Posts Tagged ‘HP Converged Infrastructure’

IBM zEC12 vs. Itanium HP Superdome 2

November 16, 2012

Last week HP introduced its newest, top-of-the-line HP Integrity Superdome 2 server. This is the closest HP offers as a direct rival to the zEnterprise mainframe family.  The new machine is based on HP enhancements and the Intel Itanium processor 9500 series.  It can handle transactions 3x faster than previous generations while using 21% less energy. According to HP, users can realize a 33% savings in TCO over the previous generation of Superdome 2.

The new HP server is part of HP’s Converged Infrastructure, which is designed to provide hybrid computing by supporting its HP-UXHP NonStop and OpenVMS operating systems. Over time, HP notes, the Converged Infrastructure will encompass mission-critical x86 platforms to deliver a single, unified infrastructure for UNIX, Windows Server, and Linux environments.

This sound a lot like IBM’s zEnterprise hybrid computing strategy, which IBM introduced over two years ago through the zEnterprise-zBX combination.  Earlier this week IBM reported over 150 zBX cabinets and 1200 blades have been shipped, somewhat more than the 100+ zBX cabinets IBM had reported shipping a few months back when DancingDinosaur last checked. Clearly, z-based hybrid computing is gaining traction. Today, the IBM top-of-the-line hybrid computing server is the zEC12.

The Intel Itanium 9500 is the latest Itanium processor; the one Oracle prematurely announced dead and planned to stop supporting. It took HP over a year to win its lawsuit against Oracle, currently being appealed, to even get to this point.

The Itanium 9500 indeed is high performance.  It contains 3.1 billion transistors and supports up to 8 cores, twice as many as the previous-generation Itanium. According to published specs, it offers up to 54 MB of on-die memory, and enables up to 2 TB of low voltage DIMMs in four-socket configurations. The Itanium 9500 provides up to 2.4x performance scaling and 33% faster I/O speed over the previous generation, with frequencies ranging from 1.73 GHz at a power level of 130 watts, to 2.53 GHz at 170 watts.

In most key ways, the zEC12 beats the Itanium 9500, starting with its 5.5 GHz core processor and an increase in the number of cores per chip (from 4 to 6). Itanium touts 8 cores per chip but they are slower, half the speed or less.  The zEC12 indeed is faster and brings 101 user-configurable engines. IBM calculates the machine delivers 50% more total capacity in the same footprint.

The zEC12 also supports up to 3TB of RAIM (Redundant Array of Independent Memory), which protects against memory loss. In addition, the processor has been optimized for better software performance, particularly for Java, PL/1, compilers, and DB2.  Like its predecessor it handles out of order instruction processing and multi-level branch prediction for complex workloads. Large caches, almost 2x more on the chip, speed data to the processor. In addition, Flash Express provides 1.6 TB of usable capacity (packaged in pairs for redundancy, 3.2 TB total) to streamline database paging.

The Itanium 9500 chip and HP’s Superdome 2 server certainly won’t be a dog.  Let the chip geeks and benchmark zealots debate the finer technical points in the coming months.  But with the zEC12’s new availability and security enhancements, and a robust hybrid infrastructure it will be hard to beat the zEC12 for almost any mix of workloads, and that may be the key—the wide mix of workloads. While maintaining IBM’s core mainframe strengths in data serving and transaction processing, the zEC12 also brings a scalable and secure data repository for the enterprise, especially with the new Crypto Express4S card.  More than that, it can perform as a private enterprise cloud almost out of the box, and it is also a cost-effective solution for large-scale consolidation. With the zEC12, DB2 for z/OS, and the IBM DB2 Analytics Accelerator (IDAA) you can run both your OLTP and data warehouse workloads as one integrated workload in real time.

And then there is the new Transactional Execution Facility, which IBM brought over from supercomputing and is designed to help eliminate software locking overhead that can impact performance. It uses parallelism to drive higher transaction throughput. IBM’s Java Runtime Environment is expected to exploit the Transactional Execution Facility in an upcoming maintenance release. The new XL C/C++ compiler also is planned to leverage the Transactional Execution Facility. And there is much more.

In short, DancingDinosaur does not expect z shops to flock to Itanium.  However, a little reinvigorated competition is always good to drive innovation and restrain pricing.


Follow

Get every new post delivered to your Inbox.

Join 666 other followers

%d bloggers like this: