The promise of UCS
The principal appeal of UCS is simplicity. "It's the right choice when you're looking for simplicity; ease of integration, ease of support, ease of acquisition," said Greg Schulz, founder and senior analyst at The Server and StorageIO Group, a technology analyst and consulting firm headquartered in Stillwater, MN. "Everything is pre-packaged, pre-configured, pre-integrated."
But such simplicity isn't just a matter of writing one check. A UCS installation (sometimes dubbed a "data center in a box") allows management of servers, storage and networking through a single interface. This eliminates the need for administrators to master a different interface for each resource and stops the tendency for different administrators to specialize in different aspects of the data center. In effect, fewer administrators can handle all of the infrastructure's provisioning and management tasks effectively. A single interface can also provide more automation and cross-check provisioning choices, so management is more efficient with fewer errors that can waste resources or open potential security holes.
The simplicity and management ease of UCS also provides more convenient scalability in virtualized environments than might be possible in most heterogeneous data centers. New computing resources can be brought online quickly and added to available resource pools.
"I needed to add 16 additional ESX hosts to my VMware environment, and I needed to be able to do it very quickly to meet project deadlines," said Michael Heil, technical infrastructure manager at Moses Cone Health System in Greensboro, NC. "UCS was a great fit because I had a large influx of compute capacity in a short time period."
UCS reality check
But while the potential advantages of UCS are compelling, the technology may not be appropriate for every organization. There are numerous issues that must be considered before making a move to UCS.
The most obvious issue is cost; this includes the initial UCS acquisition costs, installation and initial setup costs, decommissioning costs for displaced equipment, training costs, and ongoing service/maintenance costs. This makes UCS a more difficult choice for large organizations that have already made a significant investment in their existing infrastructures, though larger organizations can justify UCS deployments in new (green field) data center constructions, branch offices or large workgroups. It's usually easier for small and mid-sized organizations to make the case for a "rip and replace" transition to UCS.
Cost is an additional concern in long-term UCS growth and scalability.
"I did a pretty deep analysis of acquisition costs with a Cisco UCS solution versus other approaches," Heil said, noting that other products showed lower costs up to 16 servers. "I looked at that 17th server and I would have to spend about 90 to 100 thousand dollars in additional I/O infrastructure," he said. "Because of the way that Cisco UCS scales, I would not have that huge I/O infrastructure hit."
The features and capabilities of UCS solutions can also render potential cost savings. For example, an ability to boot a UCS platform from storage area network (SAN) and migrate between blade and other servers can allow UCS replication between sites -- which may not have been practical (or even possible) before.
The next part of this tip focuses on the potential for vendor lock-in, interoperability issues, and deployment.
This was first published in March 2010