Cisco announced that it is getting into the server hardware business with the introduction of its Unified Computing System (UCS), which unites network, compute, storage access and virtualization into a single platform. So far, Cisco has only announced the overall Unified Computing System architecture, and it has not yet revealed any pricing or details about the Unified Computing System blades themselves (see additional technical details about the UCS design in this IDEAS Technology Trends article – free registration required). However, Cisco’s fundamental value proposition for the Unified Computing System is that it promises a new degree of simplification, lowering costs by removing server-side infrastructure that can be expensive to deploy and manage. Specifically, the Unified Computing System eliminates the need to have separate switches in every blade server chassis; simplifies the blade chassis backplane; and reduces the cost of planned and unplanned downtime with stateless blades.
Cisco is entering an extraordinarily competitive market that has been targeted for years by the leading server hardware suppliers in the industry, all of whom currently partner with Cisco to some degree. In designing the Unified Computing System, Cisco clearly focused on innovating with I/O infrastructure virtualization, a key area of differentiation for the blade server platforms from IBM, HP, and Egenera (which also supplies its I/O virtualization technology to Dell and Fujitsu-Siemens). As Cisco fully releases the functional details of the Unified Computing System, we plan to scrutinize its technical tradeoffs with those other platforms, particularly in terms of its ability to improve availability, Disaster Recovery, and policy-driven automation, the areas in which I/O infrastructure virtualization can potentially deliver the greatest payoffs.
However, as customers start to develop strategies for applying virtualization across the datacenter, I/O infrastructure virtualization is emerging as a strategic technology. The Unified Computing System incorporates I/O virtualization into its design from the ground up, helping server, network, and storage administrators to work together in a new way to create resource management policies and instantiate servers. In describing its goals for the Unified Computing System, Cisco portrays a vision in which servers become "fluid objects in the network". In this vision, the network - rather than the server - clearly becomes the dominant force of control.
Cisco also emphasizes that it was able to use a clean-sheet approach in designing UCS in a way that maximizes the integration between servers and the network. Critics point out that UCS has a number of proprietary aspects that risk locking in customers, but at a time when complete datacenters can be delivered as an appliance, some users may be willing to overlook internal technical details if the overall value of the solution is sufficient. Right now, organizations are focused above all else on the twin objectives of cost-cutting and networked computing. Cisco is betting that by combining a fresh perspective and key tactical strengths to respond to these concerns, it will be able to exploit a unique opportunity to take on a new role as an IT supplier.
Comments