Cloud Computing Fabric Capacity May Be Here (Debatable), But Is It Manageable?
February 1, 2012 Leave a comment
- While software-defined networking (SDN) offers a means to normalize and simplify/automate a portion of the switch infrastructure configuration, work remains on policy deployment and security.
- The deployment and integration of management software has been a severe pain point for enterprise infrastructure, with cloud services only compounding the issue.
Historically, enterprises either “built their own” management tools for script control and device configuration, used vendor-specific element managers and resigned themselves to running many different platforms (usually one per vendor), or attempted to integrate some of each of these into a framework from one of the major management platform vendors (HP OpenView, Computer Associates, BMC, IBM Tivoli, etc.). This issue has only been compounded with the needs driven by cloud applications and associated experience management (centrally managed QoS, security policies, and of course, bandwidth assurance) across a virtual infrastructure which has its own challenges. Part of the answer lies with SDN, particularly the OpenFlow initiative, which will provide a common management framework across a multi-vendor infrastructure, enabling consistent policy deployment (QoS, security, etc.) and configurations. However, the vendor-agnostic orchestration and automated deployment of an end-to-end cloud experience, whether application-driven or organization-dictated, is still in the process of jelling and remains the purview of startup vendors. There has been a significant amount of venture interest in this space in the last few years (VMTurbo, Puppet Labs, and Joyent, to name just a few of the many). These start-ups will also be competing to some degree with the larger framework players mentioned before, as they, too, seek to address this growing need.
If you think the lines separating carrier and enterprise are blurred, this was intentional; while each vertical has unique needs, they have many commonalities as well. Both need a resilient, easy-to-deploy, automated, and integrated platform that configures multi-vendor products (compute, storage, networking, and application). Though the need sounds lofty, rest assured that several start-ups, including some mentioned here, are aggressively pursuing this goal. 2012 is looking to become a year of software-enabled infrastructure, and those vendors which gain the most mindshare and momentum in this space will be poised to benefit handsomely, as the reality of the cloud hinges on the management of the complex elements within it.