This is a big week for converged infrastructure: HP announced the Converged Cloud, IBM announced its PureSystems today and we hear another major vendor will make a converged infrastructure announcement this week as well. DevOps teams should pay attention to what’s going on here.
Last year Greg Shields proposed referring to “converged infrastructure” as “hardware that is created with virtualization in mind.” With that definition in mind, it’s not hard to see how converged infrastructure relates to DevOps’ goal of treating infrastructure as code – especially for teams working with on-premise infrastructure.
Private clouds aren’t just for conservative old enterprises any more. Many startups are opting out of the public cloud, whether that means building their own data centers or running their own hardware in colos like Geoloqi. Others like like Zynga and Tumblr are opting for a mixed architecture that uses the public cloud for some things, but still relying primarily on their own data centers. using the public cloud for some things but still maintain large on-premise systems (these are mixed architectures, not hybrid clouds).
In other words, hardware just isn’t going away, even for cutting edge Web scale companies.
Watch for the network layer to be increasingly important. As I’ve written before more network virtualization and automation are on the horizon. There are already a number of converged infrastructure systems that support virtual i/o, including those from Cisco, Dell and HP. This week VMware announced that it’s getting into the network virtualization game as well. Virtual i/o provides many advantages: Less hardware, easier configuration, faster changes and simple bandwidth throttling.
Also watch for more hardware automation all around. HP’s Gen8 line of servers integrate a higher level of automation. This is where converged infrastructure is heading: more automation at the hardware level means an expanded ability to treat even physical hardware like code.
Photo by dalbera