RavenDB or SQL Server 2008 Filestream

A project I am working on at the minute involves taking a massive set of results stored in a file and doing calculations based on those results. I have been looking at using either RavenDB or using the SQL2008 with the Filestream as the storage for the results. I am not sure which technology would be best suited to my problem.

Has anyone any 开发者_如何学编程views on which of these approaches is best for massive storage and possible searching of results?

I tried to insert a lot of data to Raven and was also disappointed in the first place, because I did it all form one thread. It turned out that inserting to SQL Server was faster (without doing much configuration on both servers).

Then I realized that I probably misused Raven, which is - as far as I understand it - designed to handle massive parallelism. So I played a bit with TPL opening a lot of threads / processes doing inserts in parallel. Raven handled this very well - but keep the "eventual consistency" in mind - not all the results showed up in the database immediatly but with a short delay.

I think a direct comparrison is hard unless you adjust your architecture to fit the choosen storage technology.