[ome-users] Resources needed for providing an OMERO service

[Douglas Russell]

Hi,

Hardware wise it is difficult to estimate as it will really depend on the
number of users, type and volume of data, and the workloads introduced by
whatever workflow you decide upon.

As far as provisioned human resources though, I would say that this is not
too different than provisioning any kind of managed service. Installing,
patching, updating, adding extensions and similar, if your IT department is
good, should only be about as time consuming as the constituent parts (e.g.
database schema upgrades, nginx configuration, etc). The configuration and
optimisation of the OMERO service itself could be quite time consuming if
your IT department have to learn all about it which they probably would not
have to do with the database and nginx.

I feel like the biggest difficulty we have faced at HMS is in integrating
distributed components together. We have run into issues that existing
services may not suffer from, not because of faults in OMERO, but simply
because the resources we are making use of have not been integrated on-site
in this way before. For example, we hit a problem where firewall timeouts
between OMERO and the db and filesystem share were causing system failure,
this was difficult to track down. This is a long way of saying that I think
the ideal way to get started is to minimize the complexity of the
installation. If having OMERO, OMERO.web, database and block storage (SAN
or local) in one machine is possible for the workload you are planning, I
would go with that. If you then need to start making use of NFS for the
filesystem, or a non-local database server for the db, or a second server
for OMERO.web, then I would introduce those later as required, but not all
at one time so that each can be tested independently with the otherwise
operational monolith.

Finally, I would add that provisioning a service like this really requires
heavy buy-in from your IT department if it is to be reliable for large
scale use. Here at HMS, our quad wide rollout has (roughly speaking) ~1/2
project manager , ~1/2 developer (need for this depends on customizations I
guess, but if this person has OMERO expertise it helps a lot to have them
as a consultant to IT) and ~1/4 IT devops. The project manager also has
support personnel who are involved in the logistics and the devops has
specialists in storage, networking, etc to call upon. We also have a
Glencoe Software support contract which is also extremely valuable,
especially when we run into new errors that have not been seen before which
have often proven to be caused by the local deployment environment.

Sorry it is so anecdotal, but I hope it is useful.

Cheers,

Douglas

On Tue, 21 Feb 2017 at 09:37 Sascha Neinert <sascha.neinert at uni-koeln.de>
wrote:

Dear all,

i did not manage to find much information about the resources that would be
needed if one would want to provide an OMERO service - there are example
production server set-ups so we could estimate the required IT resources (=
hardware), but we wondered about the amount of time needed for running the
system („human resources“): updates, patches, error handling, maybe some
performance analysis, maybe some optimisations, maybe installing an
extension or two…
Any experiences on this would be most welcome!

Thank you, best regards,
Sascha Neinert


--
Sascha Neinert
CECAD RRZK IT Service
Regional Computing Center (RRZK)
University of Cologne
Phone: +49-221 478-84051 <+49%20221%2047884051> / +49-221 470-89624
<+49%20221%2047089624>
Mail: sascha.neinert at uni-koeln.de





_______________________________________________
ome-users mailing list
ome-users at lists.openmicroscopy.org.uk
http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openmicroscopy.org.uk/pipermail/ome-users/attachments/20170221/c86a5aac/attachment.html>