Bob Goldstein wrote:
With our current experience with our implementation, we have been
considering some sort of Federated solution – entirely separate
instances of RT that know about each other and that can communicate
effectively (transfer/link tickets).
Not sure what “entirely separate” means, if you can transfer tickets.
Yeah that was a bit unclear. We are considering an extension which
would allow you to “transfer” a ticket to an other instance (separate
database, etc) of RT.
I might be interested in that extension.
I had crafted a one-way extension, so that tickets from
a Clarify system could be transferred to RT. Basically,
Clarify would send email with a special format, and RT
would parse that email and create a new ticket, tracking
the Clarify ID in a custom field. (So any future appends
to the Clarify ticket could be sent to RT and appended
to the proper RT ticket.) However, this was one-way.
Clarify never received any updates when the RT ticket
was modified, and there was no back-transfer mode.
Depends on your need, of course, but an easy course
would just be to create an RT ticket, and have it contain
a web link to a ticket in the other RT instance. In my case,
though, we rarely need this, so I’ve ignored it.
We do set up entirely separate instances (same core codebase, but
separate databases, separate config files, separate customized code
(of which there is little) for departments. But they are truly
separate, no transfer of tickets. I tell people I don’t
want to set up multiple instances in a single department, in part
because they probably do want to transfer tickets, even if they
don’t know it yet.
Do you charge departments for the service?
No. But we do minimal support. It works for us because
RT pretty much works as needed, and consultants don’t really
need much training to use it. And the way I have the
core code shared, any mods I make to my own instance
can be shared across all instances (on the vendor branch)
or not (on the user branch), so there hasn’t really been
any call for customizations that I didn’t want for myself as well.
BTW, I did propose that the rest of the university replace
Clarify with RT, and that I’d be willing to do that for a fee.
The people involved really like preminum service, formal SLAs,
etc, which is ok but beyond my “free” resources. But so far
I don’t think they are interested.
How are you hosting the various instances? We’ve considered dedicated
servers, dedicated VMs, or a cluster of that services multiple vhosts.
At the moment, I have two boxes, one for apache and one for mysql.
But it was set up initially so that I could have multiple
web servers (via DNS rotation, or other load balancer), or
could put the db for each instance on a separate mysql box.
I can’t imagine a single instance so large that would overwhelm
a single mysql server (at least before we move to mysql 5.x clusters)
so we’re prepared.
bobg