Capacity Planning Guidance

[Sohail Aboobaker <sa...@gmail.com>]

Hi,

We are adding Solr to the mix of other servers in our environment. We are
looking for some guidance on hardware requirements for supporting following
loads:

a. Two Schemas --> 1 with 10,000 documents and other with 2 Million
documents.
b. Index Update Requests --> 20,000 per day. 50% of these are in batches.
c. Facet Queries on first schema --> 50,000 per day.
d. Facet Queries on second schema --> 40,000 per day.

What type of memory and CPU allocations should we do for the above load in
production and in a development environment.

Thank you for your help.

Regards,
Sohail

Re: Capacity Planning Guidance

[Erick Erickson <er...@gmail.com>]

This question, reasonable as it appears, is just unanswerable in the
abstract. About
all you can do is prototype and test. Take "facet queries". The
hardware requirements
vary drastically based on the number of unique values in the field(s)
you're faceting on,
as well as whether they're multi-valued or not....

However, your data set isn't very big as far as document count is
concerned (although
how big are those documents? 1K? 1G?).

I'd expect, assuming that you're not talking very large documents,
that you could
fit this on any reasonable machine and test, see jMeter or SolrMeter. I've put
11M documents on my 2009 laptop easily (Wikipedia english). Admittedly just 5M
of them have text.....

At any rate, set up jMeter or SolrMeter on a test machine and keep
adding documents
until you strain it and go from there.

Best
Erick

On Thu, Jul 12, 2012 at 5:46 AM, Sohail Aboobaker <sa...@gmail.com> wrote:
> We will be using Solr 3.6

Re: Capacity Planning Guidance

[Sohail Aboobaker <sa...@gmail.com>]

We will be using Solr 3.6