Tuesday, May 12, 2009

I think the mainframe’s too big!

Getting back into the daily routine in Simi Valley, with a cup of coffee on a warm but overcast afternoon, it’s hard to believe that only the day before I had been in Prague, the Czech Republic. The occasion had been the annual European BASE24 User Group (EBUG) meeting, and the picture is of me in the old town square.

Prague is a city I always enjoy visiting and it has been featured a number of times in this blog. In the November 9th ’07 posting, “Want to be my partner?” I wrote of how I had participated in a “Tandem Partner-to-Partner conference put on just for ISV partners,” and in the March 12th ’08 posting, “It's still the same, just different!” I wrote of how I had come “to Prague to catch up with HP BCS folks,” before adding “Prague is a very pretty city … (its) medieval architecture has been wonderfully preserved.”

EBUG has developed quite a tradition for holding its events among the more interesting cities of Europe – from Istanbul two years ago, to Vienna last year, and then this year Prague. While the attendance was a little down from last year’s Vienna event, the BASE24 user community support was still highly visible and it was clear that the support for BASE24 on NonStop across the community isn’t showing any signs of weakening.

The picture below was taken alongside of the GoldenGate, XYPRO, and HP exhibits and, as is typical for user events these days, it’s where conference delegates spend time networking over coffee. If you look carefully, you can just make out Steve Saltwick leading a user in the HP booth. Steve had just given a short update on HP NonStop’s presence in the Financial Services industry where he built on the material he had covered at last year’s event in Vienna.

In the posting of April 24th, ’08 “Vienna Dialogue!” written at EBUG Vienna and shortly after ACI announced the shift in strategic alliances from HP to IBM, I said that “there were many options open to HP – they could have simply elected to skip the event. In a very calm and measured presentation, Steve Saltwick, of HP BCS, started with the observation that ‘if it ain’t broke, don’t fix it!’ … (and he) went on to present HP’s vision of the future, how payments are viewed by HP as a change agent, and that HP was taking clear, low-risk, short term steps with an eye on comprehensive long-term options.”

This year, Steve opened with a slide of a roundtable executive meeting and where there was an elephant standing off to one side. Steve calmly observed “yes, it’s hard to miss the elephant in the room!” In this case it wasn’t a reference to ACI’s partnership with IBM but rather to the unprecedented times all users face in terms of regulations, bail-outs, outsourcing, M&A activity!

But it was hard to ignore the elephant and not think of IBM. “HP has 586 finance industry customers in 78 countries where 79%, or some 462, of them run NonStop. In the last two quarters alone, there had been 49 BASE24 ‘upgrades’ / BASE24 eps ‘new wins’ with 3 new BASE24 / Integrity NS Blades wins.” Low risk, short term steps? And an eye on comprehensive long term options? Clearly, HP hadn’t lost any market share in the year since Vienna’s event. Despite the shift in ACI’s strategy.

Before closing, Steve made the point of how HP had enjoyed a “100% win rate with competitive tenders coming from the installed base.” In other words, almost two years into the strategic partnership with IBM, and there had been no movement across the ACI user community from NonStop to the IBM mainframe System z solution. None! Nadda! Zilch! And there it was – IBM blinked!

Those present from IBM had sat rather nonplussed through the presentation by Jim Johnson of the Standish Group that highlighted the better TCO figures for NonStop on Blades, but now only an hour later, listening to HP highlighting its success of the past years, didn’t please IBM. But was IBM struggling? Uncompetitive? Hardly.

And yet, whether symbolically or not, the picture I have included here is of a System z10 Business Class (BC) mainframe that didn’t make it onto IBM’s booth. It was a tad too big, and remained outside the room for the duration of the event. Enclosed in glass panels, as it was, its internals completely exposed and illuminated by soft green lights, it gave me something to talk about during my own presentation.

Electing to bring a System z10 BC mainframe to the event reminded me of the exchanges I have had with IBM mainframe users over the past year. The Enterprise Class (EC) package, it could be argued, justified the complexity of the “Book” package – but since the Books in the EC couldn’t be reused in the BC package, why didn’t IBM use this opportunity to embrace Blades? Aren’t they thinking anymore? Hadn’t they just bought Platform Solutions Inc that had the firmware to run zOS on Blades? And then it happened – IBM chocked!

Just as HP retains the NSAA architecture, used with the NS16200 / NS14200 for instance, and positions it as a “top-of-the-line enterprise computing solution” so too IBM positions the System z10 EC as its top of the line. However, HP has added Blades packaging and has the ability to support all of its major operating systems with the one package, whereas IBM remained the contrarian. In not pursuing the Blades option, and sticking solely to Books, I have to believe it adds to the cost of the mainframe and, even as I listen to IBM’s counter arguments, it just doesn’t make a whole lot of sense to me. This shouldn’t come as a surprise to readers of this blog – I have been public with my opinions on this matter for some time.

In previous postings to this blog I have been highly critical of the proprietary nature of IBM’s chip packaging for the mainframe and of their reliance on Books rather than Blades. In the posting of January 21st ’09, “HP and IBM? Moving in opposite directions ...” I wrote of how, “unlike the IBM mainframe, HP’s largest servers (including NonStop) now support standard blade packaging and have embraced industry-standard interfaces and controllers … (on) the other hand, IBM is continuing with its proprietary ‘book’ packaging, completely shunning the blades that are available for the System p and System i user … in this instance, maintaining a proprietary approach (and) I think they are headed down the wrong path.”

Talking with IBM earlier in the day, they had said to me that moving the BASE24 community to the IBM mainframe wasn’t proving to be as easy or as straightforward as they had thought. Other delegates that I had talked to suggested that some within IBM’s management still considered migrations as nothing more than a 30 – 90 day professional service. And there were others within IBM who thought that NonStop users considering an upgrade to Blades were facing a migration away from NonStop – not aware that NonStop was supported by Blades!

Yet many delegates were surprised by my presentation. After listening to the presentation of Steve Saltwick, and earlier to Jim Johnson’s presentation, it was clear to me that the EBUG audience remained very committed to NonStop. When Barclays gave a presentation on their move to NonStop on Blades, and on how they had migrated their ATM network from the IBM mainframe to NonStop – I sensed the mood of the audience had become a lot more positive toward NonStop on Blades and I elected to give a lot more emphasis to IBM’s platform and packaging miscues.

And it occurred to me that perhaps part of IBM’s dilemma is that it too faces choices. Perhaps my presentation only confirmed their earlier observations about how difficult it had become to convince NonStop customers to migrate to System z. Should they have been surprised – the audience was made up of NonStop users, after all.

I have written extensively on how ACI’s preference for IBM, as its sole strategic partner, exposes BASE24 users to more choices – and how it will be these users who ultimately end up deciding the fate of the relationship between ACI and IBM. In the posting on March 16th, ’08 “ACI Strategy - it's all about choice!” I noted that “(while) ACI executives still reiterate the strategic nature of the partnership with IBM, it will be the customer that controls the final outcome and will be choosing the platform that meets their needs … (everything) I now know about the strategy suggests that ACI users will be given a choice – and this cannot be a surprise for anyone.”

But is it ACI customers that are the only ones having to choose? Pursuing a strategy based solely on migrating NonStop customers to System z now appears to be a rather poorly thought-through strategy for IBM and perhaps it is time for IBM to rethink their choices. There’s such a large population of home-grown solutions on the mainframe today – surely the better choice would be to pursue migrating them to BASE24-eps.

While IBM would push back firmly, and point out how this wouldn’t generate additional mainframe sales or improve its market share, I am not sure what other options are left for them. After all, and as symbolic as it was, IBM may just have to find other rooms to accommodate the mainframe – and ensuring they retain residence in their “parents” home, where the mainframe will make it through the front doors, has to be a key consideration.

And that does make sense to me. IBM blinked, and IBM choked, but after all those folks at IBM sure are able to think, and to revise their strategy! Aren’t they?


Kevin Callanan said...

Richard....as always....your insight is interesting and much appreciated.

Anonymous said...

As you quite rightly point out the z9EC and z10EC are both based around the concept of books of processors as was the z9BC. However, a fundemental flaw of the z9BC was the fact it had only 1 book of processors which technically constituted a single point of failure.

This was highlighted both internally and via external partners.

As a consequence - for the z10BC machine there was a departure from the traditional book (MCM or Multi Chip Module) - there was a movement to Single Chip Module (or SCM's). As you may have noticed in the demo display box there are six physical quad core processors in the z10BC - 10 of these cores are user configurable.

In terms of Blades - IBM are progressing with their Blade strategy for the POWER Range - which combines the previously separate ranges known as iSeries (AS/400) and pSeries (RS/6000). Conceptually there is very little difference between a Blade and a MCM (Book) - except that a book contains far more processing power than an average blade! Most Mainframe installs for medium to large customers come with a full complement of four books - with only the number of processors they require from day 1 enabled - this allows on-the-fly processing upgrades as-and-when required without requiring maintenance outages whilst new hardware is installed - it also means a customer is only paying for the processing power consumed - not paying for a large box that is only typically utilized between 10 and 30% (whilst waiting for that seasonal peak).

Richard Buckle said...

Your comments on the differences between the Books used in the z10 (and indeed, the earlier z9) are correct and that was my point.

IBM didn't reuse the Book out of the z10EC for the z10BC but rather, created a "special" for the reasons you noted. What disappointed me here is that they didn't elect to look at mainframe Blade options and simply add to the System p, and System i, product lines.

The timing looked good and the positioning of the z10BC as a product not needing the data center environment (and most likely running lots of Linux) could have opened more opportunities for them. Again this is strictly disappointin from my side - and an opportunity missed.

Dan Lewis said...

Interesting read, thanks Richard.

PS I think that could be me sitting at the XYPRO desk in your picture!

Richard Buckle said...

Yes Dan - it is!

Arnold Yan said...

I had some conversation with big bank managers. Someone of them told me, they will go to other vendor if ACI force them to switch to BASE24-ES. There are also a few struggling BASE24-ES projects going on. Customers are not stupid. They know what they want.

Richard Buckle said...

Customers are never stupid (although I have seen a few of them make dumb moves over the years), and whenever they face change they will always look at their options. In other markets and geograhpies, I have seen HP NonStop do a pretty good job at retaining their support and indeed, helping them migrate to newer NonStop servers. And this has been the real surprise for me.

But will other ISVs emerge - I have no doubt that a couple of them have already began looking at the opportunity and in the coming months, I expect to read a lot more about them ....