The Mainframe Blog Home

Server Hardware Trends: A Commodity Market Plus IBM

The New York Times summarized the latest IDC and Gartner server marketshare reports, highlighting the rise of the non-branded custom-built commodity server makers that supply big Internet firms such as Facebook. "Others" is now the #3 server "vendor" on a hardware revenue basis and #1 on a volume basis. (On a revenue basis, IBM is #1 and HP is #2.)

These long running trends are fascinating, and I've described them before in various ways. I think it's important, though, to distinguish between IBM and HP because they have very different positions in the overall market. IBM is now the only remaining credible vendor of "high-end" servers. We've seen time and time again in many markets — retailing, to pick an excellent example — that getting stuck in the middle is a bad place to be because competitors are both attacking from below and above. The attack from below is based fundamentally on price, particularly acquisition price. Those are the "Others." The attack from above is based on value, sustained high levels of research and development to deliver innovation, and best-of-breed capabilities and qualities. That's IBM. In the middle is HP, the JCPenney of the server market. In a few more quarters Dell will probably be right there, too, but we'll see.

I very much like IBM's position given these market trends, and I'm not too worried about the slight hardware revenue dip IDC and Gartner reported given the structure of that dip. IBM's high-end got higher, to put it succinctly, and there's some good evidence IBM's margins improved. Moreover, most of IBM's revenues associated with its servers are not measured by its hardware revenues alone, and that's unique to IBM. When HP sells servers they typically don't include much else from HP that customers buy. In contrast, it's very rare that an IBM server gets sold without substantial IBM content that customers buy.

I don't know exactly how big the high-end server market will be, but it will continue to be a terrific business amidst the continuing explosion of information, long-term economic trends, and increasing quality demands. As long as IBM keeps finding ways to differentiate and to innovate up and down their solution set, the company will do fine, and more importantly so will its many and growing numbers of customers. However, while IBM is very much pursuing its high-end strategy with gusto, IBM is also eager to push into volume markets as well — IBM in the role of Target (and/or Costco) to offer an alternative to Walmart, metaphorically speaking. I'm referring of course to IBM's OpenPOWER Consortium with Google, NVIDIA, and others.

So these Gartner and IDC reports are really not good news for HP in particular. As I've said before I don't know how HP gets out of its shrinking box. HP's CEO Meg Whitman has a tough job.

IBM Announces the OpenPOWER Consortium

IBM's Tom Rosamilia describes IBM's OpenPOWER Consortium announcement. IBM is sharing the complete blueprints for its POWER microprocessors with several major industry partners: Google, NVIDIA, Mellanox, and TYAN. Others are welcome to join. Yes, that Google, the search giant that buys many thousands of bespoke servers but which also has some of the most challenging data center-related problems in the world. Now Google gets an entire, more advanced microprocessor design to use as it pleases.

It's no secret that the traditional RISC UNIX market has struggled. IBM has been steadily gobbling up UNIX server marketshare for several years as other UNIX vendors, lately HP and Oracle/Sun, collapsed. But it's not good enough to dominate a (probably) declining market, so IBM is wisely trying to expand the whole market and go all-in on Linux cloud infrastructure. IBM has got some superb launch partners in that effort.

I think it's a bold IBM move but a calculated one. IBM is basically trying to replicate ARM's success in the processor licensing business but in a much different market, a market Intel currently dominates with its proprietary X86 architecture. I'm referring to massive, horizontal scale-out computing architectures in remote (typically) data centers: large Linux-based public clouds, notably Google's, but also with NVIDIA-infused GPU technologies for supercomputing (as another example). Not competing with ARM at all which, despite a few rumblings, isn't charging into data centers. Optimizing microprocessors for mobile use cases is quite different than optimizing for public cloud backends.

So will Intel get "squeezed" in the middle? The middle has proven to be a dangerous place to be in the server processor business. Which is why I also remain extremely bullish on zEnterprise, by the way (and which is doing very well indeed). It's certainly an interesting development, and it's really good news for customers. Frankly IBM had to do something bold, and this move definitely qualifies.

It also puts IBM's acquisition of SoftLayer into better focus. I was a little unclear how SoftLayer would fit into IBM's strategy, but now it makes a lot more sense. It also makes complete sense for IBM's launch partners to join the OpenPOWER Consortium.

I like this.

Happy New Mainframe Day! Introducing the zBC12 (Post #1)

We saw some clues that IBM was getting ready to announce a new mainframe today, and here it is: the new zEnterprise BC12. There are a lot of IBM announcements related to the new zBC12, and I'm sifting through all the information. Here are some of the highlights as I see them. I'll provide further updates as I read all the materials IBM has released.

• First, the hardware itself: processor clock speed increased to 4.2 GHz, core count increased, and both total capacity and per-core capacity up more than I would have expected. A single zBC12 can provide almost 5,000 PCIs for z/OS, z/VSE, and/or z/TPF with its maximum 6 CPs. That still leaves another 7 engines for any mix of specialty cores (zIIPs, IFLs, etc.) Uniprocessor performance is above 1,000 PCIs. Also, the maximum memory is now up to 496 GB of usable RAIM-protected memory, which is another very nice bump. This new zBC12 can soak up a large amount of workload.
• There's a new "LPAR absolute hardware capacity" setting on the zBC12 which presumably also will now be available on the zEC12. This setting will be mainly of interest to Linux on zEnterprise customers who want to set particular IFL capacity limits mostly for software licensing purposes.
• IBM is introducing exploitation of 2 GB memory page support in the zBC12 and zEC12 starting with Java 7 for z/OS, to improve Java performance and capacity yet again.
• There's a new high speed memory-to-memory adapter ("10GbE RoCE Express") which provides something analogous to HiperSocket connections but now between machines, to speed up data transmission and reduce networking overhead. This new adapter is available for both the zBC12 and zEC12.
• There's another new adapter for both models called the zEDC Express which accelerates data compression.
• I always wondered why IBM had 101 customer configurable cores on the zEnterprise EC12 machine. It's an odd number, and that's unusual for mainframes. Now we know: IBM reserved one core for a new Internal Firmware Processor (IFP) which is invisible. But this IFP, also included as a standard feature on the zBC12, supports the new RoCE Express and zEDC Express functions. I expect IBM will use this "hidden" processor for progressively more supporting functions, much like how SAPs provide various accounting and support services for I/O. We'll never really deal with the IFP and its control programs, but they'll be there, supporting particular new functions.

Much more information and analysis will follow. Stay tuned.

HP to Retire OpenVMS as Early as 2020

I expected HP to place OpenVMS into a virtualization package then continue supporting that package on X86 servers "forever," but HP isn't doing that. HP says it will not carry OpenVMS forward to the next (last?) Itanium-based server hardware — which wouldn't actually require much work. HP says they'll end support for OpenVMS as early as 2020.

Yes, if you're running OpenVMS you should be looking at z/OS as a place to move your remaining OpenVMS-based applications. I would recommend starting that project within the next 12 to 24 months given HP's forecast timeline.

Now what does HP do about NonStop (formerly Tandem)?

Sad Day: Unisys Abandons Its Mainframe Processors

I frequently write about IBM's virtuous mainframe cycle. IBM continues to pump lots of investment into its mainframe-related research and development because it makes business sense. Growing sales promote more investments, which promote more sales, and so on. The new zEnterprise EC12 and IBM's latest round of truly new and innovative software products for zEnterprise are among the direct results of those ongoing investments.

Unfortunately Unisys hasn't been able to sustain that sort of virtuous cycle. Unisys's announcement this week is, in my view, a sad closing chapter in that long history. The announcement is that Unisys has given up on its own processor designs even for its high-end mainframes. From now on everything will run in emulation.

Unisys was formed when Sperry and Burroughs merged many years ago. Sperry heritage mainframes trace their lineage to UNIVAC and Remington Rand, and they run the OS 2200 operating system. Burroughs heritage mainframes run MCP. There are still a few installations around, but Unisys's clear path (pun intended) is to manage their decline as best they can. Now with software emulation on server hardware sourced from Dell.

If you look across the industry right now you see that the most successful enterprise IT company (IBM) has intensified its focus on vertical integration: hardware and software co-designed and optimized together in common purpose. And the most successful consumer IT company (Apple) is doing exactly the same thing: the chips inside that new iPhone 5 are Apple's design, with Apple's unique power saving features. Oracle, with its Exadata products, and Microsoft, with its announced (but not yet shipping) Surface tablet products, are trying to do the same thing. It's a recipe that works and works well if you can sustain long-term investments to stay in the lead.

I completely understand why Unisys abandoned processor design. Good R&D is expensive, and Unisys can't afford it. I understand it, but that doesn't mean I have to like it.

With all due respect to Timothy Prickett Morgan (the author of the article linked above), emulation is not special, and HP should not buy Unisys (or vice versa). HP effectively outsourced its processor designs to Intel many years ago (Itanium), and that has been an unmitigated disaster for HP and HP's shareholders. HP needs market-relevant innovation, badly, after over a decade of corporate mismanagement and plundering. Again, look at IBM and Apple. Both companies have made relatively small acquisitions to bolster their own R&D efforts, and to nurture and grow those other great ideas as part of a common purpose.

Finally I should point out that Unisys's announcement is one more reason for their customers to consider switching over to IBM zEnterprise and z/OS. UBS made that move, as an example. That really is the clear path forward.

Cool Chips: "IBM's Mainframe Engine Makes Each Clock Count"

The Register's Timothy Prickett Morgan explores the fascinating details of the z12EC's new CPUs.

Happy New Mainframe Day! Introducing the zEnterprise EC12

Good morning in the Americas, and happy New Mainframe Day! Today IBM is announcing the zEnterprise EC12. You can order one (or more) today, and they start shipping in about three weeks. (IBM is just like Apple that way: no vaporware.) And the world's flagship enterprise server is practically perfect in every way.

I'm still reading through the wealth of information that IBM has revealed today. Every major general press outlet is reporting on this announcement. Here are my quick takeaways:

• Another giant leap higher in processor clock speed, and another world record. Wow, I am in awe. IBM people tell me they expect to maintain industry clock speed leadership "indefinitely."
• Hexacore architecture with about double the number of transistors per die compared to the z196. That's a wonderfully nice increase — but it amazes me that IBM's engineers figured out how to run and cool 6 cores per die at mainframe service levels and at such incredible, continuous speeds.
• About 25% performance improvement per core and 50% performance capacity improvement per server. Those are big jumps. Has anybody else noticed that's with only an amazingly short 25 month product cycle since the z196 debuted?
• Cache is up big, although the cache mix across levels is slightly different. Where you "spend" your cache real estate matters, so apparently IBM found a better arrangement for spending that much bigger cache budget.
• New "server class memory," i.e. directly addressable flash storage, another industry first-and-only. Including this server class memory you can have about 9.4 TB of directly addressable memory per machine -- all redundantly protected (RAIM), another industry first-and-only. Right out of the gate z/OS exploits this storage class memory for things like paging, memory dumps, etc. for faster problem determination and recovery. More exploitation to come, of course.
• New processor instructions and loads of new software that exploits those new, faster instructions, with a particular emphasis on information analytics.
• zEnterprise is IBM's first and most advanced hybrid computing server, and now IBM has upgraded the zBX to a Model 003 along with the zEC12.
• A new "zAware" option (which I need to read more about).
• Up to 101 customer configurable cores per server, up from 80. (N.B. There are scores or even hundreds of other processor cores inside such a zEnterprise server, even excluding the zBX option. Be careful if you count and compare.)
• IBM held the line on space/power/cooling requirements. Bravo. They've also added a new, even more redundant/more reliable cooling system.

I will probably post some updates a bit later but, in the meantime, let's read on together.

UPDATE #1: I've corrected the total memory figure above. Computerworld has an interesting story about the new zEC12, including these additional highlights:

• The storage class flash memory is encrypted, presumably because conceivably someone could physically get to it (unlike zEnterprise DRAM). IBM does think of these things. Moreover, IBM zEnterprise CTO Jeff Frey explains that IBM plans DB2 and Java exploitation of this new storage class memory. One example: in the not-too-distant future DB2 will be able to keep extremely large databases in directly addressable nonvolatile memory. Now that would be very interesting and useful.
• As usual, some workloads do better than others when IBM increases the performance of its mainframe, although all should do very well. Some of the ones that should do particularly well are big multithreaded Java applications, compute-intensive C/C++ code, DB2 (especially in analytics), and SAP.
• While IBM's z114 (and predecessors) have had the option to install without a raised data center floor, now that's available for the big zEC12 too. Eliminating the raised floor makes it easier to install a zEC12 in a crate, I would point out, so you can have interesting portable data centers.

The IBM announcement also mentions the fact that the zEC12 is the first commercial server to include processor instructions supporting transactional memory, not counting IBM's unique supercomputer for Lawrence Livermore Laboratory. Now, I have a bone to pick here: the computer science engineers (presumably) haven't given this feature a good name. The word "memory" is misleading, in my view. But so they've decreed, so that's the name. The new transactional (blank) instructions make it easier and faster to program with concurrency control. And that's a very good thing.

UPDATE #2: Mainframe Watch Belgium has much more information, including these highlights:

• IBM has introduced 2GB page sizes and supporting instructions in the zEC12, which are great for DB2 and Java, in particular. So now that's 4K, 1MB, and 2GB page sizes. Considering that we've had only 4K forever, and 1MB debuted only recently (in the System z10 in 2008), this is amazingly rapid evolution.
• IBM is going to eliminate zAAPs in favor of zIIPs. Initially with the zEC12 IBM will provide z/OS PTFs (enhancements) for running all zAAP-eligible workloads on zIIPs, even when zAAPs are installed. But the zEC12 is the last machine to support zAAPs. Translation: your capacity planning just got simpler. I expected this at some point, but I didn't it expect it yet. It's great news.
• There's a significant software technology dividend across IBM software. Or, said another way, you can do more with less money. Nothing wrong with that.
• The machine looks really spiffy, even artistic. No, it's not pink or purple, but perhaps IBM will entertain even that request if you want.

UPDATE #3: IBM's Nick Sardino takes us on a guided tour of the new zEnterprise EC12:

Musings on the Evolution of the Server Market

I've been doing a lot of thinking lately about trends in the server marketplace, in part based on conversations with server buyers but also as a student of history. To summarize, I think the world has decided there should be three basic types of servers:

• High-end servers (aliases: enterprise servers, mission-critical servers, vertically scalable servers, large SMP servers, "system of record" servers, "data center in a box"). These servers are becoming more important every year as businesses and governments cope with massive amounts of data, as security needs grow, and as the consequences and costs of business process failures rise. Preeminent among these servers are IBM's zEnterprise machines, although IBM's high-end Power servers are also firmly in this category.
• Workload-specific servers (aliases: appliances/appliance servers, application-specific servers, workload-optimized servers). These servers generally run one or a few highly related workloads that are packaged together and, at least to some degree, tweaked and tuned. They often include integrated disk and/or solid stage storage. There's a range of capabilities in this category, from the highly flexible IBM PureSystems to more closed offerings such as Oracle Exadata as examples.
• Commodity servers (aliases: X86 servers, volume servers, horizontally scalable servers). Obviously there are lots of these servers sold every year, and some buyers, such as big Internet companies, design their own hardware to their own specifications, bypassing vendors such as HP and Dell. Vendors such as Lenovo will also put pressure on HP and Dell in particular. IBM and Cisco have been trying to differentiate themselves with unique capabilities in this crowded market segment. Intel dominates this part of the server market, although it's still an open question whether ARM and/or AMD will be able to make further inroads as CMOS microprocessor technologies reach their natural limits.

There is some blurring between these categories, notably in the small but important supercomputer market which blends characteristics of workload-specific and commodity servers.

IBM has been carrying the high-end torch for years, building the world's most impressive, capable, and innovative enterprise servers. Consequently IBM always has a target on its back — it has always been thus. After all, if you don't have anything to compete, then the only option is to try to dismiss the category entirely. Also, while one of the defining characteristics of a high-end enterprise server is its general purpose nature — the ability to run multiple disparate workloads with varying service level requirements even within one operating system image — I would expect IBM to continue improving the "consumability" of its servers. An excellent recent example is the IBM DB2 Analytics Accelerator which combines zEnterprise, DB2 for z/OS, and Netezza technologies into a single, integrated, extremely high performance information system for both transactional and business intelligence/data warehousing. It "just works." In such ways these servers are workload-optimized while still retaining the full, open flexibility that general purpose high-end servers also have.

As I've mentioned before, the server business is tough. In some ways it resembles the global commercial aircraft industry which is highly competitive but also highly concentrated with only two major manufacturers: Boeing and Airbus. In the server market it's much the same now: IBM and Intel, perhaps with Oracle lately playing the role of Bombardier or Embraer (regional aircraft manufacturers for niche roles) — an even more volatile market segment. There's a simple reason: server (and processor) research and development is extremely expensive. Both IBM and Intel have built business models that, while different, work for them. IBM's server business model is analogous to Apple's, with lots of vertical integration, high value added, revenue stability and predictability, and a broad range of in-house capabilities. Intel's is more like Microsoft's, still its closest partner as it happens.

I am expecting the server market to continue along these three basic tracks for some time to come. I'm also expecting IBM and Intel to continue leading the industry. The vast majority of businesses and governments need some of what both these companies produce.

Oracle Loses Court Case v. HP, Will Appeal

Judge James Kleinberg has ordered Oracle to continue porting some of its software products to HP Itanium servers for as long as HP cares to build the machines. The Oracle software products include those that existed on HP Itanium servers as of September, 2010. The judge even ordered that the products must be complete, fully tested, tuned, supported, and released at the same time as other platforms — which is hard even when you want to deliver something. Tuning is especially different on Itanium.

In a previous post I wondered whether any court would order Oracle to continue doing something it clearly didn't want to do, doubting whether such a remedy would be viable. Well, apparently a judge thinks this particular remedy is operable. We'll see.

Naturally Oracle is appealing the decision.

Would you buy such a software and server combination?

IBM PureSystems: Simple Is Good

IBM officially unveils its new PureSystems today. With it, simplification takes a big step forward.

Enterprise applications and their interdependencies have become extremely complicated: hard to deploy, hard to manage, hard to scale, and impossible to secure. IBM is really working overtime to tame that complexity. Do take a close look.

Keep in mind that if you've got a zEnterprise server, with its Unified Resource Manager, you're already taming complexity like nothing else. For instance, IBM PureSystems initially support 4 operating environments across 2 processor architectures in harmony, which is a tremendous accomplishment. With zEnterprise you've got 8+ across 3+. (I'm using plus symbols because it depends on how you count, but 8 and 3 are the minimum counts.) In other words, IBM PureSystems are part of a continuum, and your zEnterprise server leads the way. It's extremely likely you'll want some of both in your data center.

So that's my instant reaction, with more comments to follow no doubt. What do you think? What are your most urgent issues?

Oracle's Hardware Sales Down Sharply Again

Three months ago, Oracle reported financial results for the company's second financial quarter. Its hardware sales declined 14% year over year, to $953 million. Oracle predicted that hardware sales would fall again anywhere from 4% to 14% in the next quarter (at constant currency).

Actually they fell 16% year over year, to $869 million. Quarter to quarter they fell about 9%.

For perspective, Oracle claims that their nascent Exadata/Exalogic/Exalytics business is fast growing, but that only means the far larger Oracle/Sun Solaris business is collapsing even faster. Also, the earnings report reveals that "hardware systems support" (a.k.a. hardware maintenance) declined only 3% (constant currency, year over year). So Oracle's remaining customers are caught in the perfect storm of a cratering Oracle/Sun Solaris business combined with escalating maintenance prices. Fabulous.

Of course I realize that hardware sales can be cyclical, but Oracle's hardware problems are deeply structural. Oracle introduced servers with the new SPARC T4 processors in September, 2011, which took the SPARC CPU up to 3.0 GHz. This past quarter should have been a terrific one, or at least a decent one, given that Oracle/Sun model cycle. Instead it was awful again.

I predict that Stuart Alsop now has the opportunity to correctly predict when the last Oracle/Sun Solaris server will be unplugged.

Strange Happenings in the PC Market

Welcome to 2012, faithful Mainframe Blog readers. The year 2011 is history, and now the results are starting to trickle in. Late last year IBM made some predictions about the next five years. I'd like to spend a little more time analyzing prediction #4 and its relevance to mainframe computing. There are two additional pieces of information available to me after writing that post. One is Gartner's report on 4th quarter PC sales, and the other is some careful observation of my family, particularly my sister, during the Christmas holiday.

First let's consider Gartner's report: PC unit sales fell 1.4% globally in the fourth quarter, and U.S. sales dropped 5.9%. HP's PC sales fell 16.2%. Note that these figures do not include Apple's iPad. If you also take out sales of Apple's Macintosh desktops and laptops, total U.S. PC sales (of Windows PCs) fell by 8.6%. (Apple's Mac sales grew 20.7%.)

Those are startling figures, but they are in perfect agreement with IBM's prediction. If the PC were the only way (or at least the "best" way) to access our increasingly digital world, we would expect the so-called "digital divide" to persist for a generation or more. Instead what's happening is that smartphones and tablets are rapidly becoming the most prominent access devices, while the importance of (and sales of) the PC are diminishing.

And I also observed my sister. She has an iPad and an iPhone. I'm not sure if she has a PC, and I don't think she cares whether she does. And for most of the Christmas holiday period and no doubt beyond she was glued to that iPad. She had everything she needed and more to support both the business and fun aspects of her life. And clearly she found the iPad nearly effortless to operate and worry-free. It's hard to break an iPad, in either software or hardware terms. She, and millions of other people like her, across all countries and social strata, are finding non-PC mobile devices much more suited to their lifestyles and needs. And this change is occurring very quickly.

Thinking as an architect, I was also struck by how much she was able to do in such a short time. The intensity of her iPad use was quite impressive and not, it seemed, a temporary phenomenon. As I mentioned previously, the infrastructure required to support the information delivery and transactional requirements of all these cloud-managed mobile devices is going to be astonishing. And it'll be mainframes of one stripe or another that'll do most of the heavy lifting.

I'm very bullish on the future of the mainframe as we begin this new year and enter the post-PC world. Be sure to keep stopping by in the coming weeks and months as we continue to explore the growing world of mainframe computing.

Happy New Year!

Mainframe Mobile Apps

IBM has introduced mobile apps which provide mainframe-related information. The apps are available for iPhone/iPad/iPod touch, Android, and Blackberry.

HP Lays an Ostrich Egg During Turkey Week

One of the fundamental rules of public relations is that if you want to make an announcement that you don't want to be noticed, do it just before a big holiday. That probably helps explain why HP picked the Tuesday before the big Thanksgiving holiday in the U.S. to announce its "Project Odyssey."

I have to applaud HP on picking an exceptionally appropriate name. Odysseus means "trouble" in Greek, and his journey took ten long years after defeat in war. During his journey none of his fellow travelers survived. Sounds like a perfect name for the latest plot twist in the Itanium Meltdown, doesn't it?

I can summarize HP's announcement thusly:

1. There's nothing we can do to save HP-UX, OpenVMS, or NonStop. We might have another unexciting Itanium chip or two in the pipeline, but who cares? Our turkeys are cooked without software vendor support.
2. Unfortunately our lawyers aren't giving us high odds of prevailing over Oracle in court. If by some miracle we did, it wouldn't help our customers. We'd just collect some cash.
3. You'll be able to stick obsolete-but-repurchased Itanium blade servers into the same chassis that also hold Intel X86 blades running Windows and/or Linux. This plan is so exciting that we're announcing it when nobody will notice. (And didn't we already announce that?)
4. Yes, we know blades are horizontally scalable, not vertically scalable. Yes, we know many NonStop customers aren't thrilled with blade architectures for availability reasons. Did we mention we're getting out of the high-end server business? Did we mention we couldn't afford the R&D a decade ago to be a credible high-end server vendor, never mind now? We just wanted to be a box pusher and let Intel worry about the CPU and everybody else worry about the software. It seemed like a brilliant idea at the time, at least to our accountants.
5. We'll strip the HP-UX corpse of any interesting bits and make those available for Linux and/or Windows. No, we don't have any idea what those bits might be.
6. There are no new versions of HP-UX, OpenVMS, or NonStop Kernel to announce. Are we being clear enough yet?
7. Isn't it cute that Microsoft said something nice in our press release? Of course they would: they abandoned Itanium before Oracle did. Of course they'll sign their name to our Itanium retirement party book. Ditto Red Hat.
8. In an announcement like this one normally we'd boast about all the new Itanium customers, our revenue growth, etc., etc. Of course, we can't do that when the opposite is true. We couldn't even persuade a real customer to say anything nice about this announcement. We got tired of hearing "F**k you!" every time we asked a customer for a quote.

zEnterprise and SAP: Perfect Together

I've seen some past criticism that IBM doesn't publish benchmarks for its mainframes. That's another "mainframe myth." IBM does publish benchmarks from time-to-time. Here's the latest example: a new world record for SAP core banking performance and scalability.

Benchmarks aren't useful at all unless they resemble some projected reality, in total. They're simply tools to try to understand how a system will behave when you put it to real, productive work. Unfortunately every business (and government agency) is unique, and when you're looking at a thoroughly mixed workload environment (IBM's zEnterprise), running a couple off-the-shelf tests isn't going to help much.

That said, this SAP benchmark is at least a better attempt, because it's trying to simulate what a real bank would do during the day, night, and across multiple channels and business functions. And the zEnterprise results are extremely impressive, even measuring SAP by itself. (IBM mainframes can and do of course run lots of applications on the same footprints.)

It's tough to benchmark "run most of my entire business" scenarios. But it's not necessarily impossible. I've worked with a lot of customers who pack up, go visit a benchmark center, and figure out how big their systems should be, how they should (or should not) architect their applications, etc. That's the best way to gain confidence about how well particular infrastructure(s) will support future business demands. (They can also test quality of service issues, such as how quickly they can fail over to a second data center, how much if any interruption there would be when upgrading software versions, and so on.)

Alternatively, if you've already got a mainframe, you're probably benchmarking all the time, perhaps without knowing it. You probably already know that your mainframe can handle the next merger or acquisition, and the next five applications, and then some. Maybe that'll require adding some capacity, maybe not, but the IBM mainframe inherently yields lots of performance-related data so you already know that you can deliver a quality outcome. After all, the easiest benchmark is the one you don't have to run. There are also affordable Capacity On Demand (COD) options for running special high-capacity tests on your own machine(s) — or to save your business if the programmers delivered some poor performing code that they can't fix right away.

Of course, if you have 150 million bank accounts, and you want to run SAP core banking, just go buy a couple zEnterprise machines from IBM (and a zBX or two). You've now got that benchmark, too.

Google Acquiring Motorola Mobility: Hardware Matters

Google announced a friendly takeover of Motorola Mobility less than a day ago as I write this. No fewer than seven law firms have already announced civil lawsuits against Google and Motorola Mobility in relation to the acquisition, which only reminds us that anyone can complain about anything, with or without legal merit. Instead of lawsuits I'll offer some quick analysis, similar to my previous HP Itanium meltdown analysis.

Mobile competition is turning out to be much more interesting than the PC-related battles of the 1980s and 1990s. Most of our planet's inhabitants don't have PCs or PC-like devices, most probably never will, and it's even possible that PC penetration will diminish over time. These trends are disturbing to some incumbent vendors, notably Microsoft. The one-two punch of the Internet followed by mobile computing have upended the technology world, with some potential winners and losers starting to emerge.

Apple is clearly a winner. Michael Dell famously recommended that Apple's management liquidate the company and give the proceeds to shareholders. Fortunately Apple rejected that free advice. Apple has recently surpassed Exxon Mobil as the world's most valuable publicly traded company, depending on which trading day you check. Apple is now over ten times more valuable than Dell. Most of that shareholder value derives from the tremendous success of the iPhone which has spawned the largest and most popular mobile application ecosystem that now rivals the PC. Later this year, Apple is cutting the cord for good: iPhones, iPads, and iPod touches with iOS 5 will no longer require a PC or Mac for keeping in sync and for updating their operating systems. Inevitably that means many people who don't yet have PCs won't buy them, and people who do have PCs will use them less and less. Mobile computing is just more...mobile. (One of the PC's few remaining distinguishing characteristics, the full-sized keyboard, doesn't offer much advantage for Asian languages such as Chinese and Japanese. And if you really want a full-size keyboard you can add one to your iPad or iPhone via Bluetooth.)

Apple is one of the best examples of the newly rediscovered maxim that "hardware matters." Like Exxon Mobil, which is a vertically integrated energy supplier, Apple is the ultimate vertically integrated mobile computing supplier. (IBM mainframes represent the ultimate expression in vertically integrated hardware and software in the business server marketplace. IBM i systems represent another excellent example.) Google has emerged as the other serious participant in mobile computing with its Android operating system and its related application ecosystem. In effect, in the shift to the mobile world Google replaced Microsoft as the dominant client software provider. I think Microsoft has already lost that fight, and the company's partnership with fast-fading Nokia won't work. Microsoft is to mobile computing what Digital Research was to microcomputers in the early 1980s — an odd irony.

In making this acquisition, Google has acknowledged that, as with mainframes, "hardware matters." Oracle has also come to that realization, but I think Oracle could have solved their problem at much lower cost than its acquisition of Sun. Oracle didn't need Sun to build Exadata-type products. I think Oracle was just petrified that IBM would acquire Sun and reacted accordingly. IBM wisely walked away, leaving Oracle with the carcass — and angry Sun customers. Google mostly solves an immediate problem: the attempted intellectual property-based attacks against Android. Google dramatically beefed up its patent portfolio thanks to a big patent purchase from IBM, and now Google picks up Motorola Mobility and its rich patent portfolio in mobile communications. That's smart when faced with the ongoing stupidity that is the patent system.

But Google has also come to realize that hardware and software integration matter, to deliver a smooth, trouble-free customer experience. Google is one of the few companies, even in the technology sector, that's just crazy enough to build its own servers to support its unique in-house software. Google's purpose-built software has co-evolved with its hardware over time. Likewise, Google should be able to take Android to new levels of ease-of-use and function thanks to Motorola Mobility, its most loyal Android partner.

It'll now be interesting to see what Samsung, LG, ZTE, HTC, Sony-Ericsson, and other mobile device manufacturers do. So far they have simply followed the fads, meaning that they've built a lot of Android devices, quite successfully. I don't view any of them as credible software companies. (No, Samsung's Baidu really doesn't count, except perhaps as insurance for Samsung.) Google has proven to be a good partner in many competitive situations. Despite competing in the Web browser arena (with Chrome), Google continues to fund the bulk of the Mozilla Foundation's budget in return for preferential search engine placement in Firefox. Likewise, despite some tension, Google and Apple maintain their partnership in search and in mapping. Google says that they will continue to improve Android and supply new Android versions on a business-as-usual basis. I believe Google, because Google's business model is and will remain based on maximizing eyeballs, i.e. based on advertising revenue. In the early U.S. television industry, NBC's parent, RCA, manufactured televisions, but NBC was also very happy to have viewers tune in using their DuMont and Philco sets. Google's business model is quite similar, and that'll continue.

So, fundamentally, Google's move is about protecting the Android ecosystem from attack and "getting their feet wet" in developing practical Android innovations that they will also readily share with other device makers to maximize their advertising revenue. I'm not the first to say it, but if you want to develop better software you should also build hardware. I think Google made a very smart move here, and I think Motorola Mobility's talents will accrue to the benefit of Android and of all Android partners. Those Android improvements will help widen the gap between Android and, in particular, Windows Phone and Blackberry, both of which seem to be fading fast. Rival handset makers might get a bit nervous, but I think they'll stick with Android and sell a lot of new, much improved devices. They'll also still be free to innovate atop Android if they wish, and they'll have a more powerful partner in terms of fending off IP-related FUD. Google is paying about $12.5 billion for Motorola Mobility which seems like a much better value than Microsoft's $8.5 billion for Skype.

It'll be interesting to see whether Microsoft buys Nokia now. That would be the obvious competitive response, and Microsoft certainly has the cash. It looks like Microsoft could pick up Nokia for about $30 billion. The trouble is, I don't know what Microsoft can do to establish Windows Phone as a significant mobile platform even with Nokia.

This post was updated with additional detail on Motorola, Skype, and Nokia acquisition prices.

Introducing the zEnterprise 114 (Updated)

It's new mainframe day! IBM has announced the zEnterprise 114, a particularly clever mainframe package which replaces the System z10 BC. The z114 is the little brother to last year's z196 super-mainframe, but keep in mind the z114 is in every way a real mainframe. A single z114 can handle more processing than IBM's fully configured top-of-the-line mainframe from just a few years ago. In no particular order, here's a list of the z114's improvements that I particularly like:

• For the first time that I can remember, and certainly for the first time since 64-bit z/Architecture debuted, all of IBM's mainframe models feature flexible, modular processor configurations. That is, you can order a z114 with either one or two processor "drawers" installed. I love that flexibility, particularly because it means that IBM has found a very smart way to lower the costs of entry into the mainframe world, so more businesses and governments can buy their first mainframe and enjoy its benefits. You can order an M05 hardware model and then, when/if you need more than 5 configurable cores, you can upgrade to an M10. (And you can still continue upgrading to z196 machines when/if needed.)
• Processor power and capacity are, of course, improved. That wasn't at all a problem with the System z10 BC, but it's always nice to see more performance in the smaller mainframe package. Uniprocessor performance is up 18%, for example. Clock speed is up to 3.8 GHz, which is a rather high-end number. Maximum z/OS capacity per z114 machine is up well past 3,000 MIPS, not including specialty engines.
• Coupling facility capacity is up a lot, too. If you need a dedicated coupling facility machine, the z114 is more likely to meet or exceed requirements.
• The z114 supports more memory than the z10 BC, but most of that increase is allocated to memory protection in the form of RAIM (Redundant Array of Independent Memory). Yes, the "little" mainframe gets exactly the same mainframe-unique memory protection that its big brother got last year — and which no other server has.
• Likewise, the z114 M10 provides a mandatory minimum of two spare cores, just like its big brother. The M10 also offers processor drawer redundancy. In the incredibly unlikely event a drawer or a core fails, the system stays up and running. If that capability is valuable, you can order the M10. For the rest of us (which is most of us), the M05 is perfectly fine and still better than its predecessors. Adding or replacing a processor drawer requires a planned outage, at least for now. Considering that everyone did perfectly well without that feature for decades, until the System z9 EC in late 2005, I think IBM can be forgiven for not getting that bit of work done just yet. (I think it's a moot issue anyway if you have a physical Sysplex.)
• You can run the z114 on high-voltage DC power or on AC, with or without a raised floor, and with top or bottom cable exit. In other words, you can put a z114 pretty much anywhere, including inside a mobile data center.
• Yes, you can add the zEnterprise BladeCenter Extension (zBX) to your z114. You can run the IBM Smart Analytics Optimizer, DataPower blades, all the other blade features — all centrally managed by the z114.
• You can order z114 machines now, and you can take delivery in September. Model upgrades from System z9 BC and z10 BC machines are available at the same time.
• IBM introduced "Advanced Entry Workload License Charges" (AEWLC) to go with the new z114, aimed specifically at smaller mainframe customers running single, standalone machines (with or without internal Sysplex clusters). Otherwise, AWLC is available for physical z114 Sysplexes, so that's good, too. AEWLC, AWLC, and IWP represent the technology dividends: you can enjoy the same performance and throughput for a substantially lower software license charge, or you can enjoy substantially more performance and throughput for the same software license charge compared to the System z10 BC. All I hear about is that everybody else everywhere else is increasing software license charges and maintenance charges. For well over a decade IBM has been slashing software prices on its mainframes, so you get way more bang for the buck (or euro or yen). I really hope people understand this stuff by now.

All goodness. Nice job, IBM.

UPDATE: IBM has issued a press release highlighting the new z114 mainframe. Indeed, IBM says that the z114 has a U.S. "starting price of under $75,000." That's the lowest price ever for an IBM mainframe and a big 25% reduction from the previous entry price. Also, according to Timothy Prickett Morgan at The Register, on an equal performance comparison IFLs (Linux processors) on the z114 are priced 37.5% lower. There are big price reductions across the board, though, for both hardware and software. It's clear IBM is determined to convince many new customers to buy their first mainframes.

More Itanium Meltdown News: HP Sues Oracle

In the wake of Oracle's decision to cease all development of new Oracle software releases for Itanium-based servers, Hewlett-Packard has filed a lawsuit against Oracle. HP accuses Oracle of anti-competitive behavior. Oracle issued a statement which ends with this sentence: "Intel’s plans to end-of-life Itanium will be revealed in court now that HP has filed this utterly malicious and meritless lawsuit against Oracle."

I'm not a lawyer, but it's hard for me to imagine HP prevailing in a lawsuit. Even if Oracle is anti-competitive, what's the remedy? Would a court order Oracle to continue writing software for a particular platform? Which products, for how long, and would anyone want to buy products Oracle didn't want to create but which a court forced them to ship? In all the cases involving anti-competitive behavior that I recall, any legal remedies involved stopping a particular activity, disclosing information, and/or a financial settlement. None of those remedies are going to solve customer problems in this situation. Also, Oracle will no doubt point out that they continue to develop software for IBM platforms (AIX and Linux on System z), for Linux running on HP servers, and for Microsoft Windows running on HP servers. Furthermore, Oracle would argue that they were only following Red Hat and Microsoft in abandoning HP's Itanium servers. In other words, the market has spoken, and no vendor should be required to support a fading UNIX platform longer than they wish — and, besides, Oracle has done more to sustain Itanium than anyone. Moreover, history is littered with fading UNIX platforms for which vendors no longer develop new software: Silicon Graphics, NCR, SCO, NeXT, Data General, Wang, etc. HP has discontinued development and/or support for many operating systems, too. Are any Tru64 or MPE/iX customers happy with HP?

There were a few HP customers that were lobbying Oracle to reverse or at least modify its March announcement. The odds of that happening were always low, but the lawsuit (and Oracle's response) have dropped the odds to zero. Oracle isn't coming back to Itanium.

I suspect HP knows they cannot prevail in court, but sometimes it still makes business sense to file. As I mentioned previously, most Oracle-HP customers, now that they are forced to choose sides, will choose Oracle. Software trumps hardware, quite simply. However, some percentage of customers find alternatives to Oracle. HP probably hopes that this lawsuit will encourage more of those customers to side with HP. We'll see.

For prior analysis of the Itanium Meltdown, including recommended solutions and strategies for Oracle-HP customers, please see here, here, and here.

Reports of Oracle's New Hardware Maintenance Terms

I have been reading in the trade press that Sun customers are not happy with Oracle's acquisition because they're seeing sharp increases in support costs. Now I'm getting more details from readers detailing the problems. For example, Sun's hardware maintenance was available but selectable per machine, as with nearly every hardware vendor's practice (including IBM's). That is, you could choose specific machines for 24 hour coverage, for business hours coverage, and for zero coverage. That makes sense: some machines are more critical than others, and you could pay for whatever maintenance coverage you need per machine serial number.

In contrast, Oracle thinks that hardware maintenance should be like software maintenance: all or nothing. If you want 24 hour coverage, you have to buy it for every Oracle (Sun) machine you own in your entire organization. And the maintenance price per machine isn't any lower. If anything, it's higher after Oracle's acquisition of Sun. Maintenance price increases might be more tolerable if Oracle had better support than Sun, but that's certainly not true.

Oracle's strategy seems to be to extract as much revenue as possible from Sun customers as they exit the platform. I suppose that's one way to run a business, but, fortunately, Sun customers have alternatives, including IBM Power and zEnterprise servers.

Hey, Gartner: You Blew It

Analyst Joe Clabby wrote an insightful article last year criticizing Gartner for its frequent advice to many customers to move their applications and databases off mainframes and onto other, more "modern platforms." He posed a few simple questions to Gartner, including this one in particular: what platform is more modern than the IBM mainframe?

As far as I know, Gartner never responded directly to Clabby's article. But I'm quite sure that at least some of Gartner's analysts recommended migrating mainframe applications and databases to HP Itanium servers running Oracle middleware on HP-UX. (You can see my previous analysis of the HP Itanium meltdown here and here.)

So now I've got this new question: shouldn't those same Gartner analysts apologize for giving such profoundly bad advice to their customers?

My congratulations to Joe Clabby for vastly outperforming Gartner in IT analysis and forecasting. Clabby simply got it right, and Gartner got it oh-so-wrong. More precisely, many (but not all) of Gartner's analysts blew it.

UPDATE: This video speaks to the Gartner analysts who blew it: