We are facing a reorganization of the way we handle licensing and putting some order with floating license servers we will keep on premises.
SW SNL is currently running on a VM, with like 2 cores @2.1GHz 8GB RAM and 90GB virtual disk, VMWare hypervisor.
The host server (windows server datacenter edition) will be dismissed in the near future so we need to plan how to cope with it while managing other two license servers for other CAD platforms switching to a floating license model.
Windows datacenter edition license allows to run as many VM as needed, but it costs an eye and it is not worth the workload of three license servers.
We have two options:
1.buy 3 low specs servers to run the floating license server service.
windows 2022 std edition
buy one decent server to run 3 low specs VMs, and since windows 2022 standard edition allows to run up to 2 VMs, we buy an additional OS license to cover 3 VMs.
As for option 1, to keep costs down spinning rust and software raid are an option, but honestly I would go with a HPE rack mounted server solution with a dedicated raid controller. I am thinking about a mirroring with an hot spare. ssd drives are less likely to fail than hdd ones, but they cost more.
On the other hand I have no clue on VM and their host requirements.
VMware will be no more licensed to us, so we may need to switch to a hyperV environment.
This is for 3 machines running a flexlm like service, so they basically poll the network for some dozen machines at intervals. Nothing more.
Check out cost.
SSD do fail. They fail to read only. Don’t think there are much read/write to license server. So SSD should last.
Don’t think you need RAID on license server. Just need a way to recover fast.
I like 3 servers. Not much chance for all 3 to die at the same time.
While I agree on a fast recovery process, Raid 1 is only to minimize downtime and a faulty disk is likely to happen on a machine turned on 24/7. For mechanical hdds it was like every 5 years.
I prefer to save money on memory and cpu rather than disks. Having three small VMs on a single machine coukd optimize costs and simplify the backup operations of the 3 Vdisks, but in case of a fault we need another machine setup asap.
HPE has different kind of assistance packages, iirc 3 years next business day with parts delivered on site?
Not sure why you feel you need 3 servers… We routinely run a dozen or so different app’s license managers on one server. As long as the port number is different for each (change in the license file), there is no issue. The server is a Hyper-V VM in our datacenter which can be moved to any other host automatically if the current host starts to fail or does fail so it is also resilient. I believe we were even running SNL when we were testing out Draftsight a long while back.
Same here, one server hosts licenses for SolidWorks, PLECS, Quartus, and Mathcad. Each has their own folder with flexlm files and each runs its own Service.
Yes, only time we have ever had a problem is if 2 apps license files have the same port number defined. It’s an easy fix to edit the license file and define a different port if that occurs.
IT wants to keep them separated so if you have a reboot or a botched update you have only one product down noy all of them.
They told us (cad managing group) they are going to dismiss the datacenter edition server running the VMs and we have to take the problem on our not so big budget.
@bnemec well you are right, but it is more a company political issue than a technical one right now.
IT still dictates the server guidelines…
On a side note in case of a SW license upgrade we do a complete SNL uninstall after deactivation and license log backup, that usually triggers a reboot of the whole server since some libraries are still in use somewhere and it may affect other software in case of multiple license server running concurrently, but I do not know how long is the timeout for those licenses, but I think it should be long enough to cover a machine reboot or at least in the minutes (10?20?) order of magnitude, but from my not so long experience about a HPE server running on HDD, a reboot takes literally forever.
Running multiple machines at the same time could also help redundancy in case of fault, but I have yet to study the issue and I admittedly no experience with it.
I suppose…but I’ve never done that in 25 years. Well maybe once or twice that I don’t remember for some reason. But in general I just do an inplace upgrade.
we do version up on weekends and our VAR is normally not available unless we explicitly require them to be on standby to support us.
If something happens we are likely to wait until monday. At least this is what I was told.
And I think we are invoiced separately for that kind of support…
oh, well just to put it in perspective, they are the IT Dept. and do whatever they deem need to be done, while I am the guy blamed for ALL SW troubles.
