Cat is out of the bag now, and the #vTax is dead. My bet is that VMware might be a tad bit pissed that this leaked ahead of schedule (like the CEO change) due to leaks and or rumors even though the article notes that this has not been confirmed (code for yes its been confirmed but no one wanted to lose their job for going on the record). I had been hearing about this for a while now but as usual did not want to discuss it openly as it appeared to me as an embargoed topic of discussion. I’m guessing that this was going to be release at VMworld as part of either a keynote or some other conference discussion. That said, this news will make a lot of the SMB crowd very happy.
The original #vTax was highly unpopular, so much so that I had to re-evaluate how our clusters were going to be built and deployed. I think initially VMware had bad data behind the vRAM entitlements based on the very small consolidation ratio’s they cited as part of their reasoning (5:1) not to mention they may have simply misjudged how badly the idea would go over within the community. Yes they made a change to the memory per socket allotments by doubling them, but even then, that wasn’t enough for a lot of customers who had scaled up significantly, and with RAM being dirt cheap the whole concept of tying licensing to memory just sucked. Not to mention, no one can out Oracle Oracle when it comes to licensing evil.
This also takes away a key leg of the Microsoft Hyper-V 3 marketing playbook. Dropping the convoluted memory based licensing and moving back to per socket will make the smaller customers more likely to not focus as intently on Hyper-V3 as they may have been. With its imminent release this October, the announcement now may take a little wind out of those sails. Still its hard to discuss a product that isn’t even shipping.
Now if only all VMware products were licensed per socket (*cough SRM/VCOPS/Etc *cough).
Edited to add:
One thing I did notice when I was railing against the #vTax when it first was announced was that after the dust had settled, and I did the calculations, I wasn’t actually going to have to initially change anything to my existing clusters as they were conceived and built, it was only as I was planning to move forward that the licensing changes would start to affect me negatively. I think that if you started to actually analyze a good swath of the customer base, you would find that they were in the same boat and what essentially you had was a disconnect between the customer and the client and some poor execution and messaging. Ultimately though, the idea did rub a lot of people the wrong way and the chatter involved (much of it negative) gave what many view as a company that really has a lot of “love” thrown its way getting its first significant black mark against it.
Nate over at TechOpsGuys has several posts that elaborate on his thinking on the situation and I’ve followed those posts since their inception. Worth a read as well.
Pingback: 3 Reasons These VMware vTax & Licensing Rumors Are Great - The Lone Sysadmin
I think the big problem with the vRAM tax was that the licenses were still tied to physical CPUs. On the one hand they were trying to make the licensing all cloudy by basing it on memory, which is probably the biggest constraint in any cluster, but still trying to tie it to physical hardware. Just choose one or the other – if you want to go consumption, then go all the way and let me run the software on as many physical boxes as I see fit. Otherwise stick with physical – a hybrid model was never going to work.
Pingback: Why VMware shouldn’t give up on new licensing - The Virtualization Room