Tuning the JVM

Hardware

SkyVault degrades gracefully on low-powered hardware, and small installations can run well on any modern server. However, for optimum performance, we recommend the following:

• Use 64 bit systems only.
• Use a system with a clock speed above 2.0 GHz.
• Reserve enough RAM for your operating system beyond the memory required for your JVM.
• Keep search indexes on your local disk instead of on network storage.

Disk space usage

Use a server class machine with SCSI Raid disk array. The performance of reading/writing content is almost solely dependent on the speed of your network and the speed of your disk array. The overhead of the SkyVault server itself for reading content is very low as content is streamed directly from the disks to the output stream. The overhead of writing content is also low but if Solr is installed on the same machine, additional overhead should be allowed for the indexing process. For more information, see Calculate the memory needed for Solr nodes.

Virtualization

SkyVault runs well when virtualized, but you should expect a reduction in performance. When using the rough sizing requirements given, it might be necessary to allocate twice as many resources for a given number of users when those resources are virtual. Para-virtualization, or virtualized accesses to native host volumes do not require as many resources. Benchmarking your environment is necessary to get a precise understanding of what resources are required.

JVM memory and CPU hardware for multiple users

The repository L2 Cache, plus initial VM overhead, plus basic SkyVault system memory, is setup with a default installation to require a maximum of approximately 1024 MB.

This means that you can run the SkyVault repository and web client with many users accessing the system with a basic single CPU server and only 1024 MB of memory assigned to the SkyVault JVM. However, you must add additional memory as your user base grows, and add CPUs depending on the complexity of the tasks you expect your users to perform, and how many concurrent users are accessing the client.