Virtual Management for the Masses

Arthur Cole

A new wave of virtualization management systems is due to hit the market in the latter half of the year, a testament that vendors want to spare the broad enterprise community the realization that early adopters have already encountered: Vast new numbers of virtual servers and partitions can produce a data management nightmare.


This week sees the preview of a new Systems Director Virtualization Manager for IBM's Power-based servers. The company is betting that a Web-based interface will be the preferred means of automating management tasks across multiple virtual and physical platforms.


Key to the system is its ability to manage logical partitions (LPARs) within the Power-based virtualization engine. LPARs can be created either with the Integrated Virtualization Manager of the System p or i servers, or with an external Linux-based Hardware Management Console. Because the two types of partitions communicate with the wider network in different ways, the new Systems Director offers a common interface to both that allows admins to oversee processor cores, main memory, I/O subsystems and even storage partitions, regardless of which operating system is being used.


A virtual network without a proper automation stack is like an orchestra of the finest musicians without a conductor, says Novell's Joe Wagner. Lack of centralized automation not only leads to virtual server sprawl, but can significantly hamper business-critical applications and systems in the event of a physical failure. It also makes it nearly impossible to perform routine maintenance without costly blackouts.


That's part of the reason many leading networking firms are partnering up with virtual management solutions providers, rather than taking the more time-consuming approach of developing them in-house. NEC, for one, recently forged an alliance with XenSource to deliver what the two companies are calling "Integrated VM" on NEC's Real IT Platform. The goal consists primarily of marrying the XenEnterprise system with NEC's SigmaSystemCenter and packaging the solution across Intel Xeon 5300-based hardware configurations.


Despite the development of new management and automation tools, overseeing an ever-widening menagerie of virtual server, storage, I/O and other elements will be no cakewalk. But the flexibility you get from virtualization at least ushers in the possibility of a management regime in which policies, not network limitations, determine how data is used.

Add Comment      Leave a comment on this blog post
Jun 29, 2007 7:12 AM Roy Agostino Roy Agostino  says:
Hello Arthur,Could agree more with the premise of this piece, I see this virtualization as another driver of complexity that is forcing a new approach to systems management. To manage through this complexity I think we need to shft our thinking on four concepts: - we have to think of an end to end service instead of managing just the individual elements- the deterministic "monitor everything and collect all data" must give way to a probabilistic "sampling" technique- human correlation of incidents must give way to automation- Mean Time To Resolution has to give way to the notion that we have to prevent systems outages before they cost us a lot of money The complexity you write about is here to stay, and it brings several benefits, but systems management paradigms must adjust to the new reality.Great piece... Reply

Post a comment





(Maximum characters: 1200). You have 1200 characters left.




Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.