Page History

...

If VIVO's dynamically-generated pages do not exhibit acceptable load times, you may wish to enable HTTP caching.  See Use HTTP caching to improve performance.  With this configuration, subsequent requests for pages whose contents have not changed will result in those pages being served directly from a cache instead of being regenerated from data in the triple store.

Alternative triple stores

While VIVO is tested with and configured by default to use Jena's SDB triple store with the MySQL database server, VIVO also includes support for TDB and Virtuoso as well as the ability to connect via HTTP to a SPARQL 1.1-complaint triple store.  Use of a different store may yield performance improvements, offer additional possibilities for performance tuning, or enable features such as clustering and load balancing.  In addition, configuring SDB to use a database server other than MySQL may offer advantages for your installation.   Note that some of the SPARQL queries employed by VIVO in page rendering have been optimized for SDB/MySQL with somewhat unusual use of UNION.  These queries should be modified for optimum performance with other stores that do not exhibit the same quirks.  

Misbehaving robots

In some cases, poor VIVO performance has been traced to search engine robots that either ignore or misread directives in VIVO's robots.txt file, or which issue requests for large pages at a rate that greatly exceeds the demand otherwise encountered in typical production use.  If the search engine in question is not critical to VIVO's visibility, it may be advisable to restrict access to the associated robots.  In some situations, institutional search appliances are responsible for the excessive server load.  Here, discussions with local IT staff may be warranted.

...