Going Full Spiral, Not Coming Full Circle

January 17, 2011 Timothy Prickett Morgan

The Mayan calendar says the world is going to end in 2012, and in the unlikely event that doomsday should indeed come to pass, then 2011 will be the last year we have to worry about anything and make predictions about the future. But I doubt very much humanity will get off that easily. We surely have not for the past several million years from the studies I have read. The end of the world is a metaphor, and things long gone often return again, albeit slightly differently.

I was talking to a dear, old friend this week about the strangeness of life, about how people and places come and go, and then come back again, but usually in a way you didn’t expect. If you think of history–your own personal one or the big one we all participate in collectively and study in school–we are often told that it is cyclical, that things come full circle. As far as I can tell, this is not an accurate metaphor. What I think happens is that things go full spiral, pushing out from the center while also orbiting a point in the space-time continuum. You come back to the same relative position on the curve, but you’ve moved outward too, and now you can see things with a little perspective. And then you do it again. You can’t go back, you just keep moving and then some day, you get a sense of déjà vu, that you have seen this all before, but it is different. This makes sense if you think about it. A spiral unites circular motion with linear motion, the sense of the past with the onward progress of the future.

Maybe I have just had too much coffee today. Or too little.

So what does 2011 hold for the Power Systems-IBM i platform, and all of us who are in this ecosystem? A good question, and it would be refreshing to have more certainty than we do. But I have been thinking about this a lot, as I do this time of year, and I have a few thoughts to share.

To begin with, I expect IBM to post better results for its Power Systems line in 2011 than it did in 2010. By definition, this would have to be the case.

As we go to press with this issue of The Four Hundred, IBM has not yet revealed its financial results for the fourth quarter of last year, so it is hard to say what the relative compare will be in 2011. For the past decade or so, IBM has managed to get about $5 billion a year in hardware sales for its Power-based server lines, regardless of the labels it slapped on them, but in 2009, sales fell by 15 percent to $4.3 billion and in the first nine months of 2010, sales for Power Systems fell further by 13 percent to $2.6 billion. So if IBM has a great Q4 2010 and fills in some of the gap, it will very likely not be enough to get anywhere close to $4 billion in revenues. Crawling back to 2009’s sales levels in 2011 will not be a return to normal. Or more precisely, having a Power Systems business that generates a fifth lower revenue than IBM, its resellers, and its ISV partners are used to might be the new normal. We’ll see. I remain skeptical that Power Systems–and indeed, any proprietary or RISC technology–can return to pre-2008 sales levels because of the maturity of Windows and Linux and the very good bang for the buck that Xeon and Opteron systems offer.

That said, I will be happy to be proven wrong here. Thrilled, in fact.

One very easy and safe prediction is that there will be more consolidation in the IT industry, including private equity takeovers as well as company mergers. This will be the case so long as there is cheap money available to fund takeovers, thanks to the efforts of the U.S. government and its European and Japanese counterparts as they try to stimulate their economies. The wealthy and their hedge fund and private equity draft horses will be looking for somewhere other than the stock market and precious metals to stash their money so it can grow faster than the stock market, which is back to its pre-2008 levels now. Do you honestly think the state of the actual economy in the U.S., Europe, or Japan justifies the Dow Jones Industrial Average at nearly 12,000, the FTSE at over 6,000, and the Nikkei at over 10,000? I don’t. But please, don’t tell anyone because my retirement is in a 401(k).

So I see a year where the big get bigger, competition in established IT sectors gets calcified, and customers get turned upside-down for a good shaking to see what drops out of their pockets. And more importantly in the IBM i base, I see a lot of company founders who have been at this since the System/38 days deciding that now is a good time to cash in their chips and retire. Some of the biggest names in the AS/400 market have done it, and I expect more to do so. In fact, I would guess that a few of these privately held companies that we all know and love have already done so, but because they are private and don’t talk about it, we don’t even know it happened.

The one thing that I am not expecting a lot of is updates in the Power Systems hardware and software stack. As I already showed you early last year, the IBM i roadmap doesn’t have i 7.2 coming out until the first half of 2012. There are two updates expected in 2011, however. That would be i 7.1.2 coming in the first half of this year and i 7.1.3 in the second half of 2011. I have no idea what either of these will include in terms of features, but I plan to do some digging to find out.

I am not expecting Power7+ processors to be announced any time soon, either. IBM is not under pressure to revamp its hardware until the third quarter, when Intel will get its eight-core “Sandy Bridge” Xeons and Advanced Micro Devices will get its 16-core “Interlagos” Opterons into the field. I don’t think we’ll see Power7+ iron until early 2012, in fact. That said, I saw in the Linux for PowerPC mailing list that the Power7+ hardware type was added to the Linux kernel in November 2010. And here is a bug fix for DB2 on AIX that affects machines with Power5, Power5+, Power6, Power7, and Power7+ processors.

So maybe there is Power7+ iron that will come out earlier than 2012.

No matter what, Power7+ chips will very likely be a shrink from the 45 nanometer wafer baking processes with Power7 chips down to 32 nanometer processes. I anticipate that IBM stays with the same basic chip design, with four, six, and eight cores activated and with the same L1 and L2 caches. The process shrink should allow Big Blue to crank the CPU clocks somewhere between 25 and 30 percent, and support DDR3 memory that runs at between 1.6 GHz and 2 GHz. The new Power7+ chips will no doubt be socket-compatible with the current Power Systems 7XX line, and if not, then someone should be fired.