IBM's new PureSystems deliver bigger clouds in smaller packages

IBM recently launched a number of new products in its Flex Systems line. Some of these products were based upon X86 processors and some were based upon IBM's own POWER7 architecture. These systems were teamed up with sophisticated management tools creating the image of a single computing resource combined with flexible memory, storage and networking capabilities. The obvious goal was providing balanced system offerings targeting the ever-moving virtualized, distributed workloads we're seeing in today's data centers.

What IBM has to say about its new Flex Systems

Today’s announcement comes at a critical time, as organizations look to consolidate their existing data centers and shift greater workloads to the cloud. In fact, one study has determined that by 2014, businesses will have migrated more than 62 percent of their existing database workloads to the cloud.

These additional Flex Systems products deliver the features that provide clients with the choices they need as they seek to reduce costs and invest in technologies that provide future scalability. The offerings consist of three systems built on both IBM POWER7+ and a “double dense” x222 system built on x86 technology. Additionally, IBM has upgraded the Flex System Manager so that clients can monitor and manage up to 5,000 infrastructure end points and do so from any location utilizing iOS, Android and Blackberry mobile devices.

The key elements of today’s announcement include:

• Flex System x222 – A new double density design built on x86 architecture that allows support for up to 2,800 Windows 7 user images in a single chassis;
• Flex System p460 – Delivers POWER7+ performance for clients with larger workloads;
• Flex System p270 – A POWER7+ option that is designed for medium-sized businesses and delivers performance for technical computing and data analytics;
• Flex System p260 – A new entry-level POWER7+ node ideal for IBM i workloads;
• Flex System Manager – Provides management of more systems from a single point of control and the ability to do so from any location with iOS, Android or Blackberry devices; provides a new utilization fuel gauge to help clients monitor the status and availability of their infrastructure manage; and has been expanded with the ability to manage the System x HANA-optimized x3950;
• New switches and fabrics that help enable Software Defined Networks (SDN) and improve connectivity and performance by increasing infrastructure bandwidth up to 40Gb.

Snapshot analysis

System design has always been a complex set of compromises to produce products with just the right combination of processing speed, memory size, memory performance, storage performance, storage I/O channels, and networking capability needed for the specific tasks at hand. The balance of these and other elements can make or break a product when it is paired with a specific workload.

A system that works very well for high performance, technical computing workloads that are tightly focused on processor performance, may not offer acceptable performance when that same system is expected to meet the demands of a I/O and network intensive transaction oriented workload. Add virtualized systems and off-site processing hosted at other data centers and the compromises that can result in good workload performance became even harder to make.

In the past, in order to address these broad, conflicting demands, suppliers broke up systems into individual distributed components and offered their products in the form of  a collection of systems, storage servers, network servers and various application specific appliances. While this approach offered the flexibility needed to address a widely varying set of workloads, it also created a very complex, hard to manage environment. As companies were forced to reduce their overall costs of doing business, the use of these highly complex, highly distributed, hard-to-manage systems began to lose their luster and look more and more like obstacles to success.

The pendulum started swing back towards a more unified system environment when suppliers such as Cisco, Dell, HP and IBM recognized their customers' distress and started stuffing all of those components back into a single enclosure. These suppliers developed very powerful management tools that could present a unified image to over worked, understaffed administration and operations teams.

It's a bit like watching the industry move away from easy-to-manage mainframes to highly flexible, highly customizable and highly complex distributed systems back to a newly re-imaged mainframe based upon industry standard components, virtualization technology and powerful management tools.

IBM, among the competitors, has decades of experience in unified systems because it continued to innovate on the mainframe theme and sell products and services based upon that unified view of the world. When appropriate, IBM took what it knows about mainframe computing and applied it to its other product lines. IBM's PureSystem and Flex Systems are clearly an example of the company's capabilities.

While the competitive battle to fill the data center with these newly re-imagined mainframes continues, it is clear that IBM has the expertise to offer winning products.