Hi,

Let me get my crystal ball.... OK, now let's try inlining.

On Tue, Apr 23, 2013 at 5:48 AM, Furkan KAMACI <furkankam...@gmail.com> wrote:
>
> * I want to measure how much RAM I should define for my Solr instances,
> * I will try to make some predictions about how much disk space I will need
> at production step.

This one is easy if your index is static or grows slowly.  If not,
you'll want to set alert thresholds on disk space free/used for
capacity planning/expansion purposes.
You probably saw threads about needing about 3x the disk space (3x the
size of your index) for about a week ago.

> * Maybe I will check my answer for that question: which RAID to use (or not
> use) etc.
>
> For that questions I got answers from mail list and I have some
> approximations about them. Also I know that it is not easy to answer such
> questions and I should test them to get more accurate answers.
> My question is that::
>
> What do you suggest me at pre-production and test step?
>
> * i.e. give much more heap size to Solr instances to calculate RAM

Impossible to tell precisely, but you can launch Solr, hammer it (next
bullet), look at your monitoring tool or just JConsole, ask the JVM to
run GC (you can do that from JConsole), observe heap once everything
has been fully loaded (for sorting, faceting, etc.).  That will give
you an idea of bare minimum heap.  Increase from here.  Don't expect
to find one magic number that will be good forever, because that won't
be the case (this is where keeping an eye with monitoring and alerting
comes into play) unless your system is completely static (static
index, same type, volume, and distribution of queries, etc.)

> * use solrmeter to test qps for your cluster

Sure.  JMeter or SolrMeter will do.  The latter is written by one of
the Solr guys and gives you more Solr-specific data, so +1 for that
one. :)

> * use sematext or anything else for performance monitoring etc.

I'm completely unbiased here, of course ;)
Yes, you need some sort of monitoring (+alerting) if you are serious
about your search in production.  If you already have something, hook
that up.  If you don't have anything or don't want to bother with
maintaining a monitoring system, get some SaaS, like SPM for Solr.

> I need some advices what to test, calculate, measeure etc. Also there was a
> question about Codahale metrics and Graphite. You can advice something
> about that too.

One of the main decision factors is whether you want the
responsibility of maintaining something like Graphite in house or give
it up and focus on your service/product.  The tendency seems to be the
latter, but there are still organizations who choose the former.

> PS: I use Solr 4.2.1 for tests but if Solr 4.3 becomes ready (if it is
> tagged at repository) I will use it.

If you are in pre-production and asking questions about memory and
disk, my feeling is you should wait for 4.3. :)

HTH

Otis
--
Solr & ElasticSearch Support
http://sematext.com/

Reply via email to