Data centers are changing, and blade servers are bringing new efficiencies and flexibility to them.
The notion of a data center is transforming from discrete resources assigned to specific applications to that of a utility provider, like an electric company, in which a common pool of memory, processing, and storage is available at the flick of a switch. This vision of the data center as a utility service is driven in part by virtualization. Hypervisors separate applications from the underlying hardware, thus simplifying IT's ability to dynamically provision resources.
But this vision can't be realized by virtualization alone. Server hardware--specifically blade servers--also has a role to play. While traditional rack-mounted servers can and do run virtual machines, blade servers are better suited to the utility model.
With space and power at a premium in today's data centers, blades take up less room and use less energy than conventional rack-mounted servers. For example, 10 racked servers need a minimum of 20 AC-to-DC power supplies and 40 fans. But because power and cooling are distributed at chassis level in a blade system, a similar number of blades need six power supplies and 75% fewer fans.
Blades also increase the number of processors relative to rack space. For instance, Hewlett-Packard's c-Class BladeSystem can support the equivalent of 16 1U, dual-processor servers in only 9U of rack space. Other vendors' blade systems provide similar densities. These efficiencies provide more processing power while shrinking the overall space requirements.
Blade servers also reduce cabling in a data center because they use integrated switching modules inside the chassis, enabling internal port aggregation for Ethernet, Fibre Channel, InfiniBand, and other high-speed fabrics. Depending on input/output requirements, blade servers may reduce the number of cable pulls by 90% compared with conventional servers.
Blade systems' modular design is another advantage, making hot-swapping components much easier compared with rack-mounted servers. Because server modules are cable-free, swapping servers is about as simple as slotting a new blade into the chassis. This means new blades can be added or malfunctioning blades replaced with ease. By contrast, having to replace a failed conventional server in a tightly racked configuration may disrupt the operation of other servers around it. With conventional servers, maintenance is often delayed until scheduled outages. Such delays aren't necessary in a blade environment.
Management also is simplified. A majority of blade systems include a chassis-level control module that lets IT remotely manage the devices in the chassis. Management features include power cycling, system status reporting, and temperature monitoring. And many blade vendors support console redirection over IP, eliminating the need for KVM setups.
Of course, it's not all wine and roses with blade servers. The greater density of equipment in a rack will challenge the power, cooling, and weight-bearing capabilities of some data centers. Blade servers also lack the flexibility of commoditized rack-mounted servers, which can lead to vendor lock-in. Let's look at these issues more closely.
While increased density of blade systems can save space, it also increases power consumption and generates more heat at a rack level. For example, a 40U rack fully loaded with blade servers can require 18,000 to 24,000 watts and weigh nearly 2,000 pounds. Older data centers may need to upgrade power and cooling infrastructures and invest in added floor bracing to support the additional weight.
Buying into a blade system also means committing to one vendor's platform. While a single vendor has benefits such as a single point of contact for support and the opportunity to extract deeper discounts, it also ties IT to the hardware capabilities of that chassis design for the life of the product.
IT's Reputation: What the Data SaysInformationWeek's IT Perception Survey seeks to quantify how IT thinks it's doing versus how the business really views IT's performance in delivering services - and, more important, powering innovation. Our results suggest IT leaders should worry less about whether they're getting enough resources and more about the relationships they have with business unit peers.
What The Business Really Thinks Of IT: 3 Hard TruthsThey say perception is reality. If so, many in-house IT departments have reason to worry. InformationWeek's IT Perception Survey seeks to quantify how IT thinks it's doing versus how the business views IT's performance in delivering services - and, more important, powering innovation. The news isn't great.
InformationWeek Must Reads Oct. 21, 2014InformationWeek's new Must Reads is a compendium of our best recent coverage of digital strategy. Learn why you should learn to embrace DevOps, how to avoid roadblocks for digital projects, what the five steps to API management are, and more.