About Subscription vs Perpetual...

[fsr]

Hi,

I was doing some price comparisons for a project, looking at this page: https://www.veeam.com/buy-veeam-backup-replication.html
And i see that 10 instances of B&R Enterprise cost $800 per year, if you pay for a three year subscription, so $2400 for 3 years.
On the other hand, a two-socket perpetual license costs $1500, and then every year of basic support should be around $330, so for three years it's about $2160. Similar price, but a big difference: with perpetual i can backup whatever amount of VMs i could run on the two sockets, while on the subscription model i would be able to backup only 10 VMs. And i doubt that most people would only run 5 VMs per socket, so it's always going to cost more than the perpetual option. So, why would i want to choose the subscription model? Did i get this right?

Regards

[HannesK]

Hello,
please understand that we're with R&D, so we cannot give official licensing advice in the forums.

And i doubt that most people would only run 5 VMs per socket

agree. but reality teaches us wrong I have seen 700 socket customers running less than 1000 VMs...

Best regards,
Hannes

[fsr]

please understand that we're with R&D, so we cannot give official licensing advice in the forums

Ok, i didn't knew that.

agree. but reality teaches us wrong I have seen 700 socket customers running less than 1000 VMs...

I suppose that those VMs must be really powerful!

[HannesK]

nope - branch offices

[segfault]

I have seen 700 socket customers running less than 1000 VMs...

And then there is our environment: 750 VM's on 18 sockets (9 servers) split into two clusters (with enough spare capacity to be N+1 in each cluster).

And these are not tiny vm's; we dog-pile things like MS SQL for ESRI ARCGIS, JD Edwards, and such on them. We basically optimized our environment for socket / core based licensing models. The trick is two things: SSD based storage arrays and the RAM to socket ratio (1.5TB per two socket server). CPU is not nearly as important as having a ton of memory and very fast disk.

But back to the original point, we avoid per-vm licenses since they are so expensive vs socket based ones. The ROI of doing this is about only a few months when we look at the cost of MS SQL and oracle licenses, plus the apps are faster when you take the money that would have been wasted on licenses and purchase faster servers and storage.

[jmmarton]

It depends on the situation. Part of the equation is density: number of VMs running in the environment vs number of sockets powering the virtualization hosts. But there are other considerations, such as customers who prefer OpX over CapX expenditures. In that situation the subscription route makes the most sense regardless of density.

Also, going with the subscription model gives additional flexibility since the specific workloads are licensed and not the hosts. This means that as those workloads are migrated e.g. to Azure or AWS, you can continue to protect them with Veeam without having to change the licensing. But if you only own sockets today and then move workloads to the cloud, then you have to also add universal licenses into the mix. It may be easier to go all universal today and not have to change things down the line.

BTW, I'm in sales and not R&D so I'm happy to discuss licensing.

Joe

[fsr]

Well, it's a small enterprise with two hosts and around 15 VMs per host. Nothing on the cloud. So, i'm thinking about going with Backup Essentials with perpetual licenses, as with subscription it's more than twice the price, according to the calculator.
And i think that i readed that perpetual licenses can be converted to the subscription model if needed, right?

[Gostev]

How many sockets per host on those 2 hosts?

[fsr]

Only 1 socket has a processor in every host.

[HannesK]

to quote myself from above:

please understand that we're with R&D, so we cannot give official licensing advice in the forums.

in your situation, the socket licensing is clearly cheaper. and you can even use the two sockets of one essential package on two different hosts if you only have one processor in each host. And if the veeam.com website says that conversions are possible, then you can do it

[fsr]

So, now with the change to VUL licensing, by using the calculator, for B&R, 3 years, it shows:
B&R Enterprise perpetual (2 socket) = 4575
B&R VUL 30 instances = 9720

With VUL you get Enterprise Plus (as that's the only option with VUL), and 1 workload of any kind = 1 instance. But it does cost more than the previous considered case of B&R Enterprise instance subscription, which costed $ 7200 for 30 instances, 3 years. Workloads that used more than 1 instance, or less than 1 instance, now all consume 1 instance, and that can make quite a difference up or down.

(By the way, it seems that i made a mistake before. B&R is a 1-socket license, so the cost for 3 years should be $ 4320, exactly the double of the price i initially calculated. Backup Essentials is the one that's sold in 2-socket licenses. I don't know why i didn't just used the calculator ).

-----------------------------------------

By comparing Backup Essentials Enterprise perpetual (2 socket) vs VUL 30 instances, 3 years, the calculator shows:
Backup Essentials Enterprise perpetual (2 socket) = 2745
Backup Essentials VUL 30 instances = 6480 (NOTE: as with B&R, you get Backup Essentials Enterprise Plus and 1 workload = 1 instance)

Just wanted to update the thread to the new values, and check how this changed things.
Regards

[Gostev]

This does not make it an apples to apples comparison though:
1. You're comparing total costs for 3 years, but don't include cost of extra 2 years of support & maintenance in the Socket pricing.
2. You're comparing the price of Enterprise Plus edition capabilities (VUL) to Enterprise edition sockets price.

Anyway, this does not change the fact that with crazy VM densities such as 15 VMs per socket, socket licensing is very lucrative for obvious reasons. However, currently the median density in our customer base is exactly 6 VMs per socket, so for many customers the situation is quite opposite.

To clarify: "median" is different from "average", in this case it means that exactly half of Veeam customers have density lower than 6 VMs per sockect, and another half has density higher than 6 VMs per socket. Median value works better for statistical purposes, as it does not get skewed as much by rare or non-typical environments (e.g. VDI).

[fsr]

Well, i used the calculator, so i think that the cost of the additional two years of support should be there. And also clarified that you get Enterprise Plus with VUL. But Enterprise Plus is not needed in this case, so i compared the cheapest option that meets the requirements in both licensing options.
I just compared this particular use case. There are some very small VMs there. Others, not that small in size, but almost all of them have quite light processor requirements.

At 6 VMs per socket, VUL would be clearly better than perpetual. Not only more flexible, but cheaper. That would be 5 cores for 30 VMs and that would cost $ 11437 for 3 years in B&R perpetual vs the $ 9720 for 30 instances in VUL for the same 3 years.

[Gostev]

That's exactly right - and that was the whole point behind the introduction of this licensing model. Now, no matter where the particular environment is located on the VM density spectrum, everyone has the ability to pick the most optimal Veeam licensing model: for some it will be per-socket, and for other per-workload.