distribute themselves, and bad in the sense that these threads has the potential of really hitting the file system hard.

So if you have many storages, and run into problems, consider limiting either the number of stats done on the local file system, or change the rescan interval. The stat limit can be set both globally and per storage basis, and system will make sure that none of them are exceeded.