Administrator > Performance

A Study of Competitors’ Systems

Ben, a fictitious IT director, compares the IBM mainframe to other hardware products

In the first part of the series, we met Ben, a 41-year-old IT Director of Empire Health Services (a fictitious health-care company), who began his introduction to the world of the mainframe.

Ben is being tasked with a huge data center consolidation effort. After months of research, Ben decided to standardize around Linux and the IBM System z. Before he can close the deal, the company’s new CIO compels Ben to compare the IBM System z with a competing system from Hewlett Packard (HP). Ben’s CIO had a close relationship with HP—which manufactures the only competing hardware system.

IBM and HP have been doing fierce battle over the mainframe, with each side issuing claims and counterclaims. For example, HP touted TCO numbers without any information on specific environment and conditions. They also claimed to deliver Linux-based solutions at a lower cost, without any real data to back it up. Ben needed to look a little more deeply, since he was well on his way to making a purchase decision. Here’s what he found out.

The Competition

Ben recalled using the HP mid-range product line in the 1990s and really liked the HP9000s and HP-UX systems. So, Ben began researching HP’s enterprise systems—including how they evolved and how their history and strategies compared to IBM’s mainframe. IBM’s 40-year history of innovation is second to none. While HP was right up there with IBM in current revenue, with approximately $100 billion annually, their bread and butter is PCs and printers. Compaq’s merger with HP in 2002 enabled them to substantially increase their PC market share. HP’s largest systems are its HP Integrity Superdome servers, which were introduced in 2000, and run HP-UX. Some versions can also run Windows and Linux. Ben found it interesting that HP’s top-of-the-line servers didn’t even have a decade of history behind them. While HP9000s were definitely a force to be reckoned with in the 90s, at that point IBM’s mainframe already had decade of maturity behind it.

Ben’s CIO provided him with some literature HP had distributed, which illustrated how HP outshined IBM in the server area. Ben was interested in doing the research to see if he was right, and that the mainframe really was the best solution for his company.

TCO

Ben was surprised to hear HP claim they used less energy than IBM, and their solutions had a greater TCO than IBM’s solutions. HP claimed they used 45 percent less energy than a System z—though there were no facts explaining how.

Ben did some research and found by using the IBM System z, floor space can be reduced by 80 percent and power and cooling by 60 percent. In fact, he found the HP Superdome actually take up a larger footprint, use more energy and require a higher management expense than the System z. Ben plugged the following numbers into a table for his management (see Figure 1).

The results show Ben would save more than $60,000 per year by choosing the System z, which provides a crucial benefit in the current global economic crisis. At the same time, he could also help the environment, which is extremely important to Ben and his new CIO.

Another important benefit of selecting System z is the time to provision a server. With IBM’s System z, you get Dynamic CPU Management, dynamic memory upgrades and dynamic I/O reconfirmation. Furthermore, there is no performance impact due to provisioning, nor is any operator intervention necessary for upgrades.

With the HP Superdome, however, dynamic CPU management is limited to eight cores. There can also be a performance impact due to smaller cache on floating cells. Furthermore, operator intervention is required to un-assign and re-assess cells, as well as to recycle Workload Manager (WLM) and Single Shared Resource Domain (SRD) tools. The HP Superdome doesn’t have the ability to do dynamic memory upgrades, given that memory is pre-attached to cells, nor can it do Dynamic I/O reconfiguration, since I/O chassis are also attached to given cells.

Market Share and Economic Conditions

Ben discovered HP made many claims regarding market share. For Ben, market share is an important barometer of a technology’s relevance. Plus, the more people using a given technology, the greater the resources available to help support that technology and implement solutions.

Ben’s research, which was done through reputable sources from IDC and the Gartner group, found in 2008 System z shipments grew 44.8 percent. During this time, total Systems z revenue increased over 22 percent. In fact, during Q2 of 2008, IBM’s total revenue increased almost 12 percent. At the same time, HP’s Itanium Unix servers revenue increased approximately 9 percent, though HP’s total server shipments dropped almost 8 percent. Finally, according to IDC, IBM’s lead over HP is now 19 percent, an improvement of 3.4 percent from a year ago.

Ben also researched economic conditions, while reflecting on the global economic crisis. The mainframe line actually exhibited double-digit revenue growth in all geographies. This clearly showed Ben the IBM mainframe continues to perform strongly, even in today’s market, and further illustrates how business continue to build around the IBM System z.

Performance

Ben investigated claims that the Superdome outshined the System z in raw benchmark numbers. In fact, Ben discovered the raw numbers simply did not take into account how oftentimes it took more Superdomes to execute comparable workloads. Ben found the previously released System z9 broke all kinds of banking transactional records against the Superdome.

The System z10, built with quad-core technology, is designed to be 50 percent faster than its System z9 predecessor, and is the equivalent of approximately 1500 x86 servers with an 85 percent smaller footprint and energy costs.

Centralized vs. Distributed systems: With his analysis Ben learned a lot about how the mainframe offered clear and distinct advantages over distributed systems. His research showed centralized systems provided increased high availability, more secure backup and recovery, lower network infrastructure costs, lower software licensing costs, fewer maintenance costs, stronger availability and manageability of servers, better resource utilization and people savings through virtualization. Ben knew from experience he was lucky when a Unix server was 20 percent utilized, and it wasn’t unusual for a mainframe to approach 90 percent utilization. Other important System z benefits include:

• Better Price Performance—This is due to the wider use of server virtualization, which helps drive more servers into fewer platforms. IBM itself consolidated more than 4,000 distributed servers in six locations into just 30 mainframes running Linux—saving its client $250 million. This is part of its Project Green initiative, and is a direct result of IBM’s virtualization capabilities.
• Energy efficiencies—Numerous studies have shown the mainframe is more energy efficient than rack-mounted servers.
• Better Disaster Recovery (DR) strategies—Clearly, the fewer physical servers, the easier it will be to manage a DR solution. Furthermore, IBM has played in this arena for decades, before the term Business Continuity was even invented.
• Stronger ability to scale up rather than out—This is the current trend. Scaling up enables you to save on footprint, increase energy efficiency, increase virtualization possibilities and ultimately increase performance.
• Workload Consolidation—This can translate into significant costs savings by allowing you to manage the workloads.

Direct comparison: In an impromptu meeting Ben is asked to compare the two vendors' solutions side-by-side. Ben puts a table together comparing the features of both the System z10 and the HP Integrity Superdome. Ben is astonished the Superdome doesn’t compete with the System z10 in any single criteria. This includes virtualization capabilities, partition granularity, raw horsepower and the number of logical partitions. Ben compared some of the features of both, in Figure 2.

Security

Ben does a similar analysis, comparing the security of the System z10 with the HP Integrity system. Perhaps the most important difference is with System z, security is built into all system layers: the hardware, operating system and the middleware. Rather than an afterthought, the IBM System z server was built with security in mind.

Here are some other important findings:

• HP only supports SSL encryption, while IBM encryption is built into the hardware, the operating system and the middleware. System z/OS ISF also offers a single interface for all cryptography features. Encryption acceleration is offered with every engine (CPACF). This is non-existent in the world of HP.
• From a virtualization standpoint, HP nPars lack EAL4+ certification, while IBM’s logical partitioning is certified at EAL5. z/VM was also certified at Common Criteria EAL4+. Furthermore, IBM offers HiperSockets for secure networking. HP offers nothing here.
• IBM has centralized identify and access management through RACF and HP requires the usage of third party tools.
• IBM protects systems from malware and viruses, whereas HP really offers nothing.

This is only the start of Ben’s report to his CIO. There are still plenty of claims he wants to refute. He must also use his soft skills, do the politicking and enlist the right resources in his efforts to offer a compelling reason to use IBM and go against his CIO. At the end of the day, Ben will need to clearly demonstrate why the mainframe is the past, present and future of large-scale, mission-critical computing.