Unified Computing: A New Game Plan For IT?

Unified Computing systems like we covered in Next Generation Data Center, Delivered promise to be one of the "next waves" of IT computing, and it directs itself primarily toward mid-size companies that have distributed IT assets like Wintel and Linux servers and networks. This is the space that has been targeted by virtualization initiatives over the past five years, but while virtualization has taken aim at the reduction of physical servers, it has not provided end-to-end solutions that can automate resource provisioning, failover, load balancing and allocation over entire networks.

Initiatives like Cisco's unified computing system (UCS) are now promising to deliver integration of computing, networking, storage access and virtualization for Linux and Wintel-based data center resources. They are also challenging major IT players like IBM, HP and Microsoft to weigh in with UCS options. But for corporate IT, which is the ultimate target of all of this work, the integration required to facilitate UCS goes far beyond installing UCS software.

It begins with evaluations of dozens of IT processes built around manual provisioning, backups, resource allocations and load balancing for the network. Over time, these processes have cemented themselves into IT's daily workload and written policies and procedures. As much as technologies like UCS promise to simplify and even eliminate many of these labor-intensive processes, there is no way IT can automate end to end network infrastructures overnight. Instead, sites embracing solutions like UCS evaluate where they want to deploy it first. In many cases, they do this in the face of significant resistance from staff, which is already comfortable and competent in its day to day routines.

Despite the challenges, IT departments are penciling UCS into their technology roadmaps because it offers solutions that can help them implement strategies like cloud computing and resource deployment faster. They are doing this at the same time that recent surveys like the 2009-2013 IDC Forecast and Analysis for Worldwide Network Consulting and Integration Services reveal that only 20-25 percent of applications today are virtualized. The fact that only a fraction of applications are virtualized, with the remainder of applications continuing  to function on siloed hardware platforms, is a major argument for implementing  UCS, since UCS can automate and operationally integrate both virtualized and hardware-based resources.

"This is a market space that is forming, because hardware and IT infrastructure can be repurposed for cloud computing, scalability and failover—with or without virtualization," said Ken Ostreich, VP of Product Marketing for Egenera, a UCS solutions provider for x86 servers. "Sites can do this instead of using clustering and relying on hypervisors, because UCS virtualizes not only servers, but I/O, infrastructure and addressing."UCS doesn't cover high-end computer failover scenarios involving data mirroring, etc, but it does provide failover for mid-range, distributed computing environments. If a site is running an Oracle database on a physical server and SAP services on a hypervisor, the site can utilize UCS to perform a successful failover, even if an entire computing cluster goes down, because UCS can replicate the infrastructure somewhere else.   "All you need for UCS is a SAN (storage area network) with replication," said Ostreich. "UCS replicates your environmental profile. From an infrastructure perspective, your cost for switches, cables and I/O cards will decrease, because UCS can virtualize many of these devices."

An economical virtualization and automation strategy that is not dependent on actual virtualization should be a research item on every IT department's technology roadmap.  But along with this, IT needs to consider the changes to its operations, infrastructure, policies, procedures and asset management as well as the retraining of staff, and in some cases, the overcoming of staff resistance. For many sites, UCS will be a new game plan requiring a play book that has yet to be written.