For this week’s Data Center Deconstructed we’re setting the Wayback machine to 1998, when Cisco opened a new engineering Data Center at its headquarters in San Jose, California.
As with any server environment approaching its 14th birthday, its physical infrastructure is lightweight by current Data Center standards. The oldest section supports just 1.5 kW per cabinet and 50 watts per sq. ft. of cooling – one fourth of the density in new Cisco facilities. Structural loading is limited, too. Two-thirds of the data hall is above the ground floor and supports about 900 lbs. across two floor tiles, a far cry from the 2,500 lbs. today’s storage frames can weigh.
Patching isn’t showcase quality, either. The room is lightly managed – internal groups are assigned row space and then install, cable and maintain their own hardware. Conditions are nothing like the entries in Cisco’s Crazy Cabling Contest mind you, but not all patch cords are neatly routed through wire management, either.
Why bother examining an underpowered, legacy Data Center with imperfect cabling? There are lessons to be learned here. As whiz-bang and worthy of attention as new server environments by Cisco and other industry leaders are, most Data Center managers don’t have state of the art facilities. They have legacy rooms, like this one. And despite its age this Data Center in San Jose remains productive and reliable, serving as Cisco’s largest hosting space for non-production applications, which is a credit to its design.
Come with me, then, on an archeological dig of sorts. We’ll explore the original section of the Data Center that opened in 1998 and see how Cisco’s Data Center design strategies evolved through subsequent expansions of the room.