Is there anyone who can say with certainty what the next generation of enterprise data centers will look like? A recent Network World article, for example, handicaps the field of data center vendors, picking HP, IBM, Cisco and VMware as the odds-on favorites (where is Dell?). Given the move toward clouds and fabrics (in which computing, networking and storage are interconnected tightly), this group makes perfect sense for the enterprise. To some degree, they all are partners with each other, as well as with specialized outsiders. But what about the impact of Internet data center design?
Google is the poster child for how to build a web-scale data center. And for Google, commodity is the word on the hardware front, with software as the secret sauce that makes everything work together. Anyone looking to follow in Google’s footsteps — not a bad idea given the near-legendary efficiency and flexibility of Google data centers — probably isn’t looking at specialized Cisco UCS equipment or choice-limiting VMware software. FedEx, for example, will explain next week how it built a cloud using Appistry software, which abstracts applications from hardware and lets customers achieve auto-scalability and hot-swap commodity hardware, a la Google.
I say probably, though, because even some cloud providers are on the premium hardware bandwagon. For example, managed hosting providers building cloud computing platforms are touting best-of-breed hardware and software. IBM’s cloud offerings, too, revolve around specific hardware for specific workloads. This trend almost certainly is designed to sell enterprise customers what they perceive they need, and if it continues will establish a rival school of thought to the white-box-centric strategies of Google, Amazon and their ilk.
Further complicating the issue are IDC’s latest server predictions. The analyst firm has volume-class servers outpacing mid-range and high-end servers in both revenue and growth over the next four years. This suggests Google’s strategy will take hold, but Cisco wouldn’t have jumped into the data center fray without a solid basis for believing it could make money. Right? Or did the economic meltdown throw the whole data center market out of whack?
When it comes to data center composition, my crystal ball is cloudy.