Beside of the classical Benchmarking approaches to compare databases, there are some important topics to consider when choosing a Triple Store. Topics such as security, administration, hardware, maintenance, reasoning options, scalability and integration with other tools are discussed in this presentation.
I have been working for the past 3 years in productive environments using Triple stores, mostly in the pharmaceutical industry.
One use case is an ontology oriented search tool, where data validation, versioning, response time to Sparql queries and replication are important issues. The second use case was more about finding unknown relation in a large data set (a few billions of triples). We did some benchmarks with the database listed above and this is the experience that I would like to share with you.