I made a wrong turn and ended up in a power and cooling seminar at the Gartner Infrastructure and Operations Conference in Orlando last week. Having walked in a few minutes late ... I decided to stay and push my education beyond the mainframe rather than making a rapid u-turn after seeing my mistake. Well it seems there is some relatively new technology out there which sprays a non-conductive liquid on processors and system boards in rack mounts. A collector tray catches the liquid and circulates it through heat exchanger tied into chilled water to cool for re-cycling and the spraying / cooling continues.

I was immediately reminded of the new open systems technology a few years ago called "virtualization". before that it was mult-tasking ....

While I'd love to compare the costs of power for 40k Linux servers rack mounted against the costs of a mainframe with zLinux x40k .... I think I'll wait till 2011 when the cost of cooling is anticipated to be 2 times the cost of powering the dense silicon. Gartner states that the cost of cooling today is only 60% of powering .....

I wonder what new technology open systems will 'invent' next year ... maybe a bus and tag system for more robust peripheral communications.

I recently came across an article of interest entitled "Mainframes, SOA and SEX" by James Governer regarding SOA as a catalyst for mainframe growth and respect.

As an equal partner in the delivery of business services the mainframe can be leveraged as "just" another technology, and its strengths can be exploited. In order to exploit those strengths, SOA architects,  SOA programmers and SOA vendors unfamiliar with the mainframe need to get spec'd up. Exploitation is only possible with the knowledge of the capabilities and strengths of the platform. Think of the possibilities ...

When the internet bubble began to increase in size ... and the ASP model became the talk of the town there were thoughts that the ability to cloak the platform behind the internet might support a resurgence for more mature platform like the mainframe and the AS/400. There were some providers actually offering services over the internet which were delivered by these platforms. I believe psychological barriers proved to be insurmountable, and the need to have data (a companies crown jewels) within the confines of the company's walls thwarted acceptance of this model.

SOA does not suffer from the remoteness of data, but does cloak and exploit mainframe systems. It allows mainframes to do the heavy lifting, and GUI enabled platforms to make life easy for users. This is just what the mainframe needs ... a way to appeal to the non-mainframe techies ... and to continue to support business. SOA is the 'Screen' providing a gender neutral approach to computing platforms.

I’m surprised by a particular survey question in Gartner’s "Survey Results Provide Insight Into IBM Mainframe Market Momentum" (7 February 2006 ID Number: G00137484). In essence,(since this is copyrighted material I can’t be too specific) mainframe users are blaming ISVs for the high costs of mainframe computing. Interestingly enough I was at the Gartner Conference last year when the polling was done, and as a vendor, I did not vote that way. In fact I guess you might say that the poll reflects vendors and mainframe users.

The point is that mainframe software costs more than distributed software for a number of sound economic reasons and thus it is more expensive. But as most users know it has a much more rapid ROI, and provides more business value (savings) and thus delivers more value.

What is most disturbing is that mainframe software provided by vendors other than IBM has been relatively flat for years. I know for fact that is true with BMC mainframe software and market intelligence supports that the vendor constantly raising the price of software is …. IBM. Few users may have figured this out, and certainly procurement is still in the dark about the fact that with IBM IPLAs you don’t license a product you license a product number. While this may not seem like much … you need to upgrade the product every time the product number changes, regardless of ongoing maintenance. This upgrade has nothing to do with capacity changes; it is totally dependent on when a new version of the product is introduced by IBM. With each product number change carries an additional upgrade charge. Then there is the subsystem licensing, i.e. MLCs. When was the last time a new release of DB2 or IMS didn’t go up?

With BMC Software you license a product … reflected by a product name. Your maintenance insures the best possible support as well as your entitlement to new release and versions of the product as long as maintenance is maintained. I’ll let you do the math … but it is clear to see why they can afford to give away a free software license when in a competitive pinch.

The savings that IBM has delivered in regards to the hardware as been taken away with the price increases the levee on the software... and that is a reality!

I've been intrigued through the years at the manner in which IT looks at software costs. There seems to be a trend at reducing software costs no matter what the results / cost. In many cases software is considered to be an evil neccessity and rarely looked at as a facet of the total cost of operations. Years ago Gartner focused on a holistic approach to operations which it called Total Cost of Ownership. It originally looked at the dynamics between hardware, software and services (labor) to get a better picture of the fabric of IT. Currently their methodology includes Service Levels, which makes a lot of sense since focusing on the economics of IT without consideration for the reason IT exists is moot.

I recently had the opportunity to host Bill Kerwin at a webinar which outlined a holistic approach which reduces the overall costs of IT while delivering higher levels of service to the business ... and isn't that what IT is all about.

I believe the key to these reduced costs is automation, and with automation errors and problems introduced by change are reduced leading to better service levels and that can't be done without software.

The following HP webpage surfaced recently. This is so assertive, I would have to say HP is scared to death of the mainframe and envious of its dominance. Allow me to address their concerns ;)

1. Yes mainframe skills are on the decline .. as are IT skills in general. Many companies and organizations such as BMC, CA, SHARE and IBM are addressing the specific shortage in MF skills through educational programs at all levels. It is working.

2. Mainframe as always hosted more internally developed applications than ISV applications, and the adage that "if something works ... don't mess with it". The requirement for ISV apps seems to have always been less of a concern on the mainframe. Thus, nothing has changed.

3. Yes, hardware is more expensive ... but that is only one of the 5 components supporting a Total Cost of Ownership (TCO) methodology. However, you need less people to manage, the platform supports higher levels of service and provides greater business value then alternatives .... HP is mixing in Linux here ... so imagine the power and A/C required to support 40,000 rack mount Linux servers .... I'll bet a mainframe needs less juice and A/C ... certainly takes up less space.

4. Looks like a little teaming with Microsoft on this one ... titled "Linux on the Mainframe ..." but the text talks about the performance of Windows 2003. This is a real stretch ... mixing facts to get the desired results I guess. Enough said.

5. It seems they ran out of meat after bullet 1 ... I don't see any problem re-compiling applciations for a specific hardware platform ... especially given the reliability inherent with the mainframe. There is no reason for every hardware platform to mimic Intel. As the demand for Linux on mainframe apps increases the market will deploy mainframe specific compiles. Oh ... btw way don't you need to recompile for the variety of UNIX implementations including HP's UNIX?????

6. The analyst data supporting the low HP TCO appears to be a self-serving paper .. too short in length to understand the methodology and frankly my perception is that is would have been more meaningfult to compare the HP system to a comparable system. I got the impression they were comparing the TCO of palm applications on a hadheld and then the mainframe ... and the mainframe lost. DUH!

7. Yes, sales of mainframes are down but no one has ever said this was a growth market .... it is a stable high business value market.

It seems that everyone not actice in the mainframe market is standing at the edge casting stones .. interestingly enough you couldn't get all of them in a room long enough to combine forces ... could you imagine; Microsoft, Intel, Sun, HP, Dell and Oracle ... 

I wonder when Oracle will announce their new and improved mainframe migration effort now that they've assimilated Peoplesoft into their company ... it should be coming rather soon ...

In a the flury of IBM mainframe announcements I came acorss this little bit of fluff: "... BMC Software, Computer Associates, Hewlett-Packard -- tend to either ignore or isolate the mainframe in their rival products, Zollar said. 'It's a major shortcoming of our competitors,' he said. 'They create an artificial wall between zSeries and distributed systems.'" This got me a little excited. Now I know these words were most probably structured by a new marketing dude or dudess at Tivoli ... but it's a little early to expect the mainframers to forget the facts.

Hello Al .... up until the Candle acquisition just a mere year and a half ago IBM had a pretty empty portfolio in support of zSeries monitoring ... they had virtually nothing. In fact, prior to the acqusition only the Tivoli executives could spell mainframe and that was only because a number had come from  VTAM software backgrounds in Ralegh, NC.

I would not include BMC nor CA in the camp with "artifical walls". I can speak directly for BMC, and we have been supporting mainframe and distributed for years .. and now that our customers' DS and MF organizations are merging we are meeting their needs for wholistic support for all computing platforms within Business Service Management. Our current products in Mainframe and Distributed go way beyong just monitoring, and including fixing things before they break, automating trouble tickets and most importantly; linkage with a customer's business services.

Oh, by the way ... we recognize Oracle as a leading DB vendor and offer support for Oracle environments as well as UDB in the distributed space. So when it comes to your end-to-end transaction management story ... don't forget the facts.

Imagine a 24 year old excited about mainframes. Internet News captured a story which mainframers hope becomes more pervasive over time. It seems there is intelligence in undergrads and they are looking at the needs of the marketplace, and trying to differentiate themselves when selecting a career path.

IBM, BMC and CA are all making investments in mainframe education at the college level. Maybe the tide will start to turn, and the fear of insufficient backfilling of mainframe talent by the world's largest companies will become a distant memory.

What other ways can the mainframe community provide assistance in recruiting mainframe talent? What about technical schools. It seems to me that a programmer can do without art apprecitation, advanced physics et. al. 

I remember years ago, IBM retrained a lot of administrative folks as programmers when they decided to get into the software business. Maybe it is time to look at the old logic / aptitude test they used to determine the ability to be successful as a programmer. What do you think? Are we aiming to high? Should we be recruiting at a high school level and offering students a college and technical school path to the mainframe?

I attended a customer briefing last week, and was once again impressed by the due diligence being performed by IT. One company was about to revamp their business application suite. That is not unique unto itself, but they were investigating the tools available for the mainframe, specifically DB2 on z/OS. The application which they were about to standardize on was available on UNIX as well as the mainframe. The robustness and scalability of the mainframe platform was encouraging a serious look as it appeared to be a better match for their business.

This customer is not alone, but there is a tendency in this industry to look at, and publicize the new and shiny toys being deployed in IT. In all honesty, like many of its experts, the mainframe is not old ... it is just experienced.

If you think this is unique, then you might want to take a look at an article which appeared a year ago describing Ford Motor Company's flirtation with distributed systems application development. Ford I've been told looks at application deployment based on the attributes of the business service requiring the application. If the service can tolerate downtime then mainframe is probably not the best platform. If scalability, security and availability are important then the mainframe still cannot be beat.

Has a standard paperclip lost its ability to meet the requirements of temporarily binding papers? Its over one hundred years old.. and like the mainframe, it is still meeting requirements.

Three years ago I was party to a series of blind focus groups trying to understand what corporations needed to meet the inevitable IT skills decline in mainframe computing.  We talked with IT executives from around the US. and presented a concept to initiate discussions.

An interesting item arose out of these focus groups which was more organizational in nature. It seemed that there was a planned merging of organizations to breakdown the old mainframe v. distributed cultures to more directly facilitate cross-training. This view was expressed by the majority of participants. Over the last few years we've seen this trend continue with Europe adopting quicker than in the States.

The concept we introduced was a graphical user interface to existing mainframe database tools to provide a more comfortable environment for distributed DBAs. The feedback was positive, so we developed the concept and have been delivering a set of DB2 and IMS products with GUIs. In essence we are putting lipstick on the mainframe to facilitate cross-training and operational efficiencies.

Last year I made a customer visit to a leading retailer in the U.S.. I met with their senior IMS and DB2 managers. When I asked them if they had seen SmartDBA (GUI interface for BMC mainframe products) they folded their arms (in a form of protest) and exclaimed that GUIs were not for their team, they were experienced Pros. They relaxed when I informed them it had been built for their successors, not for their existing DBAs. For you experienced mainframe DBAs don't dispare ... this interface is in addition to the ISPF interface ... and is meant more the new recruits who will be toiling away as you enter your retirement.

A side note: After meeting with the DBA managers, I met their Director who disclosed that after two years of proposing, the business was finally going to allow him to reorganize the DBAs and merge the Distributed and Mainframe departments into one organization. It's nice when plans come together.