It's been said that server blades represent a strange and unusual challenge for IT departments because they somehow require a complete rethinking of management and support processes. Nothing could be further from the truth. Although certain aspects of server blades and the way their use will evolve will require IT staffs to make adjustments, only a zealot would assert that this means a revolution for systems management.
It's more true to say that blades simply make manageability more important than ever. Respondents to a recent InfoWorld server blade survey say they are focusing on uptime and reliability - 91 per cent and 93 per cent, respectively - priorities that directly reflect a system's manageability.
One sign of the increased importance, as well as the evolutionary nature of systems management processes and tools, comes from the hardware suppliers. While marketing staff fall over one another to convince us that blades change everything, vendors' software engineers are quietly tweaking the bundled server management tools to incorporate blade-specific features and instrumentation, particularly chassis controls that allow the immediate replacement of failed blades. Otherwise, issues such as OS and application management are as easily managed - or not at all, in a blade environment - as they are with comparable conventional hardware.
Survey respondents holding off on adopting blade technology indicate that the lack of well-known blade standards is largely to blame. In fact, 66 per cent of those who intend to buy blades soon cite half-baked standards as the reason they're in no hurry. Intel's IPMI (Intelligent Platform Management Interface), which is backed by Dell, Hewlett-Packard and NEC as well a number of smaller fry, is finally mature enough to build a product around. IPMI defines the interfaces to "intelligent" server components, including chassis, fan, power, and temperature sensors.
Hardware vendors appear to be taking two paths on the road to managing blades. Some niche vendors have gone along with larger vendors in offering highly manageable equipment, but that's not a universal plan. IBM appears to be bucking the trend by concentrating on deploying broadly scalable blade environments that may be cheaper to purchase by the blade, if not as thoroughly instrumented. Big Blue is upping the ante by reworking its mainframe management tools for the brave new world of blade computing, giving it an advantage of no small importance.
Realistically, however, the big vendors will have little to add to their management tools as a result of the rise of blades. Major server vendors such as Hewlett-Packard, IBM, and Dell have recognised for years the extra value - and the justification for the price tag - found in offering soup-to-nuts management features for their hardware and increasingly sophisticated deployment tools for the software that makes the hardware useful.
Server blades are arriving on the scene at an interesting time. Although the consolidation of file and print servers has been a mantra for the past few years, the growth of Web services and the increasing load on corporate databases from external and internal users have forced many shops to throw more boxes into their data-crunching and page-serving pools. According to survey respondents, server blades will fill much of the need to scale database applications and Web hosting broadly across machines: 71 per cent and 57 per cent, respectively, of polled readers report planning to use blades for these tasks.
But because there's so little difference between managing 100 rack-mounted servers and 100 blades, what's the fuss over managing blade-installed software? After all, any shop with more than 100 users probably has some sort of system for automatically installing software packages on desktops.
Although basic software deployment tools address program loading, content and traffic must still be managed to meet changes in demand. The need for better content management across a number of blades used for Web hosting will provide a much-needed tonic for vendors such as Fatwire. Load balancing becomes critical in a dense environment, and to no surprise, load-balancing kingpin F5 figures that putting its Big-IP software on a blade is an excellent idea.
Ultimately, customers are going to demand unified "lifetime" management services, regardless of the mix of operating systems in a chassis. That requirement may drive lesser-known systems management vendors such as BladeLogic to the forefront of blade management.
Some proponents of blade computing will point to reduced labour costs as another benefit of the technology, but this will be a relatively minor contribution to the bottom line. But don't discount the intangible benefits of improved morale that follows reduced grunt work, which is an implied benefit of switching from conventional rack-mounted servers to server blade technology.
Whether you consider blades flavour-of-the-month or the catalyst for changing everything we know about computing, it's clear that they will affect the discipline of server management for the better. Fortunately, the solutions to many of the management challenges, especially hardware monitoring and software deployment, posed by blades have already been around long enough for IT administrators to have become comfortable with them. Although the assumption that blades will make shops jump from dozens to hundreds of servers is far-fetched, wringing the most out of an investment in blades will require careful attention.
What's a blade server?
Blade servers are a new configuration of server that is conceptually described as a "server farm in a box". Rather than having multiple servers stacked horizontally on top of each other, multiple "blades" (electronic circuit boards) that have dedicated functions (such as Web serving) are stacked vertically in a single server chassis. All of these server blades share the same power resources and create less heat as a result.