IBM Licensing Nightmares
I have to second el.b00ty's post from this past January. I almost thought it was me; our shop is very very similar. I know some folks don't think long term, but I do. I'm always thinking 5 years down the road, building my infrastructure today to meet the needs tomorrow.
But when I see PVU Licensing schemes from IBM, I start to shutter. Yes, the license is still cheaper than the hardware, but these costs add up quickly. Yesterday, I had a single core, dual processor system handling Domino. So we were licensed for TSM TDP and the BA Client at 200 PVU as well as Domino itself. Today, I have a single socket quad core HX5 doing the same job. The PVU costs went from 200 to 280. And so what? It is only 200 PVU, right? Wrong! Start multiplying those costs across all the hardware.
My costs for doing the same bloody job (nothing changed, just the hardware) doubled in 5 years! Only hedge fund managers get that rate of return (and apparently IBM). And to top it off, IBM Increased it's PVU core costs from 50 to 70 instead of leaving it at 50 (for the dual core systems). In a few years when Quad Cores are dinosaurs and all I can lay my hands on are Octicores, my costs more than double AGAIN; 560 PVU to still do the same job! That is provided IBM doesn't raise the costs of the Octicores from 70 PVU to something higher in the interim. They obviously have no problem raising their PVU licensing costs. Thank God the Insurance Companies haven't taken notice!
And I don't want to hear about the "you can disable cores" kind of talk to save on licensing. I didn't buy the hardware to cripple it, I bought it to do a job! It doesn't seem appropriate to punish a business with additional licensing costs when they are just buying hardware upgrades, and that business doesn't really have a choice on the processor cores any more.
Yes I know, more cores means more work by a single system and we can consolidate. Great! For some folks that may be true, good on them for trying to save money by consolidating. For other shops, that is not a possibility. Why would I consolidate ALL of my Domino infrastructure into a single system. Looks too much like the "all of my eggs in one basket" kind of thinking. That is a bad idea and it is just not worth the risk if the system goes down. (BTW - Quality from IBM hardware is declining if you haven't noticed ... look for loose screws in your HX5's, I found them).
So, what do I see happening? I see people dumping IBM Software in droves! That is what I see. And I have already seen it happening! The pace will only increase over the next few years. And IBM will just raise their costs on their existing customers to make up the difference until somebody gets a clue and figures out why. We almost dumped some IBM Software this past budget and you can rest assured that the question will come up again in future budgets (just for the record, I fought against dumping Domino). In the grand scheme of things, IT folks have little power on how much they can spend. You make your case and you get a budget. But no finance manager is going to let you double your software costs every 5 years! I can guarantee it! Sooner or later, you all will face the decision, "I can keep running IBM software and Dump a lot of the other things we need to actually run the business, or we switch vendors and keep the business running". Kind of a no-brainer to me!
BTW - here is the real reason for my rant. The amount of work necessary to do the calculations to determine the correct number of PVU is too time consuming. I can easily tell you how many systems I have, and how many processors (which is irrelevant for just about every other vendor software solution we employ) are in each system. But getting down in cores, processor type, processor vendor, it's a lot of work! Too much work! What IT Shop isn't short staffed already? And now I'm forced into doing work I shouldn't have to do counting cores and identifying processor types to determine if I have to buy 50 PVU or 70 PVU, or 120 PVU. So I complain face-to-face to our sales rep that a better solution should exist for licensing and that costs are getting too high with multi-core systems. His response, "We have software you can purchase that can do this for you; licensed on the number of cores it finds of course." I kicked him out! He obviously wasn't listening to what I was saying.
I hope someone from IBM sees this post and does start to listen. Next budget, I can guarantee you, IBM is going to lose some business from us (maybe more than just software). We may be one business but I know for a fact, we are not the only business considering the switch.