- This topic is empty.
29. January 2008 at 19:35 #892imstarunitMember
- Total Post: 7
I’m just trying to get a rough idea of how devices have people been able to completely image at one time. I currently have WDM Workgroup edition installed on glorified workstation hardware.
-WDM Server hardware is: IBM Thinkcentre A61
-WDM server is on 1GB Full network port
– thin clients currently on 10Half ports (plan to move thin clients (v90) to 100F Half ports)
It takes about 30 mins to image 5 devices concurrently (the max for WDM workgroup edition) if the devices are on 10Half ports. We plan to upgrade the network soon so that all devices will eventually be on 100Full but for now we are stuck with 10Half.When using 100Full, the same image takes about 15mins to complete all 5 devices.
I would like to upgrade to WDM Enterprise to increase the number of concurrent devices that I can image at one time but need to provide some ammo when proposing this to my company. So, short of actually doing a stress test with say 10,20, 50+ devices, i’m just trying to get a rough idea of what this setup can handle before we start hitting some bottlenecks or significantly increasing the time it takes to image all the devices concurrently.
Appreciate any feedback.
Imstarunit29. January 2008 at 22:17 #11516pauladamMember
- Total Post: 12
- Regular Joe
I have only attempted about 7 or 8 (Model 9150se) at once so far.
Although there are a total of 60 Terms in place at my clients site I left the rest of the imaging to the onsite IT guy. I’ll email him to find out if he’s done the rest of the terminals and how many he successfully did in one go.
The hardware in use at the moment is a fast HP Desktop (new but not 100% on the CPU) – its got 2gb RAM and is connected to a Gbit Cisco 3560 switch which is then uplinked via 2xgbit SFP etherchannel to a 100mbit 3560.
All of the terms and client switchports are all set to Auto/Auto – so they should be negotiating at 100/Full.
I would say it still took about 10/15 mins to run through a full reimage of the 7 or 8 terms I did at once – But in saying that, one terminal took that long as well.
I guess the first bottleneck Im going to hit will be disk IO – as Ive just got a single disk with the OS, SQL, WDM and its Repository…..
But I’ll still find out what has been done since I left site and will keep you posted – likewise, if you get any performance stats to post that would be good to.
Paul Adam29. January 2008 at 23:31 #11518imstarunitMember
- Total Post: 7
Through my limited testing I have produced some pretty consistent results. I’ve only tested up to 5 v90’s on Windows 2003 / WDM. Whether imaging a single device or up to 5 devices the times have always been ~30mins for devices on 10Mb Half ports and ~15mins for 100Mb Full ports.
I basically thought the same thing as yourself.. that the disk IO may be the initial bottleneck. The A61 only has a single SATA drive. I think the disk would probably suffer before something, like the NIC, becomes saturated. Currently, at about I’m not sure how caching would behave here and whether the memory subsystem would would be another bottleneck.
Anyways…it’s good to see that your results a more or less consistent with what i’ve seen (i.e ~15 mins to re-image a number of devices concurrently).30. January 2008 at 5:48 #11521thinkthinMember
- Total Post: 1707
- Jacked into The Matrix
One other thing to add is that maybe in the short term if you can team the server NIC’s to two 100 Mb/s ports and if the clients are 10 Mb/s but on a switch you should see improvements.
I have imaged aprox 20 with two 100 Mb/s NIC’s in a switch assisted team and found that was about as much as the network and disk I/O would take. CPU and memory really did not get used that much. Remember really the question is “how many concurrent FTP downloads of a 512 MB file can my server support”
If you have a WAN, WDM enterprise allows you to locate images closer to the devices and this really makes a big difference,
- You must be logged in to reply to this topic.