This is more of a poll than a technical question. I'm curious as to what everyone's server to user ratio is. I know that is largely dependent on the applications, etc., but my environment is getting a little out of control and I can't seem to get my manager to care. I've tried to find a little info on the net, but didn't find a lot. We've had a 70% increase in total server count in the past 3 years, and our total user count has actually decreased. I am talking about both physical and virtual. The ratio used to be 5.5:1, but now it is creeping extremely close to 3:1. It just seems awfully skewed to me. The biggest problem it is causing at the moment is licensing, but it also takes more time for our monthly maintenance procedures, etc. Any info would be appreciated. Mostly, I just want to make sure I'm not off base in my constant griping every time a new server is turned on and an old one isn't turned off.
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations Andrzejek on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Server to User ratio 1
- Thread starter FloDiggs
- Start date
What are you providing to your users application wise? Just file servers? Application servers? Email? You need to tell us what your environment is.
I'm Certifiable, not cert-ified.
It just means my answers are from experience, not a book.
There are no more PDC's! There are DC's with FSMO roles!
I'm Certifiable, not cert-ified.
It just means my answers are from experience, not a book.
There are no more PDC's! There are DC's with FSMO roles!
- Thread starter
- #3
I'm not asking about how many users I can have per application, or per email server. I can get those recommendations from vendors. I talking about a down and dirty total number of servers compared to a total number of users. For instance, a company with 1500 users and 300 servers would have a 5:1 ratio. A SMB with 150 users and 20 servers would have a 7.5:1 ratio.
-
1
- #4
The number of servers to users is completely baseless. It doesn't matter. What matters is whether the performance of those servers is satisfactory; whether the high availability requirements of your SLAs are being met; and whether you're providing all relevant services needed at a performance level that meets business requirements.
One business with 70 users might be completely fine with a single server, while another business might need 10 servers. It's all about what the business does, and what's needed to do that business. Server to user counts are completely groundless unless you're comparing two completely identitcal businesses.
The same applies to IT staff to user ratios.
Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
One business with 70 users might be completely fine with a single server, while another business might need 10 servers. It's all about what the business does, and what's needed to do that business. Server to user counts are completely groundless unless you're comparing two completely identitcal businesses.
The same applies to IT staff to user ratios.
Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
- Thread starter
- #6
I understand what you both are saying; I know it all too well and agree with 90% of it. But where I disagree is that the user to server ratio is baseless. There is always the possibility for overkill. If you buy 3 server to do what 1 server could efficiently handle by itself, that is over kill. It's expensive, and wastefull. Sure, you'll always meet your SLA for uptime and performance, but if the cost for meeting that SLA isn't taken into consideration you can end up going too far in the other direction. The problem I have isn't justifying more equipment; it is proving that we are wasting time and money on excess equipment. I'm just trying to get a feel for what other organizations see in their environments. I'm looking for any easy to understand statistic that might help the people here understand the problem they are creating.
The number of users connected to a piece of hardware is independent of the load upon that server created by the applications those users are using. It is the load upon the server caused by the number of users and the applications they are using that you should be worrying about, *not* the number of users connected to any one server.
I don't think you understand the issue enough to then explain it properly to your management.
In this thread, 33% of the system administrators don't understand the proper relationship of a servers load.
I'm Certifiable, not cert-ified.
It just means my answers are from experience, not a book.
There are no more PDC's! There are DC's with FSMO roles!
I don't think you understand the issue enough to then explain it properly to your management.
In this thread, 33% of the system administrators don't understand the proper relationship of a servers load.
I'm Certifiable, not cert-ified.
It just means my answers are from experience, not a book.
There are no more PDC's! There are DC's with FSMO roles!
I agree with the other guys, you are focusing on the wrong numbers(users to servers). You should look at server utilization and if 2 servers are under utilized, can they be merged? Then take that info to your boss and explain why it makes sense to merge these servers. 5.5:1 really means nothing in the context you are trying to use it.
RoadKi11
"This apparent fear reaction is typical, rather than try to solve technical problems technically, policy solutions are often chosen." - Fred Cohen
RoadKi11
"This apparent fear reaction is typical, rather than try to solve technical problems technically, policy solutions are often chosen." - Fred Cohen
Completely false. If you take those three servers and configure load balancing and fault tolerance, you're actually increasing uptime, decreasing user complaints, increasing productivity, etc.FloDiggs said:
You've had to tell us WAY more than what you've included. Looking at a particular piece of equipment and determining if it's sufficient or "excess" would take some research. And what YOU and YOUR ORGANIZATION determine for that one particular piece of equipment is completely different than what another org of the same size might determine.FloDiggs said:
I own my consulting company. Two employees. ~30 servers (physical and virtual). One of our clients, a church, has 345 employees, 6 locations, and 19 servers (P & V). Who's "wrong"? Neither. Each has a very defined set of requirements, and the appropriate resources for providing those requirements. Yet two drastically different server to user ratios.
Technical requirements and limitations also come into play. Just because you CAN put an email server application on a domain controller doesn't mean you SHOULD. Sure, you'd get better ROI for that hardware. But you'd introduce technical complexity and security concerns that make it undesirable.
Another org might have two locations, and issues might require that additional servers be placed in the second location to mimick those in the first. So geographical location and network issues cause the server count to be higher - not the user count.
Arbitrarily asking "how many servers should I have" without an in-depth conversation on what's needed is never going to yield information that should be considered tangible and sound.
What about orgs that have test labs for testing applications and resources before placing them into production? That's not only sound, but recommended. Yet it drives the server count higher.
What about organizations that do development of their own inhouse apps? They need development servers.
What about orgs that have fairly structured IT departments and separate logical resources to dedicated servers to reduce complexity?
The questions go on and on.
Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
- Thread starter
- #10
Thank you all for your responses. If nothing else, you have established that I posed the wrong question from the start. The ratio idea was a misguided attempt to find a creative way to get my lackadaisical boss to understand the need for consolidation without getting into the vast details of best practices, system utilization and how they apply to our systems. His technical competence stops with Facebook and YouTube, and anything that actually requires thinking tends to make his eyes gloss over. If you have experience or ideas for pitching server consolidation to non-technical people, I’m all ears. Problems with rack space, power consumption, licensing compliance, and increased administration hours have not triggered any red flags in his mind. Utilization reports tend to be a little too much detail for him to comprehend.
How many servers do you have and why do you feel consolidation is the proper "next step"?
Have you done the numbers on the utilization and determined that less servers would indeed handle the current load and then some?
I'm Certifiable, not cert-ified.
It just means my answers are from experience, not a book.
There are no more PDC's! There are DC's with FSMO roles!
Have you done the numbers on the utilization and determined that less servers would indeed handle the current load and then some?
I'm Certifiable, not cert-ified.
It just means my answers are from experience, not a book.
There are no more PDC's! There are DC's with FSMO roles!
- Thread starter
- #12
We have 46 servers. We used to have 28. We've added a few applications, but we've also removed a few. Some of the increase has been VM's for test, but the majority are old server that had their primary application migrated to a new server and have small apps that just haven't been moved yet. So, they are hardly utilized, and since they aren't causing my bosses boss to breathe down his neck, he's content to let them sit until doomsday. Much of what we need for those servers to be killed is our dev team to address some issues, but the sys admin and I do not have the authority to put pressure on another group. So, it goes back to convincing my manager that the servers need retired and then convincing him to spend some political capital to make it happen. Adding the new test servers, and taking into account a couple apps that used to share hardware but now have dedicated hardware, we should only have 35 total servers (P&V). To complicate it even more, I'm not the sys admin. I'm the net admin, but I back the sys admin up. Part of me just wants to let it go and let those two deal with the consequences, but when major issue arise, I tend to be the one that has to figure out how to fix it.
Honestly I don't think it matters much if a virtual server is hardly utilized. In some cases it makes a lot of sense for some small business app to be runing on a dedicated server as long as it is virtualized. If it really concerns you too much you could schedule it to turn on and off at certains times, like if it's only used once a month or something strange like that.
From a consolidation standpoint, setting up a Hyper-V cluster will give you a LOT. Among the usual virtualization points of reduced power, reduced rack space, reduced cooling needs, reduced network ports, etc, you would also get high availability ("less downtime" to your manager), quicker deployment of new servers (cloning), etc. When paired with System Center Virtual Machine Manager, you also get P2V tools, and the ability to quickly move guests from one host to another (migrations during hardware refresh are MUCH faster). This greatly speeds administrative tasks ("increased productivity" to your manager).
Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
Pat Richard MVP
Plan for performance, and capacity takes care of itself. Plan for capacity, and suffer poor performance.
- Status
Similar threads
- Locked
- Question
- Locked
- Question