A well-observed characteristic of the computer industry is that everything goes full circle: centralise/decentralise/centralise, insource/outsource/insource and there are plenty of other examples.
In March 2009, networking equipment manufacturer Cisco played its part in those cycles by announcing what, to the untrained eye, could be easily mistaken for a mainframe.
The Unified Computing System (UCS) is in fact a new computing platform for the data centre that for the first time puts Cisco in competition with the likes of HP, IBM and Dell.
The system is based on blade servers using Intel’s next generation Nehalem chip, storage equipment from EMC, management software from BMC and of course Cisco’s own networking kit.
Crucially, VMware’s virtualisation software will be used to wrap a potential 320 compute nodes, housed in up to 40 chassis into a single compute fabric. (Arguably, the breakthrough that justifies Cisco’s entry into the server hardware market is a virtual switch that it has developed in partnership with VMware).
In other words, UCS should in theory allow an organisation to run its entire data centre out of a single box. Plus ça change.
Jim Houghton, CTO of IT consultancy Adaptivity, who sat on the customer consultation panel for UCS, is impressed. “It is considerably more dense and operationally efficient than anything that is available out there today,” he says.
But elsewhere, commentators have questioned the wisdom of getting so much of the enterprise IT stack through a single vendor. While Citrix’s Crosby acknowledges that concern, he adds that “what they are doing is addressing the complexities of managing virtual environments. Maybe a single vendor approach could fly.”