Redis To Go vs Websolr for faceting

I'm trying to find the mos开发者_运维百科t performant solution to a project that contains a large dataset. I would like to filter the dataset with faceting.

I'm running in the cloud so I would be using Redis To Go or Websolr. Sunspot has faceting built-in which I'm tempted to use for that reason alone. However I have my eye on performance and I'm wondering if well-formed Redis sets for the faceted attributes could give a performance boost.

How would these two solutions compare performance wise?

You say you have your eye on "performance" — perhaps you can clarify what exactly that means to you? Without any insight into your user stories, I would say that both Redis and Solr would be completely appropriate tools for this job from a performance perspective.

For starters, if you ever want to combine faceting with full-text keyword search, then Solr is the winner, no question. We'll just assume that's not the case here.

The biggest tradeoff from my perspective would be the number of lines of code you'd need to implement faceting with one or the other. Solr support for faceting will work for you 'out of the box' whereas indexes and facets on top of Redis would require a fair investment.

Whether that investment in Redis is worth the effort comes back to the question of how you're using these facets and how you're defining "performance." If you want to justify Solr over Redis, you're probably going to want to define and benchmark some real metrics.

Oh, right, full disclosure: I'm a cofounder of Websolr, and a maintainer of Sunspot, so there's that. But I use a bit of Redis in my own apps and I'm not afraid to give Redis its due ;)