IT’s battle against inefficiency has been raging for years. Two thirds of companies go over schedule on their project/solution deployments. Technology complexity doubles every two years. And more than 70% of budgets are spent on operations and maintenance, a percentage which is growing. Blade servers were introduced as a solution to such problems. And while they did make headway, their basic architecture failed to resolve key underlining issues.
In the first article of this 2-part series, we examined the findings of IT analyst Clabby Analytics’ 8-page white paper A Comparison of IBM’s New Flex System Environment to Traditional Blade Architectures to address the most frequently asked questions we hear from customers. In this second entry of a 2-part series, we’ll look at some of the shortcomings of traditional blades, and why Clabby Analytics believes IBM Flex System is the best blade offering in the market.
The Shortcomings of Traditional Blades
The Clabby Analytics report details the shortcomings of traditional blades. In particular, they cite four areas of concern:
- Chassis: Existing blade chassis were designed with yesterday’s speeds in mind. As a result, they limit bandwidth to 40 Gb or below, and in some cases do not support increasingly popular connectivity methodologies such as InfiniBand. In time, they will hit the wall and new chassis will have to be purchased.
- Latency: Traffic within the blades is routed in an indirect manner and this adds unnecessary latency. For those requiring high performance, internal blade latency is a growing concern.
- Storage: Blades originated as a way to increase compute density by packing more memory and processing power into a smaller space. This was accomplished at the expense of storage which was shipped externally to the SAN. That approach was adequate five years ago, but the rise of Big Data demands storage close to the CPU.
- Management: Blade management systems focus on the physical blade environment but are limited when it comes to monitoring virtual machines, networking and storage subsystems.
“To manage blade environment most efficiently, systems administrators and managers need well-integrated tools that can manage both physical and virtual resources across the entire systems environment (compute nodes, storage and networks) in a cohesive, integrated manner,” said the Clabby report. “All major blade vendors will need to rebuild their blade chassis environments in order to support high speed 100Gb Ethernet and faster connections in the future.”
How Flex System Addresses these issues
Latency is reduced by up to 50 percent in IBM Flex System due to a more direct design that facilitates rapid internal communication. Up to eight Solid State Drives (SSDs) can be placed inside the chassis which equates to 1.6 TB of superfast flash storage.
The chassis itself has advantages in terms of airflow, power consumption, and cooling and can accommodate advanced communications adapters such as the latest generations of InfiniBand, and 100Gb Ethernet. The Flex System chassis contains 14 half-wide bays whereas most competitors contain 8. It can also utilize 224 CPU cores per chassis whereas other competing chassis’ offer 128 CPU cores.
Finally, Flex System Manager (FSM) is a single pane of glass across the entire ecosystem. As well as all compute, networking and storage assets, FSM has the capability to manage all virtualization environments and related storage from one interface.
“All major system components as well as ancillary components will all be connected in a unified, converged system architecture that can be managed from a common management point,” said Clabby Analytics. “In the FSM world, systems managers have control over all of the resources needed to ensure that applications can run optimally.”
How IBM Flex System compares against HP BladeCenter and Cisco UCS
The report directly contrasts Flex System against the leading blade servers on the market, including IBM BladeCenter. Compared to the HP BladeSystem BL 460c Gen8, for example, Flex System offers 50% more memory for virtualization and twice the networking bandwidth.
While the Cisco Unified Computing System (UCS) simplified the deployment process of blades and associated virtual machines, Clabby found it to be too narrow in its application. The architecture makes it easy to install compute nodes, but is not well integrated with the rest of IT.
“The problem with this design is that the burden of integrating traditional blades with storage and switch components falls squarely on the shoulders of IT managers and administrators or onto systems integrators,” said Clabby Analytics. “These managers/integrators may or may not have deep tuning/optimization expertise. In contrast, IBM Flex System deliberately builds on what it has learned after years of infrastructure deployment to provide solutions that, in the majority of cases, outperform the self-integrated designs of most IT managers/integrators.”