Michael Nielson, VP global marketing HP Networking
Cloud has been viewed first and foremost as a solution for computing delivery that begins with server virtualization. While this slant makes sense in conceptual terms, in practical reality, dynamic provisioning of all elements in the data centre – including servers, storage and networking infrastructure – are critical to the instant app access that cloud users have come to expect. This more holistic approach is a key feature of HP’s Converged Infrastructure platform, which in turn has served as the basis for the vendor’s Converged Cloud offering. Over the last year or so; however, HP has demonstrated a concerted effort to enhance the networking component in its cloud portfolio through targeted acquisitions aimed at developing network capabilities. The most recent of these is HP’s acquisition of
F5 Networks, a leading provider of application delivery controllers.
Based on this acquisition and HP IP, the company has developed the Virtual Application Network, a solution launched at this year’s Interop conference that is aimed at resolving a number of the networking challenges in cloud environments. The first concerns management. Discussing the new offering, Michael Nielson, VP global marketing HP Networking, noted “in the era of cloud, you can’t manage the network like you used to – you can’t rely on Command Line Interface to manage a public or private cloud infrastructure.” Nielson painted a dramatic model scenario to illustrate this point: in the case of a 500 server data centre deployment, with each physical server having 50 virtual machines, which each have 10 different network characteristics that need to be configured, “just to get it online, you are now talking about 250,000 command line interfaces that have to be entered just to roll out your data centre.” When human error in command line programming is taken into account, the project becomes especially vulnerable: assuming an error rate of one tenth of one percent, which Nielson claimed is pretty standard if not good, this means 250 errors are introduced – somewhere in the network configuration. The result, Nielson explained, is a “cascading effect” that slows deployment of the network, application servers and VMs, and delay in the roll out of services to business – with traditional network configuration techniques, business app deployment in the cloud is not tenable.
