Cisco Logo


Inside Cisco IT

In an earlier post, my colleague Reid Bourdet described how we migrated our largest Cisco Unified Communications Manager (Cisco UCM) cluster to a virtual machine environment running on Cisco Unified Computing System (Cisco UCS) servers.  This was the 19-node (server) Cisco UCM cluster that serves the Cisco headquarters campus in San Jose, California; and we completed the migration over a weekend.

What makes that move even more interesting is that we’re nearly done consolidating 5 separate clusters into one virtual environment, and reducing the total number of servers by a factor of four. Virtualization on the Cisco UCS hardware allows us to consolidate multiple UCM nodes on a single blade.  In this post, I’ll provide more details about the scope of this migration, the results we’ve gained, and how we’ll continue migrating other Cisco UCM clusters to Cisco UCS servers around the world.

UCM Clusters in the Cisco San Jose Campus

In addition to the Cisco UCM cluster supporting the San Jose campus population, our San Jose location is home to four other Cisco UCM clusters, and by the end of the current fiscal year, we will have migrated all of them from 116 Cisco MCS 7845 Media Convergence Servers to 32 Cisco UCS B200 M2 Blade Servers, split between two local data centers.

Cisco UCM Cluster

Before San Jose
Cluster Upgrade

After San Jose
Cluster Upgrade

San Jose campus UCM Cluster
25,000 users
45,000 configured phones

48 Cisco MCS 7845 Servers

Multiple virtual machines shared across
32 Cisco UCS B200 M2
Blade Servers

Western United States UCM Cluster
6500 users
15,000 configured phones

27 Cisco MCS 7845 Servers

Western United States Contact Center Cluster
2000 agents
4000 configured phones

17 Cisco MCS 7845 Servers

Global Telepresence Cluster
2000 endpoints

5 Cisco MCS 7845 Servers

WebEx Conference Cluster

19 Cisco MCS 7845 Servers

Total Number of Servers

116 Cisco MCS 7845 Servers

32 Cisco UCS B200 M2
Blade Servers

 

Although migrating such vital IT systems to a new server platform requires careful planning and a large coordinated effort, the benefits make it worthwhile.

Cost and resource savings. The compact design of blade servers and the ability to host multiple systems on a single blade can have a tremendous payback in resource savings and cost reductions.

Data Center Resource

Cisco MCS 7845 Servers

Cisco UCS B200 M2
Blade Servers

Net Savings

Electrical Power

39.21 kW

7.94 kW

80%

Rack Unit Space

232 RU

24 RU

90%

Cabling

348 network cables
116 power cables (minimum)

36 network cables
16 power cables (maximum)

90%

 

By the end of 2012, we’ll complete projects to migrate the remainder of our production UCM clusters around the world. The larger clusters (such as in Amsterdam) which reside in data centers we migrate to Cisco UCS B-series blades because of the efficiency gained with the higher density of virtual machines supported.  With B-series blades, we also use SAN storage to gain the benefits of separating storage from computing.  For smaller clusters (for example in Sydney and Singapore) which reside in telecom closets or small server rooms we migrate to Cisco UCS C-Series Rack Servers. The rack servers are more suitable for the smaller locations because of the fewer number of required virtual machines that need to be supported.

For more information about our UC-to-UCS migrations, check out these posts:

Comments Are Closed

  1. Return to Countries/Regions
  2. Return to Home
  1. All Inside Cisco IT
  2. Return to Home