This morning is the last day of the IBM Systems & Technology Group Analyst Summit, packed with a couple of panels and then (for me) winter air-travel roulette through O’Hare. I’m betting on snow. See the highlights from yesterday as well. Here are some highlights from the morning:

• Rod Adkins opened with a brief talk going over revenues and R&D spend at IBM.
• Since Oracle/Sun, Rod said, we think we’ve won well over 100 opportunities against their top 300 customers (that is, a 1/3 of top account). And if you look at the GB [SMB] space, we’ve had another 100 wins.
• Research spend break out for IBM groups is: STG at 47%, SWG at 40%, Research 13%.
• IBM’s Robert LeBlanc moderated the first panel – the usual prepaired mini-talks, but hey – with Jai Menon (IBM, CTO, VP, Technical Strategy), Bijan Davari (Next Generation Systems and Technology), David Lindquist (VP, CTO of Tivoli), Matt Leininger (Livermore labs).
• The word from Bijan was that there wasn’t much more room for performance leaps in chips, which is one of those problems that’s been kicking around a lot recent. Instead of focusing just on processors to speed things up, then, you’ll have to tune other parts of the system, building “accelerators.” This fits with the System x contains more than just a generic pizza from yesterday.
• Asked to summarize the types of innovations Tivoli is working on, David Lindquist said it’s around virtualization, standardization [of parts of the infrastructure stack], both of which lead to “extensive amounts of automation.” The goal of using these new things in IT management is “fundamentally lowering the cost of managing systems and dramatically increasing the pace at which you can deliver services.” I’m always a big fan of the second and, actually, the whole bucket of ideas is about the best hope for what “private cloud” will be, even if the moniker is a bit rankling.
• As Merv summarized it, Matt said that “simulation is now a key pillar of science, enabling revolutionary breakthroughs.”
• The take-away from this panel – as has been the general theme of most all talks here – was that integrated systems are the best way to go for performance, cost, and all that. As one of the panelist summed it up: “It’s hard to innovate within silos [the opposite of being integrated –Coté]. Really the next generation of [things] can only come in an integrated way.” Looking at new innovations in hardware, like SSDs, he continues, “to really take advantage of those things, it’s going to change the operating system, the database system – you’ve got change the whole stack.”
• The next panel was run by Mark Shearer (VP Marketing Communications, Sales Support) and packed with IBM Fellows and Distinguished Engineers: Cod Barrera (Chief Technical Strategist Storage), Gregg McKnight (VP, System x), Guru Rao (Systems Chief Engineer, System z), Brad McCredie (VP, Power), Satya Sharma (Systems Software). Here, the idea was to speak to innovations in the works – driven by new customers needs like better power management, mostly – across the different systems and storage.
• The panel opened up discussing the impact of Smart Planet think on their various systems. Essentially, there’s more data in lumpy, rather than regular, loads. For the z and other folks, there was much focus on using new technologies to speed up and enable more analytics and raw data processing, doing OLTP and the like optimizations with integrated systems.
• Then the discussion moved to how reducing power consumption has become a key driver for systems design. The general goal here, most emphasis by Power’s Brad “No Socks” McCredie is to “keep taking the cost out of computing… To reduce [customer’s] total cost of ownership.”
• On that topics, McCredie continued (not an exact quote): 10, 5 years ago performance was all that mattered, so everything (in a computer) was on, at full speed all the time. Now we let the customer control more of those performance aspects so they can go for trade-offs between power consumption and performance. This “lets the user setup the policy, we do those trade-offs across the stack.”
• After this panel, IBM Research’s Brenda Dietrich was a stand-in for the scheduled speaker who got caught in weather. It was a reprise of the talk she gave at Connect09 on analytics, which is one of the more accessible and well put together talks on what analytics actually is, requires, and what you can do with them. As she said, “advanced analytics used to be pointed mostly at long-term problems- we had no choice. Now we can apply them faster.”
• The last panel was a customer panel, packed with people speaking to how they were using IT as a core part of their business, as a way to not only meet the usual demands on IT, but provide new services to monetize. The panel was moderated by IBM’s Jim Corgel (GM, ISV and Developer Relations, SWG) and had: Scott Smith (eMeter), Mark Moore (S1 Corporation), David Fertig (The Systems House), Nishit Mehta (HyGen Pharmaceuticals), Michael Jacobs (FIS).
• Having worked in the online banking world long ago (back at FundsXpress), I was especially interested in what Mark Moore had to say. S1 is looking at helping banks deliver very customized, “tailored services” for retail banking customers. And, having the fine grained, custom pricing that goes along with them. And, the never ending need for better fraud detection.
• Apparently, FIS has recently decided to switch from System p to System z for what seemed like performance and data integrity reasons. As Michael said, there’s only so far they can go with parrell processing in banking where account clearing, for example, is a very linear task.
• On the topic of eMeters, Scott Smith said: mostly we have large scales of data going on here, and needs for regulation and control. “there are 130M electric meters in NA – 416M meters worldwide – these are about to be changed. Large market. Large data *. eMeter says petabytes of data for utilities are coming and unprecedented. *“

Disclosure: IBM paid T&E for this event and is a client. Dell is a client as well. See the RedMonk clients list for other relevant client.

More details on what Smart Planet means for the Systems group at IBM.

Following this mornings general sessions, the afternoon brought more detail to the hardware meets Smart Planet vision-scape. There was more discussion of “growth markets,” a fun customer sit-down with Visa’s CTO Matt Quinlan, breakouts for the various server types, System Director talk, and then Q & A.

Here are some highlights:

• IBM says China continues to be a big, new market for them. They highlighted the big project types as: smarter rails, healthcare, banking modernization, Smart Grid. IBM’s Vinodh Swaminathan did the growth market presentation, and was interesting to note the lessening of the term “BRIC” and more emphasis on China, India, and Eastern Europe. ICEE, anyone?
• Tom Rosamilia, GM of System z (mainframes) hosted a “fireside chat” with Visa’s CTO Matt Quinlan. As customer talks go, this one was pretty good: it spoke both to technical details and the vision du jour, Smart Planet. As Matt summarized some Visa videos (making payments with a mobile phone and getting debit cards instead of checks for child support), these are “kind of our version in reality of Smart(er) Planet.”
• Visa has been running on mainframes for a long time, but updates to z/OS and other software throughout. To that end, when it comes to young talent, Matt said that they can get fresh college grades working on their mainframe based systems in two weeks.
• Matt on mobile & other new devices: the Visa plastic card has been around for 50 years. Even with a chip in the card, you can’t do much beyond the basics with it. With mobile phones you have a powerful chip, a computer in your pocket. “This is the first time in human history that we have truly a ubiquitous device…. What you can do with a transaction across a mobile platform is very different than what you can do with a point of sale.” Clearly, in being here Visa sees IBM as an important parter for that dry-cleaned cyberpunk vision.
• I chose to go to the System x breakout as most people I talk with are interested in x86, giving a confused look when it comes to Power and just laugh at me if I mention mainframes.
• The best, nuanced summary of the System x session was from Adalio Sanchez, GM of System z: “The hypervisor is not where the battle is going to be waged, it’s about how you bring it all together [into a system].” The idea here is that while x86 boxes may be built from commodity parts, there’s a certain type of embrace and extend IBM does to add extra value to how those “pizza boxes,” as Sanchez referred to them, are put together. That added value, along with a reasonably low price, is what would make IBM attractive vs., say, Dell.
• While it’s easy to make the Microsoft embrace and extend snark, he also clarified an anti-lock-in stance by dismissed unified computer entrants (Cisco) in the market by saying they’re “creating a proprietary approach that tends to lock people in. Integration is fundamental, but can be done in open ways that don’t create lock-in for the [customer].”
• The afternoon general session was started with STG Software GM Helen Armitage, who gave some more detail to STG’s software strategy. What’s interesting is that STG “owns” virtualization and operating systems (AIX, Linux, etc.) in IBM along with their own management stack, System Director. The goal of System Director is to provide base level, open integration and enough interfaces and UIs to use them without encroaching on Tivoli territory. While the virtualization and OSes are clear cut, things start to get fuzzy as you climb up to monitoring and other IT Management stacks. Clearly, Tivoli will do process and other ITSM tasks, but there’s enough overlap to be troubling, and many analysts said as much during the Q&A.
• There’s no easy, clear way to divide the world between System Director and Tivoli – STG can’t say it just instruments and makes monitoring possible in a “good enough” interface, and Tivoli can’t give up monitoring. Obviously, both do more than that simple distilling down, but in talking about the division of functionality, it’s difficult to get a “crisp” message. Then again, if one of them gave in on what they couldn’t do, it’d all add up. On the other hand, the metal to screen energy monitoring scenario from Pulse 2009 is a perfect example – and template – of how STG and Tivoli can share IT Management rather than throw dust up around the assumed clean separation of the two.
• As an overall note, it’s notable that the STG folks don’t speak to developers more. As Iwata commented on this morning, the new types of work loads (resulting in, requirements) that are driving new systems roll outs are not always ERP an classically enterprise software related. They’re new types of applications. Getting developers more intimate with hardware is a big challenge, as a whole generation of developers (myself included) were taught that working closing with the operating system, much less the hardware was a big no-no. Think the virtual machines of Java and the CLR of .Net. That kind of developer mentality favors commodity hardware, not the luxury toppings that come in IBM pizza boxes. However, there looks to be some developerWorks in the agenda for tomorrow, so we’ll see what Jim Corgel and crew suss out there.

In the evening we had dinner at the classicly country club feeling Winged Foot Golf Club, which was exactly what you’d expect from a golf club. Tomorrow has half a days activity and, tragically, a lab tour I’ll have to skip out on in favor of getting back to Austin.

Disclosure: IBM paid T&E for this event and is a client. Dell is a client as well. See the RedMonk clients list for other relevant client.

While at the IBM software group analyst event this year, along with several other analysts, I was invited to a lunch from the IBM systems group (or, “STG”) – hardware, from x to z, including storage. In particular, it was the new general manager, Helene Armitage, for software in the systems group talking about the growing importance of software for selling hardware. While IBM has Tivoli, the systems group maintains its own management software and, of course, operating systems and virtualization that goes along with it. (See Timothy Prickett Morgan‘s piece from today for more background on the group and execs.)

Aside from introducing a bunch of (presumably) software-centric analysts (like myself) to the systems side of software, the main themes were:

• Getting closer ties with the software group – they want to be “the foundation for what [Steve] Mills builds on,” as they GM said.
• Modernizing the talk around IBM’s platforms – see below.

There weren’t any launches or anything, but the GM did start to outline some interesting initiatives and positions, highlights below:

• From general purpose to specialized boxes – responding the Unified Compute ideas from Cisco and the like, but also the fact that massive cloud providers seem to demand customized boxes, a lot of the discussion got to providing more specialized boxes. Things like the CloudBurst box and iDataPlex which, you could lazily call an appliance, but are really big boxes and racks custom built for a specific way of using hardware.
• Jonathan Eunice of Illuminata asked about the dev/ops effect on hardware, pointing out that systems providers that cater to that kind of IT-think will be in a good position if it hits big. Those of you out there, dear readers, who put up with my incessant rambling on this topic know I’d violently agree. Indeed, the willingness that the GM represents from STG to cater to IT staff beyond box-only performance-hog mentality is a big opportunity. As Eunice also rightly pointed out, it’s certainly a chink in the enterprise infrastructure status quo‘s armor that VMWare/SpringSource is looking to build go through – it doesn’t seem like anyone else has noticed that ever widening hole. Key here for IBM is focusing on function, not the specific hardware. Outside of the traditional customer base and (to be it tongue in check) “those who know better” trying to sell anything but x86 here will be like the bad old days of American car companies competing with imports. That said, there was mention of selling 50 new z (mainframe) customers last year – I’m pretty sure we’re a long way from a mainframe renaissance, though.
• There’s a high degree of virtualization agnosticism – VMWare, Hyper-V, Xen, KVM, z/VM, and PowerVM: pick your poison. The x86 chip crew does a good enough job, it seems, tooling silicon up for hypervisors, and IBM has its own chips covered. Gear heads will go on about z and Power virtualization, which is fine for existing folks, but the same rule of dev/ops applies: either speak to x86 or speak only to function and benefit.
• To the point of all this function and benefit over spec, the GM summed it up nicely herself: on client calls “we spend more time now talking about the software instead of the hardware.” Part of that is the fact that a thick generation of IT buyers are conditioned to desire commodity hardware over specialized boxes.
• One analyst asked about networking. As the GM said, prefacing it with a “this is my view,” but, “IBM is not the network company.” Too bad that token ring thing didn’t work out.
• While STGs has it’s own stable of operating systems, Linux is still a big deal for them. Indeed, as the GM said, “I think we’ve had a fundamental impact on the Linux industry that we’re happy with.”
• The GM saw STG as being geared up for (private) cloud delivery, but the articulation of how exactly it fit in wasn’t detailed. Granted, Tivoli operates here, and the CloudBurst offerings are cloud-bound. Still, like most of IBM’s cloud programs to date, there’s a distinctly Blue cloud feel to it – not that we know what clouds any color except Amazon really look like, really. There are certainly “cloud in a box” offerings from IBM now, but I haven’t gotten a feel for how those are being received in IT departments. Part of that is that it’s too early to tell, while the other part is increasingly skepticism on my part that “private cloud” is, long term, a good thing or, best case, something that you could fairly compare to (public) cloud.

It’s worth tying in a some commentary from Steve Mills’ closing Q&A here. Judith Hurwitz asked him how he’d run software at HP. As part of an answer that he summed up at the beginning as “the way we’ve done it at IBM for the past 20 years,” Mills said the different IBM groups spend much effort on staying independent from each other. Though they don’t have to rely on each other for success, they can leverage each other for additive success, he said, making a somewhat subtle distinction. Clearly, Mills implication was that HP wasn’t being run this way.

Mapping that view in, it makes organization sense for systems to maintain its own software group rather than shove it all off onto Mills plate. And, as several analysts smirked over the STG steaks, governments might just go anti-trust crazy if the two were merged any closer – a smirk I’m not really qualified to judge.

Whatever the tie-up or the implications, STGs biggest challenge is the same as its been in recent years: convincing people to buy more than commodity (HP, Dell, etc.) boxes and storage. Typically, this means selling new functionality and benefits rather than the naked boxes themselves which, really, is mostly about software.

Disclosure: IBM paid T&E and is a client. See the RedMonk client list for other relevant clients.