Here’s a thought experiment I’m conducting based on the recent PureSystem shipment announcements (the PureSystems #10kandcounting campaign).
With IBM recently announcing that “IBM PureSystems family of expert integrated systems…have now shipped more than 10,000 units through the 4th Quarter of 2013, is this a significant number compared against the number of existing accounts and installations for the IBM i family (Power 4 and above)?
Here are my reasons for asking that question.
- In 2012, Colin Parris then general manager of the Power Systems division told us that the IBM i family (which includes older AS/400, iSeries, and System i machines) is “used” by over 150,000 companies world-wide.
- In a recent IT Jungle column, Timothy Prickett Morgan crunched the numbers and concluded there are 160,000 IBM i and predecessor platform machines installed. 70,000 of those systems were old Power5 boxes, but about 19,400 were Power 7 machines sold since IBM announced the Power 7 on February 8, 2010. Tim’s numbers were greeted with concern in the IBM i community because there were so many old Power5 boxes out there and the number of new Power 7 boxes was small in comparison.
Playing with the numbers I get:
- The number of PureSystems shipped through FY 2013 is about 1/15th of the total number of IBM i family machines throughout the world
- The number of PureSystems machines shipped in the last two years is about 5,000/year (10,000 since announcement in 2012). We’ll call this the PureSystems standard.
- The number of Power i (IBM i running on Power 7 boxes) shipped since 2010 is about 4,850/year (19,400 shipped since 2010 divided by four). We’ll call this the Power i standard.
According to these numbers, the average number of shipped Power 7 IBM i machines and Puresystem machines is almost exactly the same: approximately 5,000 per year.
So just looking at these numbers, I have to ask myself the following questions:
1) If I compare the Power i standard (4,850/year) to the PureSystems standard (5,000 units a year and gaining momentum), do shipments of Power i boxes also look like a winner for IBM based on the PureSystems shipment standard of 5,000/year is good?
2) If I compare the PureSystems standard (5,000 units a year and gaining momentum) versus the Power i standard(4,850/year), do shipments of PureSystem boxes also look like a cause for concern when compared against Power i shipments?
3) What would happen if I add in AIX and Linux running on Power 7 hardware and compare the combined Power 7 hardware shipments against the number of PureSystems (the numbers in points 1 and 2 only involve IBM i on Power 7)? Wouldn’t that put the number of Power system boxes shipped per year above the number of PureSystems boxes shipped per year, and what does that mean in terms of success?
Now, granted I’m simplifying here but my basic question still stands: is it is right to call PureSystems a success because it has 5,000 units a year shipped but we’re concerned about IBM i on Power systems because it’s only shipping 4,850 units a year. And vice versa.
What it boils down to is twofold: 1) If IBM Power i (IBM i plus Power systems) or all Power system shipments (IBM i, AIX, & Linux) are comparable to the PureSystems standard, why aren’t we looking at a Power systems #10Kandcounting campaign or equivalent? or 2) If PureSystems shipments are comparable to the IBM i standard (4,850/year is cause for concern), why are we celebrating PureSystems #10kandcounting while being concerned for IBM i?
Are we looking at apples and apples (two things that can be compared) or apples and gorillas (two things that really shouldn’t be compared).
There may be some very valid reasons these aren’t relevant questions, including the sheer number of nodes put out in a PureSystems installation, the cost of PureSystems boxes versus Power systems (the moolah factor), and the way the market is moving in server consolidation. Power systems boxes also can’t run Windows, which may be a factor. Tim Prickett Morgan has another excellent article on what’s in these 10K PureSystems installation out this week. There’s also the possibility I’m just misreading the numbers. But when I look at just the number of shipped servers (PureSystems and Power systems), I have to admit I’m a little confused as to where the dividing line is between successful server products and server products that we’re concerned about.
Maybe it will become clearer as we move along. For now, it’s an interesting thought experiment to play in the IBM i community, isn’t it?
I’d be curious as to anyone’s feedback on these ideas: where I’m all wet and where I might have a point.