Hyperscale computing is slowly but surely taking over the enterprise. The Big Data explosion is forcing organizations to beef up their storage environments, but extending legacy infrastructure is not the way to successfully achieve hyperscale status.
Traditional RAID architectures are simply too expensive to maintain sustainability beyond the Exabyte threshold. That is why Facebook and Google, two pioneers of hyperscale in the private sector, have invested billions in resolving the primary cause of the Exa-scale cost barrier: complexity.
Wikibon’s Stu Miniman says that simplicity calls for infrastructure retooling at the rack level. The analyst points out that, like Google, Facebook’s data center is based on “homogenous building blocks” rather than bulky silos, purpose-built for apps that can potentially become outdated overnight. These building blocks are racks that contain 20 to 40 “vanity free” servers whose sole purpose is to run efficiently.
The modular nature of what Miniman refers to as Facebook’s ‘disaggregated rack’ architecture has several advantages, including the ability to rapidly scale up and down depending on the resource requirements of individual services. It’s also a lot cheaper because the infrastructure requires fewer personnel to maintain, and the lack of redundant functionality means the servers consume considerably less power than feature-rich boxes.
Can your business hyperscale like Facebook & Google?
Enterprises have several options when it comes to simplifying IT operations and reaching this level of efficiency. Amazon Web Services is among the easiest of choices. But Miniman advises against AWS as a sole solution – he deems the OpenStack-based Rackspace Private Cloud a viable alternative.