Phasing Out our Socket-based Licensing

[richard.cooper]

I guess this would be the case if you upgraded your support program from Basic to Production while remaining with the socket license. Although even then it would not be "hugely" because the difference in these support programs is just a few % of MSRP (5% if I'm not mistaken).

Good Afternoon Gostev,

We requested a renewal cost for Basic but also Production - the "Production" one came back as around 25% more - however it's described as 24/7 - is that a level above Production? If that's the correct one we would have to stick with Basic as that's a huge jump.

thank you

[Gostev]

Hello! This sounds about right I think: 5% in MSRP difference translates into 25% renewal cost difference, as renewal price itself is around 20% of original license MSRP in case of Basic support. All numbers are approximate, I don't know the exact ones but overall it makes perfect sense.

Anyway, I was mostly disagreeing with the usage of English word "huge". Oxford dictionary tells us this means "extremely large; enormous" which is hardly applicable to an increase by a quarter. OK, may be I too would perceive it to be big if it was for nothing, without any benefit. But in this case you get 24/7 support and priority queue in normal hours, for which it seems more than adequate. Heck, we pay 300% more for Business vs. Economy class and even that is not usually referred to as a "huge" difference. Now, Economy vs. First on intercontinental flights is indeed huge

In any case, it is up to you to take this support program upgrade or not. Most of our customers who are currently active on maintenance do have 24/7 support (more than two thirds).

[richard.cooper]

Thanks Gostev,

When you are working with my budget, 25% is huge

As I said above we wouldn't expect out of hours support if we don't pay for it so we will stick with basic for now.

thaks

[apucher]

@gostev: I know, that you are more technician than sales. But probably you can clarify where I am wrong......

Just to make sure, that I understand the VUL pricing and licensing scheme correctly.

I have an example from one of our customers (license and maintenance costs for 5 years):

hypervisor infrastructure:
3 hypervisor hosts, 1 socket each, 32 cores.
64 VMs in total.

Socket based licensing:
Veeam B&R Enterprise, 3 sockets, 1 year basic maintenance: SMRP: 3 x 1.592,10 EUR -> 4.776,30 EUR
4 more years of basic maintenance for 3 sockets: SMRP: 4 x 3 x 350,26 EUR -> 4.203,12 EUR
Total for 5 years socket based lciensing and basic maintenance: SMRP: 8.979,42 EUR

VUL perpetual licensing for 70 VMs:
Veeam B&R, Perpetual Universal License, 70 instances, 1 year production support: SMRP: 7 x 3.236,40 EUR -> 22.654,80 EUR
4 more years of production support for 70 instances: 4 x 7 x 873,83 EUR -> 24.467,24 EUR
Total for 5 years perpetual VUL for 70 VMs: SMRP: 47.122,04 EUR

So if I got that pricing scheme correctly, that would be a 525% price increase!

[Gostev]

Indeed, I've no idea if those numbers are correct or not. But I think I can still answer the actual concern based on more general knowledge.

If you're talking about migrating an existing Socket customer, then that's not how migration from Sockets to VUL works. First, you don't need to purchase new licenses, as we will convert those existing Socket licenses into VUL. Second, there's the migration policy that ensures migrating customers see little to no increase in their annual Veeam renewals costs (depending on the edition). This is achieved by providing them a discount in perpetuity (for the life of customer) as to convert the remaining value of the "perpetuity" factor of their existing Socket license.

If you're talking about a brand new prospect: normally you should drive all your clients to VUL Subscription, which is our primary offering. It is priced most attractively comparing to other options to encourage buying behavior. While VUL Perpetual is designed for rare scenarios when prospects are subject to regulation requiring CAPEX purchases (usually governments) in which case Veeam sales team is always involved. So VUL Perpetual deals are rarely done with simple multiplication of MSPR: this same types of customers always expect massive upfront discounts, so list price no longer means all that much.

Most importantly though, here's the bigger picture you are missing above: 64 VMs on 3 sockets means more than 21:1 VM to Socket ratio, which is insane comparing to industry average. So what we're looking at here is an outlier, and its crazy density is exactly what makes Socket license looks so good here, not surprisingly. However, the reality is that half of the hypervisor hosts protected by Veeam today have less than 6:1 density ratio, which would in turn require 12 sockets to run the same number of VMs, resulting in 4 times higher Socket price.

You're correct that for customers with very high VM per Socket density, our Socket licensing provides a deal that is too good to be true. But this gives all the more reasons for you to get them on Veeam while we still carry Socket licenses on our books! Everyone wins, right?

[mkretzer]

@Gostev
21:1 is insane? We have 69:1 (and we have not one CPU with more than 32 cores).

I guess we will have no fun with VUL. Lets see what sales has to say about this, i asked for a price this week.

To be honest i am kind of worried that the Veeam license will be much more expensive than the hypervisor (VSphere with Enterprise Plus + Production) and OS licenses (Windows Server DataCenter with SA) together for us... And i will hate it when the day comes that our company decides i will have to find an alternative.

I don't really want to become a commvault legend even if something like this would exist, their SWAG can never be as nice...

[Gostev]

Yes, it's insane comparing to industry average. Please don't take my words out of context

We have 69:1 (and we have not one CPU with more than 32 cores).

And that's basically the biggest problem... Veeam socket licensing was designed and priced almost 15 years ago, when server CPUs had 8 cores at best. Fast forward to now, same customers are buying 32-64 core CPUs which are able to run 4x-8x more VMs on a single CPU comparing to 15 years ago.

So if we were to continue with Socket licensing, would it be fair for us to increase the price 4x-8x to account for the increased number of VMs that a single Socket license allows to protect now, as to be able to maintain the same revenue level? Or we just wait another 10 years until 256 core CPUs come, when you'll have 300:1 density, while most other customers will need no more than 10 Socket licenses to protect their entire data center... and just close Veeam?

Luckily we don't have to make this decision at all, as for us it makes no sense to maintain Socket licenses on the books in principle. They are simply not suitable for the majority of workload types Veeam supports and sells to our new prospects these days.

By the way, since you mentioned competitors, as far as I know we're the last enterprise backup company that still sells Socket licenses today.

[mkretzer]

@Gostev
It would be mad to demand that everything stays the same from a company with a product like Veeam! I really understand you there.
I am just no fan of the "one size has to fit all customers" licensing. Why not give the option for a license per TB for VMs? That would also work with cloud applications.

I really love Veeam and i believe that in Business it has to be fair for both sides but not every environment is equal.

How would you explain a 1000 % price increase (we might face) to your company when you gain no immediate additional features?

[Gostev]

But you can never face a 1000% price increase no matter what? Because as the existing customer, you always have the option to remain with your current Socket license. At Veeam we honor the deal we gave to our customers no matter how bad it is for us. In the end, this is the only real way to show how we appreciate everyone who put their trust in us, and stayed with us through good and bad times. Of course, it is impossible to say what happens 5 or 10 years from now - but I know I will leave Veeam even before you do, if I see this company turning into anything other then what it has been since I joined back in 2008.

Now, as far the licensing options for net new customers, as always we will "innovate and iterate" by monitoring the situation closely and introducing any required amendments to our licensing as necessary. As I mentioned, at this time nothing calls for urgent changes because we already have close to 70% of net new business transacting through VUL (while still having Sockets on our books). Plus the transition trend to VUL has been very steep and I expect we'll be over 80% VUL by the time we finally drop Socket licenses from our books. This will be the moment to watch the buying behavior of those remaining 20% of prospects, and adjust our licensing as needed.

[mkretzer]

Gostev, perhaps I have over read something but how will this work when we expand our existing clusters where workload move from old to new hosts all the time?
We buy new licenses for these clusters every year.

How can we mix VUL and sockets in one cluster?

[Gostev]

I see three options for you:

1. Buy additional sockets in advance for future years while they are still available for sale.

2. Migrate to VUL, and buy VUL in future. You don't have to migrate right now, as this will also be available at any time later.

3. Assuming you're large enough customer, sign ELA with your preferred custom licensing terms which work best for you long term. ELAs will be honored even past sockets End of Sale.

And always keep in mind that NO ONE wins at Veeam if we loose you (or anyone else) as a customer. So just keep talking/negotiating with your Veeam sales rep about your specific situation. You were absolutely right above when you said that no one size fits all when VM density varies 5x or more between customers.

[ammaross]

I just ran through our current count and we're at 30.7 VMs per socket in our environment, dragged down in part because we still have some older dual-socket Intel boxes in one of our clusters. Now, the part that is likely throwing off Gostev's numbers is that not all of the VMs are being backed up by Veeam. I don't know if his numbers are true readings of all VMs, or just the ones protected by being in a Veeam backup job.

"By the way, since you mentioned competitors, as far as I know we're the last enterprise backup company that still sells Socket licenses today."
Of course, with the "enterprise" qualifier, this could be hair-splitting all day, but Veeam is not the last company that backs up VMware, Hyper-V, Cloud, etc in one product that still has per-socket licensing.

You're right though, density is on the rise with the added core density for servers, even though you're spending a lot of effort and telemetry numbers to tell us density is still laughably light, even calling 21:1 "insane." It's not. A 32-core (64-thread) system, running normal business-need VMs doing their usual lightweight processing can easily manage 64 VMs. Total CPU utilization rarely goes over 35% and is usually steady at 32% for my 40:1 systems. I'd imagine your own virtualization team is running higher densities than you realize, assuming they're not being pushed into the cloud that is.

[Gostev]

Yes, these are true VM readings from registered vCenter servers across our entire customer base. Usually not all VMs are protected by Veeam.

But these new host systems you're talking about will certainly change these numbers significantly within the next 5 years (typical host hardware refresh cycle for the majority of our customers). You're already there

[mkretzer]

I just got my special VUL quote and compared it to the "normal" price we pay per VM, calculated by our current workload.

With VUL we pay 2,28 times as much as with sockets. I cannot in any way explain that to anyone here! And with VUL as subscription license that license would not even belong to us!
If the license would be somewhat on par with what we pay now we would be willing switch and pay more as our VCPU:CPU rates rise, but not like this.

[Gostev]

We have 69:1

It does not come as a surprise to me that Socket licensing that was originally priced based on VM:Socket density back 10+ years ago is a better deal for your environment and I'm happy you can stick with it.

[mkretzer]

We'll see... The german team did not sound like there is a simple way to get an agreement that we can still buy sockets next year!

[Gostev]

Next year you can still buy sockets normally because End of Sale is on July 1st only, if I am not mistaken.

[pirx]

I just got my special VUL quote and compared it to the "normal" price we pay per VM, calculated by our current workload.

With VUL we pay 2,28 times as much as with sockets. I cannot in any way explain that to anyone here! And with VUL as subscription license that license would not even belong to us!
If the license would be somewhat on par with what we pay now we would be willing switch and pay more as our VCPU:CPU rates rise, but not like this.

You have to be very patient - or careful - with the offers you receive nowadays. We also started with an offer to get current number of VM = VMs in VULs. Then I informed the sales team that there is/was a offer to get the number of VULs according to the density of VMs/Socket. We have a low VM density because of streched clusters and 50% free sockets. Suddenly we were offered a much higher number of VULs which made things a lot more interesting. But this took months of discussion and only with the information given here in this thread I knew what to ask for.

[FrancWest]

We were also offered VUL, but decided not to do it. We had to pay a little more than twice the amount we are paying now per year (that was with a discount). We also had to pay a conversion fee of thousands of euros, just to migrate from socket licensing to VUL. We find this unacceptable, it's the same product, just a different licensing scheme.

[fga]

I can 100% agree to this now.
We got an offer that was absolutly abyssmal.
More than twice the amount we pay now for zero benefit on our side.
This is an absolute nogo. So we are staying on Socket and will NEVER change to VUL no matter how expensive the renewal will become.

[Gostev]

We were also offered VUL, but decided not to do it. We had to pay a little more than twice the amount we are paying now per year (that was with a discount). We also had to pay a conversion fee of thousands of euros, just to migrate from socket licensing to VUL. We find this unacceptable, it's the same product, just a different licensing scheme.

Can you please double check and confirm for me if it's really "the same product"? From your description it sounds like you have a lower socket edition than Enterprise Plus, so migration to VUL means getting access to many more features than what you have right now. In other words, it's not different from instead upgrading your Sockets to Enterprise Plus (which would likewise result in an increase of what you're paying per year).

[FrancWest]

we currently already have a 14 socket Enterprise Plus edition. I can sent you the quotes privately if you're interested.

[Gostev]

Yes, in this case we definitely want to review them as from what I know, they are clearly incorrect. If you don't change the edition, then your renewal price cannot go up no matter what. I will PM you.

[mkretzer]

Same issue here - we have Enterprise Plus, the only change is Basic -> Production support.

[FrancWest]

Yes, in this case we definitely want to review them as from what I know, they are clearly incorrect. If you don't change the edition, then your renewal price cannot go up no matter what. I will PM you.

I've sent you the quotes.

I do have to make an apology though. The first quote we got was the one with more than double the cost. After escalating, we got a second quote with a discount, that one has almost the same cost as we currently have. However, we don't know if that discount is still valid after that agreement ends. So we choose to be on the safe side and renewed socket-licensing for one year.

[vmtech123]

Is the ratio still 7:1?

I get it from the Veeam side, similar to VMware's 32core limit on CPUS before you pay for 2 licenses. I think Microsoft do the same for Sql Always ON. CPU processing is getting crazy.

However, My dilemma is I just bought some new servers, based on current budgets and expectations for the next 5 years planning for growth.
I was just told we are losing a TON of budget over the next 5 years, but felt pretty good that we just increased our storage, etc. We have VERY have low density at the moment due to a bad purchase made from someone a few years ago allowing us 4 extra servers but will be removing those soon.(long story)

If anything I should be able to drop about 8 cores from our budget plans going forward, as our density is low, but I figured I'd keep them on the books for growth and the maintenance price isn't bad.

We are not "Anti" subscription here, but long term savings do matter so we will be going perpetual.

If there are negotiations in the conversion amount say going to 10:1 or 12:1, is it based on the current environment? I really hope there is some wiggle room.

[Gostev]

Normally it's based on the current number of VMs in the virtual infrastructure. But if you have an unusual environment, then you should discuss it with your sales rep and I hope they will consider your situation. I literally just now saw someone bragging about getting "alot of excess licenses" from migrating Sockets to VUL, which means exceptions are possible. It's definitely NOT how the Migration Policy is designed to work though, so for sure there must have been some other environment-specific variable.

And remember, you don't have to migrate to VUL if you're not comfortable with an offer you get... you can just stay with your Sockets, if that's what you prefer. The ONLY change we're making is we stop selling NEW Socket licenses. Existing customers can practically ignore this change completely, with the exception of those who were planning to expand their Sockets license. It is because of them that we're sending the announcement now, more than half a year in advance. We want to give them ample time to get more Socket licenses before they are removed from the books and can no longer be procured.

[pirx]

But mixing VUL and socket licenses is, and will not be possible? If this is the case I don't get the message, because staying with sockets will be a dead end if there will be any growth.

[Gostev]

Mixing Sockets and VUL licenses has always been possible, the only limitation is that VUL cannot be used for VMware/Hyper-V VMs when a Socket license is present.

As for your concern, we rarely see customers growing their Socket licenses. The reality for the vast majority is the opposite: they either remain with the same number of Socket licenses or renew less and less each year, as they are replacing their old hypervisor hosts with modern ones. And with the advance warning about Sockets End of Sale we're communicating right now, we give those few customers who do need to expand in the next few years an opportunity to do so while Sockets are still available for sale, thus reducing the number of impacted customers to an absolute minimum. Or definitely low enough for our sales team to be able to handle them on a case-by-case basis later.

[tntteam]

Hello,

Very happy customer of Veeam for many years, this licensing change is a shock here.
Every aspect of the licensing change (basic support disappear, instance licensing, non perpetual licences) is a loss for us.
We may be a small customer (40 sockets) but seems like it's time to look for another product.
Thank you veeam for securing efficiently our data all these years, now time to move on