UCCX 7.0(2) HA to 8.0(2) HA

[cbelcher]

Hey guys I've got a 24x7 Call Center that is running a pair of UCCX 7.0(2) servers.  Anybody have a procedure that I can upgrade with zero down time?

The Upgrade guide clearly states and I quote

• Be aware that your call center remains down during the upgrade process.

There's got to be a way?

Thanks in advance!

[anchoudh]

Hi cbelcher,

You can not perform this upgrade with 0 downtime, however you can reduce the downtime with the below few points.

1. Keep your CUCM already upgraded to a higher version which is compatiable with UCCX 8.x, so that when you really put your UCCX 8.x in production just perform the switch version of the CUCM , to reduce the CUCM upgrade time.

http://www.cisco.com/en/US/docs/voice_ip_comm/cust_contact/contact_center/crs/express_compatibility/matrix/crscomtx.pdf

2. If you have the same type of hardwares in the lab, you can do a upgrade in the lab (using both CUCM and UCCX replica's) and directly unplug the 7.x series boxes and out in the 8.x servers.

3. Also, while performing the UCCX upgrade from 7.x to 8.x HA deployment, the upgrade on 8.x happens only on the first node and once it completes than you will bring up the second node of UCCX 8.x and need to perform the add to cluster operation to make it HA setup. So during this time let your UCCX 7.0(2), second node continue to handle the calls.

In the above scenario (3), please note that all the call record data (Historical) after the backup taken using PUT tool will not be migrated on the 8.x setup, because PUT tool backup takes the backup upto date before taking backup

http://www.cisco.com/en/US/docs/voice_ip_comm/cust_contact/contact_center/crs/express_8_0/installation/guide/uccx802ug.pdf

4. Keep all the licenses ready, so that there is no delay during the upgrade time.

Hope this helps.

Anand

Please rate helpful posts by clicking on the stars below the right answers !!

[Jonathan Schulenberg]

The crux of the problem here is the transition from Windows to Redhat Linux. There are ways to minimize the downtime but not eliminate it entirely.

Remember that CUCM has call distribution options too. A batch of spare phones and a hunt pilot/list/line group can sometimes cover the CCX outage if you do the work during the lowest time of the day/week/month.

You'll have to choose which is more important: IVR/ACD uptime or continunity of reporting data. Here are three options; the first two keep the database records contiguous throughout but have downtime. The third would allow you to flash-cut over but will have data loss.

• If you follow the documented upgrade process you end up starting your maintenance window, running the PUT tool to tarball the database, and then formatting the server to install Redhat. At the tail end, in excess of three hours later, you then restore the database to bring the system back online. Once that's done you then need to update CAD. This keeps the database complete but you'll be down for a few hours.
• If you have new hardware or are running on VMware (not supported until 8.0 though) you could install the new CCX server in a silo. It would have to be totally isolated and mirror production: all of the same values for the platform (i.e. hostname, IP, DNS, etc.) This can take some staging (e.g. fake DNS server) to pull off though. This lets you cut out the downtime from the OS installing: install 8.x, disconnect the 7.x servers and attach the new ones to the real network, then start your restore. There is still an outage here but it's much shorter. Again, CAD has to update after you're done.
• You could take the second option a step further and entirely rebuild the CUCM/CCX servers in a silo. The advantage this would have is that you could actually do the restore and potentially even test before cutting it into production. This comes at the expense of your database though: from the time you do the PUT backup (and the DRS Backup in the case of CUCM) until the time you cut it to go live all data will be lost.

Lastly, I said that you couldn't eliminate the downtime entirely due to CAD. Even if you went the route of the third option all of the CAD clients must upgrade before agents will be able to login and take ACD calls again.

As a footnote: Cisco only sanctions the first option. The second two can work but don't call TAC if you get stuck halfway down the path on those.

Be advised that CCX 8.0.x has had a serious history of being buggy. I cring whenever I see it in the field. 8.5(1)SU3 is reasonably stable but the QA overall isn't great. Be prepared to spend time troubleshooting defects after you upgrade to any 8.x CCX release.

Please remember to rate helpful responses and identify helpful or correct answers.

[cbelcher]

Anchoudh and Jonathan thank you both for you quick replies.  I think you guys both nailed it.

I would love to hear your thoughts on this plan TAC has laid out for me, if you would be so kind

Since you’re using the same hardware, here’s the action plan I’d suggest you follow, when performing the upgrade :

1.      Take the backup from UCCX 7.0 PUB using PUT – this node should have mastership at that time.

2.      Failover to the SUB.

3.      Re-build the UCCX PUB to 8.x.

4.      Perform initial config of the UCCX 8.x PUB and here, the PUT restore.

5.      Shut down the UCCX 7.0 SUB, re-build it to 8.x

6.      Perform initial config of the UCCX 8.x SUB, and run the client config tool from the UCCX 8.x PUB.

7.      Upgrade the following:

a.       Cisco Agent Desktop (will need to download from server if agents do not have Windows admin access)

b.      Cisco Supervisor Desktop (will need to download from server if supervisors do not have Windows admin access)

c.       Cisco Desktop Workflow Admin (will need to download from server)

d.      Cisco Unified CCX Historical Reporting Client (will need to download from server)

e.       Cisco Unified CCX Editor (will need to download from server)

In this action plan, you will face two instances of downtime :-

1.      When the failover happens at step 2.

a.       This will be a minor failover and will take at the most 1 to 2 minutes.

2.      When the UCCX 7.0 SUB is shut down at step 5.

a.       This will be a longer downtime, since the agent desktop software will need to be upgraded.

b.      You have two options to enable agents to work during the upgrade –

                                                              i.      IPPA phone service on agent phones.

                                                            ii.      Create a hunt group/line group on CUCM and ensure that the agent non-IPCC extensions are present here. This will have limited support but is a temporary workaround.

[anchoudh]

Hi cbelcher,

Yes, that's right.

One more thing to note is, once you again perform the UCCX 8.x HA setup, don't forget to rerun the CAD client config tool (once only), and also the PostInstaller.exe from the client box (where there was no information of the second node of UCCX 8.x second node ip), please add it, so that in case of any failovers of UCCX 8.x later point of time, CAD|CSD will connect to second node automatically.

Hope this helps.

Anand

Please rate helpful posts by clicking on the stars below the right answers !!