But Wait, There's More

• If you are trying to keep up with PTFs on OS/400 and related systems programs, check out the OS/400 PTF Guides, put together by our partner DLB Associates.
  
  

• September 13
• September 6
• August 30
• August 23
• August 16

• IBM is offering discounts on upgrades from certain AS/400 Model 720 machines to new iSeries Model 820 machines, to get customers moving to modern boxes, as we reported last week. The IBM discounts are between 30 and 38 percent off the list price of the 720-to-810 upgrade. OS/400 platform reseller Associated Computer Systems is throwing in a 15 percent discount on top of that, but you must order the upgrade before September 30.
  
  
• Two of the three big high-availability software vendors have upped the ante on IBM's new iSeries machines, the iSeries for Capacity BackUp and the iSeries for High Availability (see last week's issue for more on these machines). DataMirror is offering a 50 percent discount on its iCluster high-availability and clustering software for OS/400 servers when customers buy an iSeries for high-availability server before December 31. Lakeview Technology is offering a 30 percent discount to customers who acquire its MIMIX V4R4 high-availability and clustering software for the new iSeries high-availability box if they buy the IBM machine and the MIMIX license before December 31 as well. Lakeview is also giving customers who buy MIMIX and the iSeries high-availability server a 20 percent discount if they acquire MIMIX for Windows clustering software. And companies that opt for the iSeries for Capacity BackUp box can get a 20 percent discount on Lakeview's MIMIX dr1 data archiving and replication solution.
  
  
• IBM's Microelectronics Division, which has been bleeding money since the chip business fell on hard times, has needed some good news. The company's researchers announced yesterday that it has made two new breakthroughs in chip technology that might give Big Blue a lead in chip-making, which could attract customers to its foundries or enable it to make better chips for its own machines. The first announcement was that IBM researchers have created the first transistor using a "strained silicon" chip-making technique. IBM first started talking about strained silicon in June 2001, and the process that was used to create the transistor is a modified version of it, called strained silicon directly on insulator, or SSDOI. Two years ago, IBM said that strained silicon techniques could increase the speed at which electrons move through the etched wires of a microprocessor by as much as 70 percent and could yield increases in processor performance by about 35 percent without significantly altering CMOS chip technologies. What IBM does when it strains silicon wires is to put a layer of silicon germanium underneath the top layer of silicon, which stretches that silicon layer. IBM showed performance improvements of 20 to 30 percent by stretching the silicon. Using silicon germanium is tricky, but IBM has come up with an ultra-thin SSDOI methodology that apparently doesn't create the problems the company had in its labs two years ago, and it doesn't rely on the use of germanium. IBM has demonstrated SSDOI techniques on 90 nanometer and sub-60 nanometer technologies in its labs. In addition to the SSDOI improvements, IBM Microelectronics has improved the ability of positive charges to move through electronic devices. You would need a Ph.D. in electronic engineering to understand how. But what anyone can understand is that the new technique for moving positive charges inside semiconductors is based on 90 nanometer technologies and can improve the performance of semiconducting layers by 40 to 65 percent.
  
  
• IBM last week rolled out three new networking options for its BladeCenter blade servers, which it believes will increase the appeal of these machines among academic, institutional, government, and commercial customers who are interested in high-speed networking and cluster computing. IBM has announced two networking enhancements for the BladeCenter machines. The first is a Gigabit Ethernet switch module that implements what is called Layer 2-7 capability. In describing network resources, Layer 1 is the physical device and Layer 7 is way up the layers of abstraction to the applications running on the network. The IBM GbE Switch Module for the BladeCenter machines implements the functionality for all of those network layers, all on a single card, according to Nortel Networks, which is making it for IBM. In addition to the Nortel Gigabit Ethernet switch, Big Blue yesterday announced a special adapter card for message-passing interface (MPI) clustering, which is commonly used in massively parallel supercomputer clusters. As the name suggests, the Myrinet Cluster Expansion Card is being OEMed from Myrinet, the dominant vendor in this HPC clustering market. This card is based on and uses the same software as the PCI-X MPI networking cards sold by Myrinet and used by many server customers on stand-alone servers today. The other new networking enhancement for the BladeCenters is called an Optical Pass-Thru Module, which allows BladeCenters to link to Fibre Channel storage products such as IBM's FAStT and Shark arrays, tape arrays, and various SAN switches.
  
  
• Hewlett-Packard announced it has set up a consulting practice within its HP Services organization that is explicitly designed to pump up sales of hardware, software, and services in the emerging grid-computing and Web services areas. Grid computing and Web services evolved from two different parts of the computer business, but are seen by many as converging into a single blueprint and methodology describing how modern applications will be built from here on out. Just how big the grid part of this market, which is evolving from academic and government research institutions, is in terms of money is unclear. Part of the problem is that no one buys a grid; they buy components and set it up. Grid computing probably comprises a few billion dollars in revenues a year, and the market could grow to $4 billion for grid software and services by 2008. (This is a number that HP cited in its press release, and it does not include hardware.) But Nick van der Zweep, director of utility computing at HP's Enterprise Storage and Systems group, says that HP thinks this market will be much bigger, because companies will want to adopt grid computing once they figure it out; also driving grid sales will be the synergies with Web services. As part of the grid announcements last week, HP announced that it would be extending its OpenView systems management software so it could help manage grids, and reiterated that its Utility Data Center product (a mix of hardware and software to virtualize and provision servers in a data center) is already compatible with grid technologies. Rival IBM, by the way, set up a similar grid computing consulting practice a few months ago.
  
  

Sponsored By PROFOUND LOGIC SOFTWARE

RPG Smart Pages (RPGsp) is for serious e-development! Why settle for a tool that's just a code generator, or just a server, or just a development environment, or just a green-screen converter, or just a compiler? You need an environment that can do all of the above. And you need the best! www.RPGsp.com RPG Smart Pages (RPGsp) has everything you need. Free trial. Need more information? Contact Profound Logic Software at www.ProfoundLogic.com.