poor network performance
First some background on our setup...
1. Dell server running Windows 2012 R2 with all latest updates applied. The server has a Broadcom Gigabit Ethernet card (quad port). Currently IP v6 is disabled as we do not use it. The Windows environment is a host environment currently with one virtual server.
2. Other old servers all running Windows 2003
3. All servers and workstations are plugged in to Netgear ProSafe 24-port Gigabit Smart Switch.
That all said, our challenge is that the performance of the network throughput seems very slow on the virtual server. Check out the attached screenshot. This screenshot represents the activity on the virtual server when we are migrating one user mailbox from the old mail server over to the new mail server (which is the virtual server). The highest peak for a short period of time goes to the 7.7 Mbps mark but generally speaking the entire process appears to be very slow for a gigabit environment
Any suggestions on what to check to improve the network performance of the W2012 server?
ethernet.JPG
1. Dell server running Windows 2012 R2 with all latest updates applied. The server has a Broadcom Gigabit Ethernet card (quad port). Currently IP v6 is disabled as we do not use it. The Windows environment is a host environment currently with one virtual server.
2. Other old servers all running Windows 2003
3. All servers and workstations are plugged in to Netgear ProSafe 24-port Gigabit Smart Switch.
That all said, our challenge is that the performance of the network throughput seems very slow on the virtual server. Check out the attached screenshot. This screenshot represents the activity on the virtual server when we are migrating one user mailbox from the old mail server over to the new mail server (which is the virtual server). The highest peak for a short period of time goes to the 7.7 Mbps mark but generally speaking the entire process appears to be very slow for a gigabit environment
Any suggestions on what to check to improve the network performance of the W2012 server?
ethernet.JPG
Moving mailboxes is not a good indicator to check network performance, how about a file transfer over network. Try to transfer one huge file (2012 iso file or something) and check speed on that?
Are you using Hyper-V?
Are you using Hyper-V?
Here is a screenshot of transferring a 2.9 GB ISO file (while that one user mailbox is still being migrating too). It started off pretty high (22-26 MBps) but then it dipped (as noted in the screenshot). Now it looks like it is back at around 18-22 MBps. That is better but is still low, right?
Yes we are using MS Hyper-V
autocad-transfer.JPG
Yes we are using MS Hyper-V
autocad-transfer.JPG
Also....if it matters any......the server from which the ISO file is being copied from is the same server where the mailbox is being migrated from.
I agree with Saschao -- start by determining your specific problem. The utility FPING (www.kwakkelflap.com) can give you a real good idea of your network round trip time using this command:
fping <target> -t 0 -n 100 -s 32
This will send 100 packets to the target. Test this from various machines in combination on your network. Start with two physical workstations next to each other; then hit the physical server; then hit the VM server, watching the difference in performance all the way. Then, re-test with -s 32000 to test for throughput.
The next step is to look at disk throughput. Copying a large file that doesn't fit in the memory cache is more of a test of the disk drives than of anything else, as the network cannot run any faster than the disks can accept data. Grab a copy of SQLIO from Microsoft and perform some testing with that. Here is a sample command that I like to test disk write performance:
sqlio -kW -s30 -frandom -o8 -b64 -LS -Fparam.txt
Again. do this on your workstation, the various drives of physical servers, and the VM drives. This will give you a relative performance of each drive.
Your file-copy tests are a good next step, but stick with a smaller file. (The 100MB files created by SQLIO might be a good choice.) By sticking to files that are already in memory, you eliminate disk latency and can concentrate more on file transfer times. Again, test this by copying in various combinations (physical->physical, VM->phys, phys->VM, and include a workstation or two in the mix, too. Again, create a list of relative performance levels in each area.
Once you have the above data, you should get a feel for which component is actually the slowest (network, disk, file transfer). Determining the WHY is just as difficult, but you have to know which issue you have before you can dig any further...
fping <target> -t 0 -n 100 -s 32
This will send 100 packets to the target. Test this from various machines in combination on your network. Start with two physical workstations next to each other; then hit the physical server; then hit the VM server, watching the difference in performance all the way. Then, re-test with -s 32000 to test for throughput.
The next step is to look at disk throughput. Copying a large file that doesn't fit in the memory cache is more of a test of the disk drives than of anything else, as the network cannot run any faster than the disks can accept data. Grab a copy of SQLIO from Microsoft and perform some testing with that. Here is a sample command that I like to test disk write performance:
sqlio -kW -s30 -frandom -o8 -b64 -LS -Fparam.txt
Again. do this on your workstation, the various drives of physical servers, and the VM drives. This will give you a relative performance of each drive.
Your file-copy tests are a good next step, but stick with a smaller file. (The 100MB files created by SQLIO might be a good choice.) By sticking to files that are already in memory, you eliminate disk latency and can concentrate more on file transfer times. Again, test this by copying in various combinations (physical->physical, VM->phys, phys->VM, and include a workstation or two in the mix, too. Again, create a list of relative performance levels in each area.
Once you have the above data, you should get a feel for which component is actually the slowest (network, disk, file transfer). Determining the WHY is just as difficult, but you have to know which issue you have before you can dig any further...
You should really use something measuring network performance only, not involving the I/O storage subsystem, like netIO at http://www.ars.de/ars/ars.nsf/docs/netio . iperf (https://iperf.fr/) is another popular tool.
one other item to note disabling ipv6 is not a recommended nor supported option for windows server 2008 onward. The server subsystem uses ipv6 a lot.
This one is on us!
(Get your first solution completely free - no credit card required)
UNLOCK SOLUTION
Can you post the version of the driver that WAS on the machine, along with the one you updated it to? Having this detail is important so that when people refer back to this ticket in future months, they will benefit from the time we all spent working on it.
Bill Bach - Good point. While I do not recall what the 'old' versions were the latest versions that did the trick are....
FIRMWARE = Broadcom NetXtreme I and II Network Device Firmware 7.10.18
DRIVE = Broadcom Windows 64bit driver update for NetXtreme I and NetXtreme II Ethernet adapters for the 18.4.2 update
FIRMWARE = Broadcom NetXtreme I and II Network Device Firmware 7.10.18
DRIVE = Broadcom Windows 64bit driver update for NetXtreme I and NetXtreme II Ethernet adapters for the 18.4.2 update
Gain unlimited access to on-demand training courses with an Experts Exchange subscription.
Get AccessWhy Experts Exchange?
Experts Exchange always has the answer, or at the least points me in the correct direction! It is like having another employee that is extremely experienced.
Jim Murphy
Programmer at Smart IT Solutions
When asked, what has been your best career decision?
Deciding to stick with EE.
Mohamed Asif
Technical Department Head
Being involved with EE helped me to grow personally and professionally.
Carl Webster
CTP, Sr Infrastructure Consultant
Did You Know?
We've partnered with two important charities to provide clean water and computer science education to those who need it most. READ MORE
Ask ANY Question
Connect with Certified Experts to gain insight and support on specific technology challenges including:
- Troubleshooting
- Research
- Professional Opinions
Join our community and discover your potential
Experts Exchange is the only place where you can interact directly with leading experts in the technology field. Become a member today and access the collective knowledge of thousands of technology experts.
*This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.