• United States

Blade servers for a plug-and-play data center

Sep 14, 20043 mins
Data Center

* Basics of blade servers

The flexible data center of the future is emerging in many enterprises as data center managers try to turn the hype of “on-demand” or “utility” computing into reality. A flexible data center implementation is primarily an automated provisioning and management system over an infrastructure of commoditized and interchangeable computing, storage and networking components.

A key component of the flexible data center is the blade server. Blade systems consist of a rack-mounted chassis with several identical slots and the blade servers themselves, which are computers on thin circuit boards that can be plugged into the slots. The chassis provides networking, power and management and is hot-pluggable, which allows administrators to pull out a blade and replace it without powering down the system. Blade systems can pack a lot of computing power in a very small space – up to 84 servers in a rack. Data center managers can take advantage of these high-density servers to grow their data centers on demand.

Vendors such as IBM, Intel, Sun, HP and Dell all provide blade server systems – but the problem is that they are incompatible.

Unlike the PC market, which is highly commoditized, blade servers are proprietary architectures with proprietary management software. In 2002, IBM and Intel announced plans to co-develop blade servers, creating some standardization in this highly fractured market. Last week, IBM and Intel announced the availability of the blade server specifications, on a royalty-free basis. This is likely to spur the development of common standards and lead to the rapid commoditization of blade systems. As blade servers are more widely deployed and standardized, the next-generation data center will move closer to reality rapidly, and blades will provide the level of flexibility required to implement vendor-independent application provisioning.

The benefits of blade systems are:

* Better server density: Increasing utilization levels and cramming more computing power into less space will allow data center managers to continue to aggressively consolidate data centers.

* Improved reliability: Blade servers are the exact opposite of “special purpose” boxes. They are interchangeable, so a failed blade can be replaced very easily. Furthermore, automatic re-provisioning of applications can be completed in a matter of minutes using the management software provided by blade vendors.

* Simple deployment and management: Blades are easy to deploy, even by inexperienced technicians. Once plugged in, they can be managed remotely and grouped into “pools” that can be provisioned on demand.

Data center managers are likely to achieve immediate reductions in operational overhead by deploying blade systems in the data center. However, there are still some problems with blade servers. Both the hardware and the management and provisioning software are highly proprietary, meaning that IT executives will have to use a single vendor or expend enormous effort on integration. Furthermore, if blade servers are deployed as densely as possible (to maximize the cost savings), they will exasperate cooling and power problems in data centers which are not designed for that level of density, leading to increased failures from overheating.

Bottom line: Blade servers are the building block of a flexible data center and can bring immediate benefits in the form of reduced operational costs. IT executives should consider the integration challenges, and select vendors carefully – because they are going to have those vendor commitments for a while.