maximum database sizes using GridSQL and hardware recommendations

Hi,

I was just wondering what database sizes people have got to using GridSQL, we're looking at using it to speed up an existing 4Tb (growing by 0.5Tb/month) postgres data warehouse. Also any minimum hardware requirements? Obviously the gruntier the machines the better, but we're trying to do stuff on the cheap.

Thanks

Alan

Hi Alan,

Your current database will load up and be usable just fine. Also, we are in the process of working with a few different folks to roll out into production data warehouses much larger than that.

Your minimum hardware requirements will depend on what response time you expect, how many concurrent users you have, what your schema is like, how you can partition it, what your queries will look like and if they can take advantage of partitioning (both across nodes, but also, importantly, using constraint exclusion partitioning such as creating subtables for ranges of data).

GridSQL will help with a cost-effective solution. For your hardware, you could consider setting up multiple logical nodes per physical server, but make sure you have dedicated storage for each of the logical nodes so that they don't interfere with each other. Think of getting dual or quad core servers where you could assign a dedicated RAID array to each (for the most bang for your buck), and get multiple servers.

I hope that helps.

Regards,

Mason Sharp

Hi Alan,

It could be useful to test performance on EC2 (see aws.amazon.com) before you buy hardware. There you can rent hardware with varying configurations. One problem, though is that they do not provide 1 gigabit networking, which could be a bottleneck.

Regards,

David