Decision on Remote Support Tool- ScreenConnect
-
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
Now if you install on Linux and never use it form a Windows host you would not say it is bad to use. It is strictly the comparison between the two install methods.
We moved from Windows to Linux. To keep the performance parity, we couldn't go below 50% cost on Linux. But the move from Windows to Linux was a slam dunk. Because of the use of Mono, it's not the 75% cost reduction we would normally expect to see, but it is still significant.
-
@dashrender said in Decision on Remote Support Tool- ScreenConnect:
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@ambarishrh said in Decision on Remote Support Tool- ScreenConnect:
So I am at this point leaning towards screenconnect and guess installing it on CentOS would be a better option? Or should i consider a windows server, assuming that it could be tied up with our local AD? Please advise
There is a slight advantage to Windows, but not enough of one to overcome the costs and overhead (e.g. throw the same resources at a Linux box and you get more from it.) We run on CentOS and it is great.
If you have the licensing in hand already, install it on Windows. Performance is much better.
They are looking at updating the Linux version to work with .Net instead of Mono. But until they do, it sucks comparatively.
Now if you install on Linux and never use it form a Windows host you would not say it is bad to use. It is strictly the comparison between the two install methods.
I've seen the performance issues then... my SC boxes seem to need to be rebooted almost monthly.
I'm not confident that that is true. Gene reboots them regularly, that's not at all the same as that being what is needed. Also, there were not dissimilar issues on Windows. So you can't read into Gene rebooting as Windows not having issues. It's possible the two are related, but there is no reason to make that assumption based on the observations.
-
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@ambarishrh said in Decision on Remote Support Tool- ScreenConnect:
So I am at this point leaning towards screenconnect and guess installing it on CentOS would be a better option? Or should i consider a windows server, assuming that it could be tied up with our local AD? Please advise
throw the same resources at a Linux box and you get more from it
No Scott, this is not true.
You mean I should have said WAY more from it? It's not close from what we've seen (using cost resources as the guide.) We get close to parity performance at under half the cost.
You obviously do not use ScreenConnect for most of your work day.
The performance difference between the two is huge.
I have migrated my system back and forth more than one time to prove it.
Make a VM with the same vCPU and memory settings. install Windows Server 2012 R2 (have not retested since 2016 was GA) in one and CentOS 7 in the other. The user performance from the Windows instance will be massively better.
Do not talk about things that you do not actually try.
The threads and issues that you guys had when migrating are still posted on ML. Additionally so are mine from when I migrated in a different thread.
-
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
Now if you install on Linux and never use it form a Windows host you would not say it is bad to use. It is strictly the comparison between the two install methods.
We moved from Windows to Linux. To keep the performance parity, we couldn't go below 50% cost on Linux. But the move from Windows to Linux was a slam dunk. Because of the use of Mono, it's not the 75% cost reduction we would normally expect to see, but it is still significant.
I highly doubt you actually have user experience performance parity as it has been proven and taken to ScreenConnect support by more than myself.
-
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
Make a VM with the same vCPU and memory settings. install Windows Server 2012 R2 (have not retested since 2016 was GA) in one and CentOS 7 in the other. The user performance from the Windows instance will be massively better.
Ah, that's the rub. You are looking at the same "vCPU and RAM" when large enough to run Windows and applications. We run ours hosted and the cost of running Windows requires double the RAM and more than double the cost of running Linux. So for the same money, we get more power on Linux, so for the same financial resources, we get better Linux performance. We flipped back and forth too, and Linux won out here.
If money was no object and we were throwing lots of RAM at it to overcome Windows bloat, then yes, beyond the "plenty of RAM for Windows" threshold without cost as a factor I'd expect Windows to be faster.
-
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
Now if you install on Linux and never use it form a Windows host you would not say it is bad to use. It is strictly the comparison between the two install methods.
We moved from Windows to Linux. To keep the performance parity, we couldn't go below 50% cost on Linux. But the move from Windows to Linux was a slam dunk. Because of the use of Mono, it's not the 75% cost reduction we would normally expect to see, but it is still significant.
I highly doubt you actually have user experience performance parity as it has been proven and taken to ScreenConnect support by more than myself.
One could say it was proven the other way, as well. What performance issues are you seeing? I don't use it constantly like a lot of people do, but slow downs are not an issue that we are seeing.
-
Consider that ScreenConnect only needs 512MB of RAM to run on Linux... it's hard for Windows to compete. We have way more RAM than that for it, but that's all that it decides to use regardless. (Uptime of 15 days for it to have built up, too.)
-
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
Make a VM with the same vCPU and memory settings. install Windows Server 2012 R2 (have not retested since 2016 was GA) in one and CentOS 7 in the other. The user performance from the Windows instance will be massively better.
Ah, that's the rub. You are looking at the same "vCPU and RAM" when large enough to run Windows and applications. We run ours hosted and the cost of running Windows requires double the RAM and more than double the cost of running Linux. So for the same money, we get more power on Linux, so for the same financial resources, we get better Linux performance. We flipped back and forth too, and Linux won out here.
If money was no object and we were throwing lots of RAM at it to overcome Windows bloat, then yes, beyond the "plenty of RAM for Windows" threshold without cost as a factor I'd expect Windows to be faster.
I never said which was more financially performant, and neither did you, until just now.
I run my instance on CentOS for the same reason accepting the shit ass performance difference.
But if I had to use it more daily than I do, I would likely accept the higher cost of WIndows for the better performance. The lag and sluggishness or horrible comparatively.
-
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
I never said which was more financially performant, and neither did you, until just now.
I did, go back and look. I originally said resources, then clarified that it was financial resources.
-
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
But if I had to use it more daily than I do, I would likely accept the higher cost of WIndows for the better performance. The lag and sluggishness or horrible comparatively.
Where do you see the lag? In starting up sessions?
-
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
But if I had to use it more daily than I do, I would likely accept the higher cost of WIndows for the better performance. The lag and sluggishness or horrible comparatively.
Where do you see the lag? In starting up sessions?
Using it. All the time.
-
I suspect that around 6GB of RAM, Windows is faster - because you get past the point where Windows is needing more and Linux keeps adding RAM that it has no way to use.
At the ranges where Linux runs well and scaling up doesn't make much sense, Windows will struggle to run just the OS. I bet if we tested both at different RAM points, you'd find a curve where Linux outperforms dramatically until you get rather high in RAM. Then Windows would take over. But only at immense cost.
-
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
I suspect that around 6GB of RAM, Windows is faster - because you get past the point where Windows is needing more and Linux keeps adding RAM that it has no way to use.
At the ranges where Linux runs well and scaling up doesn't make much sense, Windows will struggle to run just the OS. I bet if we tested both at different RAM points, you'd find a curve where Linux outperforms dramatically until you get rather high in RAM. Then Windows would take over. But only at immense cost.
Stop speculating and actually try it. It is not even close to comparatively performant.
2 vCPU
4GB RAMWindows 2012 R2 versus CentOS 7
I have done this many times.
-
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
But if I had to use it more daily than I do, I would likely accept the higher cost of WIndows for the better performance. The lag and sluggishness or horrible comparatively.
Where do you see the lag? In starting up sessions?
Using it. All the time.
Like once IN a session? I don't see that at all. Once the session is set up (which was never instant - Windows with double the RAM or not) it's surprisingly fast.
Like I said we moved from Windows on Azure to Linux on Digital Ocean with less than half of the assigned resources and did not see it slow down. If we dropped much below half, then it was slow. But at the half point, we saw parity.
Maybe more recent updates have changed this, but half the RAM, just as fast here. Way under half the cost.
-
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
I suspect that around 6GB of RAM, Windows is faster - because you get past the point where Windows is needing more and Linux keeps adding RAM that it has no way to use.
At the ranges where Linux runs well and scaling up doesn't make much sense, Windows will struggle to run just the OS. I bet if we tested both at different RAM points, you'd find a curve where Linux outperforms dramatically until you get rather high in RAM. Then Windows would take over. But only at immense cost.
Stop speculating and actually try it. It is not even close to comparatively performant.
2 vCPU
4GB RAMWindows 2012 R2 versus CentOS 7
I have done this many times.
But we DID try it. Linux whomped on Windows.
-
Just for reference, and maybe this matters, ours has long been on Fedora. It is Fedora 26 now. Maybe that is a bit faster than CentOS?
-
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
I suspect that around 6GB of RAM, Windows is faster - because you get past the point where Windows is needing more and Linux keeps adding RAM that it has no way to use.
At the ranges where Linux runs well and scaling up doesn't make much sense, Windows will struggle to run just the OS. I bet if we tested both at different RAM points, you'd find a curve where Linux outperforms dramatically until you get rather high in RAM. Then Windows would take over. But only at immense cost.
Stop speculating and actually try it. It is not even close to comparatively performant.
2 vCPU
4GB RAMWindows 2012 R2 versus CentOS 7
I have done this many times.
But we DID try it. Linux whomped on Windows.
No, you moved one time. Between totally different backends.
On the other hand, I setup a test environment on the same hypervisor with two identically configured virtual machines. Installing CentOS 7 on one and Windows Server 2012 R2 on the other. Then I backed up the ScreenConnect system and restored it into each.
-
Just so that you can see why going to 4GB would artificially favour Windows, this is the RAM usage on Fedora. Anything over 1GB of RAM is totally wasted.
$ free -m total used free shared buff/cache available Mem: 1999 473 170 1 1354 1311 -
@scottalanmiller said in Decision on Remote Support Tool- ScreenConnect:
Just so that you can see why going to 4GB would artificially favour Windows, this is the RAM usage on Fedora. Anything over 1GB of RAM is totally wasted.
$ free -m total used free shared buff/cache available Mem: 1999 473 170 1 1354 1311Well that is provably false. Here is CentOS 7
-
@jaredbusch said in Decision on Remote Support Tool- ScreenConnect:
On the other hand, I setup a test environment on the same hypervisor with two identically configured virtual machines. Installing CentOS 7 on one and Windows Server 2012 R2 on the other. Then I backed up the ScreenConnect system and restored it into each.
I get it, but that's a lot of resources. Try it at 1GB between the two, and I'm confident you'll find exactly the opposite. That Linux is dramatically faster. 4GB is a ridiculous amount of RAM for a workload that should be very light.
