IPT Lab setup

[D2406-LTO3]

Hi

I guess I'm looking for advice and opinions really on the best way forward with creating an IPT lab. Currently in our enterprise environment we have a Publisher and 4 subs, 2 Unity Servers, 2 Arc Connect servers and 2 UCCX servers.

There's a desire to recreate our live environment in a lab and it has been suggested that we import (essentially) clones of all these servers into a lab environment. Personally speaking I can't see the real benefits of this. The lab environment would not a complete replica of the companys entire network so you would end up having to reconfigure the cloned live to work in the lab.....at which point it doesn't become an accurate reflection of the live environment.

I'd have thought that 'fresh' builds of all the servers in a lab environment will give you all the testing functionality that you require. With software upgrades, you'll be looking to test the interaction between (in our case) CUCM, Unity, Arc and UCCX and you don't really need a true representation of the live enviroment for that. Surely as long as the versions are the same in both environments, that will give you a good testing bed.

How has everyone else tackled the issue of creating a lab environment to test upgrades, configurations, etc? Have you tried to replicate your live environment or just gone for a fresh build lab environment?

Cheers

[Rob Huffman · Accepted solution]

Hi There,

I totally agree with your thinking here

Our Lab "mimics" our production environment in

Build and Versioning only. The "clone" idea is pretty

great in principle but nearly impossible to keep going

on a day to day basis. Plus there are things that you want

to try out in the lab that will never happen in the production

environment, so things will always be a little out of wack!

Just my 2 cents,

Rob

PS: the licensing for a lab can be a little tough as well, so trying

to do duplicate builds would be nearly impossible.

[Tommer Catlin · Accepted solution]

i agree with Rob on this.  Also to note, there is no need to duplicate a subscriber.  You only need a publisher.  Same for ARC and Unity. Ideally, you want to have the same hardware and software versions.  This is usually where your lab testing happens.   For CUCM, its more testing Phone loads, CUCM updates/upgrades and possibly build out scenarios of internal testing such as call fowarding, shared lines, etc.

ARC is tricky.  I would not want to replicate this in the lab at all.  To begin with, its a pain in the arse to setup for production, much less mimic in the lab.  You can however at least virtualize ARC on VMware.  So you should be able to snap copies of it

Unity again will be tricky because if you are using Unity/Windows..  this is all tied to AD/Exchange and no real way of replicating this without AD.

[David Hailey · Accepted solution]

I have attempted both and can definitively say that trying to keep "clones" of your production in a lab environment doesn't work and isn't worth the time.  However, there are a lot of benefits in having a hardware clone environment - i.e., at my last employer, we built one of the first "superclusters" as it was called initially - basically it was a maxed out cluster with 1 pub, 8 subs (4 active/4 backup), 2 TFTP servers.  We did a 1:1 hardware replica of this in our lab.  The dial plan was the same but we didn't try to replicate everything - we used it to test all facets of the environment from phone firmware upgrades to CCM upgrades (this was 4.1).  Long story short, we actually were able to use this setup to pre-build a production upgrade using spare disks and a cloned network setup.  There is a lot of value here.

There is also a lot of value in going with VMWare.  Almost all of the applications can be virtualized...not all...but most.  I have CUCM, Unity w/Exchange 2003, Unity w/Exchange 2007, Unity Connection, and other apps all virtualized in my lab.  In terms of testing software or configuration changes, this is a nice lightweight way to go.

It really depends on budget.  Like T Cat said, you can get away with a Pub.  And, like Rob said - licensing on a large cluster for a lab "clone" may come with a heavy price that isn't worth it.  The most important thing is that you push for a lab, build one, and use it....I can't tell you how many companies don't do this and it's a mistake not to.

Please remember to rate helpful posts!

[Rob Huffman]

Hi Folks,

TCAT and Hailey....really great responses here my friends! +5 points for

each of you. It's nice to hear what others think about certain scenarios and

to see what you are doing in the Lab is just an added bonus.

I'm not sure how people can run this stuff without a Lab, even with a pretty good

Lab we've run into many issues over the years, I'd hate to think of where we'd be without

it

Cheers!

Huff

[D2406-LTO3]

Thanks for all you replies, it's certainly food for thought. The input is greatly appreciated. Like Rob says, it's interesting to see how everyone else has tackled the lab setup issues. I guess at the end of the day the scope of the lab comes down to time and money!

Cheers guys!