Feature

From Network Computing Vol 18 No 06 November/December 2009

… GOING BEYOND TECHNOLOGY AND PRODUCT. THE NETWORK PERFORMANCE MANAGEMENT MASTERCLASS SERIES IN ASSOCIATION WITH NETSCOUT SYSTEMS, CONSIDERS THE MANAGEMENT COMPLEXITY OF VIRTUALISED SERVERS

IN DATA CENTRES

Opportunity: To unify data centre continue to need a unified view into inter-time and historically. Armed with this rich management of physical and virtualised application transactions and performance, data, these solutions can perform long- networks. as it relates to end-to-end service levels, established performance management

Challenge: Virtualised Servers are the new and user experience. This is entirely tasks including accelerated problem "black hole" in end-tto-eend Network and consistent with how traditional (physical) resolution, intelligent early-warning analysis, Application Performance Management.

IT organisations are increasingly deploying application and server virtualisation technologies to improve efficiency, reduce costs, and ease the environmental impact of computing resources. However, virtual servers are the new network "black hole," lacking essential visibility from traditional management solutions and in turn ignoring the behaviour and performance of the business critical applications that have been virtualised.

MANAGEMENT CHALLENGE PRESENTED BY VIRTUALISED SERVERS

You can easily picture that server farm in the typical data centre of five years ago; it probably looked something like a rack of dedicated application, web, middleware, and database servers, all interconnected with network switches. However, that is all changing now as virtualised server technologies are effectively and efficiently consolidating those racks of servers into many fewer virtualised ones, each hosting many different applications across virtual machines, with a virtual switch enabling communications between them.

While tools exist for administration of the virtual platform itself, the challenge facing most IT organisations is to consistently manage, optimise and troubleshoot virtualised applications, and simplify performance management across both the virtualised and physical data centre. The need or justification is not new, but the environment most certainly is. IT staff continue to need a unified view into interapplication transactions and performance, as it relates to end-to-end service levels, and user experience. This is entirely consistent with how traditional (physical) application deployments have been measured.

THE OPPORTUNITY FOR UNIFYING THE MANAGEMENT OF SERVICE DELIVERY FOR PHYSICAL & VIRTUALISED NETWORKS

Global organisations have been managing the application performance in data centres using a technologically mature combination of passive, packet-flow based monitoring, and sophisticated, in-depth analysis. As the applications shift from individual hardware servers to virtualised servers, performance management has become more challenging due to the lack of virtual machine visibility. In turn this is exacerbated by the ease with which virtualised servers can be moved with very little effort, albeit another advantage of virtualisation. This means that Virtualised server technology is the "new black hole".

That is, until recently. Solutions are now emerging, based on long-proven packet-flow analysis technology, that provide the very visibility into the virtualised server platform that is necessary to protect application service delivery. These solutions also leverage packet-flow data, but from within the virtualised server platform, using an embedded virtualised performance management data source or software agent.

Mirroring traffic on the local virtual switch, the virtual software agent can collect all of the essential metrics from the virtual machines; this can then be analysed, reporting on intra-server traffic, both in realtime and historically. Armed with this rich data, these solutions can perform longestablished performance management tasks including accelerated problem resolution, intelligent early-warning analysis, optimisation and planning, and protection of the user experience.

As you strife for end-to-end service delivery management in your data centres, including virtualised server environments, you may want to strongly consider the following.

A packet flow-based solution for in-depth decode analysis to troubleshoot end-user impacting degradations Both real-time and historical monitoring and trending metrics between virtual machines for optimisation and planning Early warning, automated behavioural analytics to isolate the movement of virtual machines Complete end-to-end views of intraserver activity, inter-server packet flows, and data centre-wide application traffic, in order to protect overall service delivery of all essential business applications. Finally, to ease the overall complexity surrounding managing application performance across virtualised servers through the data centre to a user in a remote office, a unified solution will help deliver a much more simplified approach. In the next issue of Network Computing, NetScout will continue this Masterclass series with its in depth look at Service Delivery Management.

Network Computing and NetScout invite reader comments and questions relating to items discussed in this series. Mail: Ray.Smyth@BTC.co.uk

Feature