What is a good db for high update interval and aggregated reads?

I need to find the solution for the following problem:

A system that has 10,000s of parent object, each one contains a few hundreds of sub types with valus which might be updated a few times each second.

On the other end, I need to generate aggregated data That can be calculated 4-5 times a second. There can be 10,000s of this different aggregators.

The good thing is that rows are rarely added during runtime, only updated. Also, the value types are small int and floats.

What framework should I use?

I thought about using a key-value nosql method, but that doesn't enables me to do agg开发者_Go百科regated calls. Any idea is welcome, thanks

You should consider an OLAP solution, like SQL Server Analysis services. OLAP systems pre-calculate aggregations on several levels and facilitate therefore very high performance for scenarios such as yours.