Infrastructures change, so automate the process

Change, infrastructure terms, used to occur rarely, and usually constituted a major upheaval that was at best partially planned and full of pitfalls. Now it is an everyday, even hourly occurrence, which makes the issue of managing the change one of the most important tasks facing IT managements.

By Martin Banks, 10 Dec 2007 at 12:50

`Managing', in this context, now means much more than simply organising what tasks should done, when. It now has to incorporate the automated implementation and deployment of the changes.

The result of this is that most customers are already managing complex composite applications over many tiers, so the ability assess the impact of change is increasingly difficult. The knock-on effects of any individual change in an infrastructure can be significant, and not always obvious. HP refers to such situations as `server collisions' and it is what its Business Technology Optimisation (BTO) suite of tools is setting out to overcome.

BTO brings together HP's existing expertise in infrastructure management tools such as OpenView and Business Service Manager with the technology acquired with Mercury Technologies last year and, more recently, Opsware.

The issues surrounding change management in an infrastructural context are now being made worse by the fact that server virtualisation is creating new servers, in effect, out of nothing. Some of them do not even `exist' for very long. HP itself offers the Utility Data Centre with a highly virtualised approach. This allows users to specify the creation of a virtual server or environment that exists for a specific period of time. IBM is now also offering this approach with its Blue Cloud system. Estimates predict that some 72 percent of all servers will in fact be virtual systems by 2013, with the average population being four virtual servers on every physical system.

In addition, the range of services that will be available via such infrastructures will have grown significantly. For example, it is expected that the position of VoIP in the enterprise will soon warrant words such as endemic and ubiquitous applied to it.

The cost of managing this complexity of infrastructure is already a major problem in its own right. For example, in 2005 the cumulative cost of managing the installed base of 27 million servers was estimated at $110 billion. The expectation is that next year will see an installed base of 35 million units. Even if the per-unit management costs remain static that will still mean a management bill of over $140 billion. The increasing complexity of the infrastructure, however, means that the potential for the per-unit management costs to rise is great indeed.

This means that issues such as Configuration Management are now moving centre stage for IT managements. This is the lynchpin of HP's BTO offering, which has been largely built around the Universal Configuration Management DataBase (CMDB) that came to the company as a core part of its acquisition of Mercury. Around this, both feeding and fed by it, are HP's Business Service Management and IT Service Management tool sets. The final component - for now at least - is the Business Service Automation tool set which came with HP's acquisition of Opsware in September. According to Ben Horowitz, co-founder of Opsware and now VP in charge of HP's BTO operation, the package represents `Automated Operations 1.0'. This has the objective of overcoming many of the silo'd and often labour-intensive processes involved in managing infrastructures that are now required to change rapidly and, sometimes, continuously.

For individual users, of course, all this represents the problems of the `back office' and the datacentre. Their immediate issues tend to be more focused around the growing number of options that are continually appearing in the personal client space. The volume of devices being used at the edge of corporate infrastructures is growing rapidly, and the functional richness of each of them is expanding by the day. It is now easy to predict a significant reduction in the number of `traditional' PCs and laptops, to be replaced by either functionally rich personal clients, or user ID tags of some description that allow a user's personal workspace to follow them to any available workstation or terminal. In either case, this will be all about service delivery by streaming content, rather than up- or down-loading specific applications and files, putting far greater reliance on the integiryt and reliability of the infrastructure.

1 2

Email to a friend

Print this page

< Previous Server : Analysis & Insight Next >