[rt-users] Scalability question - RT appropriate for 50K+ requests per year?

[Kenneth Marshall]

On Thu, Jun 11, 2009 at 10:16:48AM -0400, Lander, Scott wrote:
> Folks,
> 
>      We are taking a look at RT as our helpdesk, and the basic functionality seems fine.   However, I am wondering how it does under a fairly heavy load (ok - maybe it isn't heavy - but, I think of it as fairly large)  of upwards of 50K requsts per year, about 200 requests per (work) day.  Is anyone running that large a system?
> 
>      If so, could you describe your system - CPU/disk/memory/OS, etc, as well as how it performs?   Anything you wish you had done different because of the size?    We envision keeping tickets (closed) at least 3 years, so we would have upward of 150K tickets in the system after some time, if that matters.
> 
> 
> Thanks for any input.
> 
> Scott
> 

Scott,

We have been using RT here for the past few years and our ticket
load is close to what you describe above. We currently have just
over 200k tickets in the system. We are currently on the initial
hardware purchase 5 years ago of 2 Dell 2650's with a internal
RAID controller to manage the disks. We purchased two to provide
a redundant environment that would allow our ticket system to
survive the failure of a complete box. On the assumption, that it
will be hard to manage repairs to critical systems without an
out-of-band trouble ticket tracking system to consolidate work.

The web frontend runs on one of the boxes and the PostgreSQL DB
runs on the other box. The DB is replicated to the frontend box
and a copy of the frontend software is available on the DB backend
to allow it to also serve as the frontend. The boxes each have
4GB of memory, 2 3.0GHz Xeon, 6 x 146GB SCSI + battery backed
NVRAM cache. This system is currently running RT 3.4.5pre1 and
PostgreSQL 8.3.5.

We are currently preparing to upgrade to RT 3.8.x with a
PostgreSQL 8.3.x backend on 2 similarly configured Dell 2950's.
My comments about the hardware are that it is easily able to
handle the number of requests that you need. I can not really
tell you how little resources you can get away with using.
Others have been using VM's to good effect. I would say that
one of the major complaints that we have had concerned the
ability to perform full-text searches within the attachments
and large custom fields. I would recommend a backend DB that
supports that functionality: Oracle and PostgreSQL.

Good luck and I hope this helps.

Regards,
Ken