Cloud Managed Services Resources per user

Forums › IT Pro and developers › Cloud Managed Services Resources per user

boflit

82 posts

Master Geek

#247862 27-Feb-2019 15:43

I used to work at the BBC managing a team of IT engineers. When issues arose with performance on TS or VMs then we had a formula based on the user count to work out what server resources were required. I can't remember what the numbers were though, as it was 5 years ago now.

However, back to the here and now, I'm not in IT unfortunately, miss it, but hey ho, my current employer has third party support with everything in the cloud on a TS Server. Since I've started we've added 3 new users, not a lot, but when the available resources were already low, in my opinion, its even worse now. I've asked them to increase RAM and allocate another couple of cores to us but they want more money, even though I have said that as we pay per user, our resource allocation should go up.

I wasn't here when they agreed the contract, and it doesn't explicitly state it that way, although it reads in a way that suggests that, and the management team here understood that was the case too.

So my question really is, what is the normal allocation of RAM and CPU per user these days and in a standard cloud based hosting situation, does a per seat cost cover not server resources as well?

Cheers

dfnt

1512 posts

Uber Geek

Lifetime subscriber

#2188280 27-Feb-2019 16:04

There is no normal allocation when it comes to TS/RDS/Citrix/etc/VDI

It all comes down to resource usage per user, depending on their workflow and what apps they use, then extrapolate from there

There's a few calculator spreadsheets littered around the net that should give you a rough idea

The cloud service is likely priced on resource usage, not per user, usage because its a server object at the end of the day.. so makes sense that they're passing the cost on to you

boflit

82 posts

Master Geek

#2188286 27-Feb-2019 16:16

Surely not? How can you plan how big a TS farm to have if you don't have an estimate of users? As I said at the BBC we required it for support of our financial services division, we had 150 users connecting from India, which is fine for just a connection, but when you add the inevitiable purpose of that connection, i.e. an application, then there is a finite number of users you can expect to go on a server. Hence load balancing across a farm as well. I think from memory in this one farm we had 9 servers.

However, with a TS on a VM, then it should be easier. Assuming not all resources are allocated from the getgo, atleast in terms of RAM anyway, so increasing the RAM on a VM for a TS should be easy enough, and I think should be part and parcel of the manager user cost. If I have 25 users, then 10 cores and 40Gb might be enough with memory utilisation at about 35Gb, +\- a few. Add in 5 more users and all of a sudden you get performance issues. The understanding is, you pay a per seat cost, the beauty of the cloud is that it covers all resources and licensing. When we add new users they don't sting us for more Office licences, cos its part of it, our understand was hardware was also. Now they want to sting us for $200 a month for another 10Gb RAM.

Just doesn't seem right, so I'm wanting clarification from other IT companies that provide managed service per seat pricing as a comparison, I feel like our providers are slinging kak, but maybe that is how it works, but if so then it should be clear in the agreement. Just a shame they didn't have an IT (Ex) person to review the document beforehand, or atleast something with the foresight to think of this scenario.

Beccara

1469 posts

Uber Geek

ID Verified

#2188315 27-Feb-2019 17:07

It's different for every client, One client may have a young workforce living inside full RDP sessions with 10+ tabs of Chrome open that needs 2vCPU/6gb Ram/100gb SSD class storage per user whilst another may use RemoteApp for office applications with an older workforce may only need 0.5vCPU/1gb Ram/30gb Hybrid storage

As for per seat covering server costs, both ways a valid, I've seen MSP's run a per seat + per human fees for everything and others say that the per seat cost covers desktop support and server costs are additional

Most problems are the result of previous solutions...

All comment's I make are my own personal opinion and do not in any way, shape or form reflect the views of current or former employers unless specifically stated

dfnt

1512 posts

Uber Geek

Lifetime subscriber

#2188384 27-Feb-2019 18:37

boflit:

Surely not? How can you plan how big a TS farm to have if you don't have an estimate of users? As I said at the BBC we required it for support of our financial services division, we had 150 users connecting from India, which is fine for just a connection, but when you add the inevitiable purpose of that connection, i.e. an application, then there is a finite number of users you can expect to go on a server. Hence load balancing across a farm as well. I think from memory in this one farm we had 9 servers.

However, with a TS on a VM, then it should be easier. Assuming not all resources are allocated from the getgo, atleast in terms of RAM anyway, so increasing the RAM on a VM for a TS should be easy enough, and I think should be part and parcel of the manager user cost. If I have 25 users, then 10 cores and 40Gb might be enough with memory utilisation at about 35Gb, +\- a few. Add in 5 more users and all of a sudden you get performance issues. The understanding is, you pay a per seat cost, the beauty of the cloud is that it covers all resources and licensing. When we add new users they don't sting us for more Office licences, cos its part of it, our understand was hardware was also. Now they want to sting us for $200 a month for another 10Gb RAM.

Just doesn't seem right, so I'm wanting clarification from other IT companies that provide managed service per seat pricing as a comparison, I feel like our providers are slinging kak, but maybe that is how it works, but if so then it should be clear in the agreement. Just a shame they didn't have an IT (Ex) person to review the document beforehand, or atleast something with the foresight to think of this scenario.

Because during the project you have discovery and/or a BA that sits with business users to determine what they use and how they use it, then you enter that data into a calculator and base your initial specs off that

I guess it depends on how the contract was drawn up, so its probably better to start there..

There is no one size fits all solution, and it sounds like you want a DaaS solution which would be a true per user cost with no care about the resourcing behind it

gehenna

8518 posts

Uber Geek

Moderator

Trusted

Lifetime subscriber

#2188407 27-Feb-2019 19:25

In general there are usually sizing guides available for particular services. Eg this one for remote desktop services https://blogs.technet.microsoft.com/iftekhar/2010/02/10/rds-hardware-sizing-and-capacity-planning-guidance/

timmmay

20587 posts

Uber Geek

Trusted

Lifetime subscriber

#2188408 27-Feb-2019 19:26

I agree that if you're paying a per user fee you probably shouldn't be paying for additional hardware when things get slow. Can you take your business elsewhere?