1. Cheryl Watson’s Tuning Letter 2011 No. 4
2. Neat Mainframe Study
1. Cheryl Watson’s Tuning Letter 2011 No. 4
The 55-page 2011 No. 4 Tuning Letter was emailed to paid subscribers on October 13. You may visit our website at www.watsonwalker.com to obtain subscription information and the table of contents. The following is the Management Summaryfrom that issue, talking about some of the contents of this latest Tuning Letter:
The New System Programmer
We know that staffs are being squeezed today, and fewer people are doing the same work or more. One person on the staff, your system programmer(s), has more to do than there is time for. Additionally, many of the seasoned system programmers are retiring and are being replaced by younger people who don’t have the many years experience of the retiring staff. That’s why IBM introduced their z/OS Management Facility, normally referred to as z/OSMF.
z/OSMF is a web-based (and no-charge) tool that allows your system programmer to handle many of the day-to-day tasks on a Windows system, even remotely. Every feature of z/OSMF is designed to save the system programmers valuable time when performing typical tasks, such as modifying the Workload Manager (WLM) policies; diagnosing and tracking system dumps and forwarding them to IBM; configuring TCP/IP; cloning systems for distribution; monitoring the status of z/OS, zLinux, and AIX systems on one screen; monitoring capacity on demand; and even using ISPF from a Windows PC. z/OS 1.13 will also start implementing DASD management tasks in z/OSMF. Benchmarks by IBM show that a system programmer can save hours of time simply by exploiting z/OSMF.
If there is no charge, and it saves staff time, why isn’t z/OSMF installed everywhere? Well, that’s a long story, and one we describe later in this issue. But, in a nutshell, it is difficult to install, and it takes too many resources (CPU, memory, and DASD space). IBM is fully aware of the problems, and is working to reduce them. They also provide very extensive support if people have problems during the installation process.
A White Paper, WP101779, published in February made a lot of people nervous be-cause it said that z/OSMF needed a minimum of 2 GB of storage, and 120 MIPS. That’s not really true, but it scared off a few installations. The steady state (normal load) of z/OSMF takes less than 1% of a z10 uni-processor. The White Paper has just been updated to indicate that those resources are only needed during rare occasions, and they explain what those occasions are, so that you can schedule those tasks during off peak period. It also explains why the first installation of z/OSMF, which is normally tested on a very small sysprog LPAR, runs extremely slow.
But Cheryl believes in the need for z/OSMF so strongly that she thinks this is as im-portant to an installation as Workload Manager was in 1995. I’ve seldom seen her so enthusiastic about the future of a product. We believe that everyone will be using z/OSMF at some point, so we recommend that you take the time to install it and start getting the benefits immediately. The Focus of this Tuning Letter is to provide a description of the enhancements of z/OSMF 1.12 and 1.13, describe several user experiences, and provide recommendations on installing z/OSMF easily and in a minimum of time.
We featured the latest z114 processors from IBM in our Cheryl’s List #150 on July 12, 2011. We provide much more detail on the z114s and what they mean to you in our CPU Charts and on page 35 in this issue. If you didn’t see that Cheryl’s List, here’s what we included:
IBM announced their latest line of mainframes, the z114s, on July 12. The US announcement letter is 111-136. There are 130 new machines that corre-spond almost one-to-one in capacity and speed with the older z10-BC line of machines. With few exceptions, z10-BC sites can move directly to a similar z114 machine with similar MIPS and MSUs. IBM has even kept the model names the same. For example, both the z10-BC model 2098-B01 and the z114 2818-B01 are 29 MIPS and 4 MSUs. The largest models, the fifteen X01 to Z05 models offer between 5% and 18% boost in speed and capacity (with a corresponding increase in MSUs), but the other models do not. In fact over half of the other models come with a 1% to 7% reduction in speed compared to the z10-BCs. And sometimes, there are not corresponding reductions in MSUs.
The z114s, however, come with a lot of advantages over the z9-BC and z10-BC models, so they still provide a price advantage. The hardware maintenance is lower, the price of specialty engines is lower, and you can still have up to five specialty engines on each model. In additional, the z114 can attach to a zEnterprise BladeServer (zBX). The M05 comes with 5 processors, while the M10 comes with 10 processors, but only 5 processors can be configured as general purpose processors (CPs).
In positioning the z114 as an alternative to UNIX and Windows servers, IBM has lowered the cost of the smallest z114 to $75,000. And instead of reducing the MSUs to provide a technology enhancement, IBM has introduced yet another pricing option. This one is called AEWLC (Advanced Entry Workload License Charges), and is described in US announcement 211-250.
Elsewhere in This Issue
You’ll also find many other useful or interesting items throughout this newsletter: Full results from our Tuning Letter Survey • Missing I/Os from DB2 • How to analyze new system tasks • Update on zEnterprise High Performance FICON (zHPF) • CUNUNIxx parmlib member update • Important publications and papers from IBM • SMFand Information APARs to help you identify useful maintenance.
2. Neat Mainframe Study
While sitting in an IBM meeting in Poughkeepsie today, I heard a speaker mention a study done by Dr. Howard Rubin about the cost of doing business on z/OS. His papers can be found at www.rubinworldwide.com, with a direct link to a presentation at www.rubinworldwide.com/files/Mainframe_Cost_Effectiveness.pdf. A link about his study is found at www.ibm.com/systems/info/z/zenterprise/newaccts_roi.html. From that site, here is the introduction:
An independent study conducted by Dr. Howard Rubin of Rubin Worldwide shows that companies with a greater-than-average mainframe mix in their IT infrastructure deliver better business results than their distributed-centric peers.
These companies have realized significant savings in IT, for example:
44% lower IT cost per credit card transaction
31% lower IT cost per consumer loan
25% lower IT cost per mega watt hour produced
24% lower IT cost per hospital bed
20% lower IT cost per airline passenger
26% lower IT cost per new vehicle
25% lower IT cost per retail store
23% lower IT cost per barrel of oil
A very interesting interview with Dr. Rubin was provided in IBMSystems Magazine –www.ibmsystemsmag.com/mainframe/Business-Strategy/Consolidation/Breakthrough-Economics/.