VMware Cloud Community
Phylum
Enthusiast
Enthusiast

Help Troubleshooting Throughput Problem

I'm seeing some really poor network performance on a specific VM and I'm wondering how to best troubleshoot this.

  • Is it the NAS?
  • Is it the Host?
  • Is it the physical network?
  • Is it the virtual network?
  • Is it the VM itself?
  • Is it the destination VM?
  • Something else?

While transferring some data to an AWS EC2 Instance, I noticed it was taking an incredible amount of time to transfer a small amount of data.  Even after several minutes the box had only managed to transfer 400MB which to me (on a 25/25 connection) wasn't normal.  I decided to run through some very simple tests on the VM in question then on the destination EC2 instance.  (Just so we're clear: the destination machine is not hosted here, its an Amazon EC2 VM.  So all this data is going out over the Internet to the destination machine.)

I found a 30MB file on Softlayer's site that I've been using to build a very simple baseline:

wget --output-document=/dev/null http://speedtest.sea01.softlayer.com/speedtest/speedtest/random4000x4000.jpg

I ran through these tests from (yes I have wget for Windows)

  1. a physical desktop (Windows 7): 3M/s
  2. the NAS (via ssh): 3M/s
  3. the CentOS VM in question: 600-700K/s
  4. a Windows 7 VM on the same host as the CentOS: 3M/s
  5. the destination EC2 instance: 3M/s

Note: I realize this only tests downloads & not uploads.  I expected to see maybe high download speeds & a poor upoload speed but when the download itself was low, I figured there may be some sort of misconfiguration somewhere.  I ran the GUI upload test (http://speedtest.sea01.softlayer.com/speedtest/) on the physical and virtual Windows 7 systems, and the results were roughly 18Mbps up and 25Mbps down.

With 110GB to transfer, I'm a little worried it'll take ages at this rate!  Help!

I don't know what information might be useful at this point so I'll give you what I have.

Local VM Info
  -  CentOS release 5.8 (Final)
  -  Linux version 2.6.18-308.13.1.el5 (mockbuild@builder10.centos.org) (gcc version 4.1.2 20080704 (Red Hat 4.1.2-52)) #1 SMP Tue Aug 21 17:10:18 EDT 2012
  -  1 vCPU / 2GB RAM

Remote VM Info

  -  Linux version 2.6.18-xenU-ec2-v1.2 (root@domU-12-31-39-06-0D-E1) (gcc version 4.1.2 20070626 (Red Hat 4.1.2-13)) #2 SMP Wed Aug 19 12:57:15 UTC 2009

  -  Similar CPU & Memory configuation

Host Info
  -  Dell PowerEdge T710 Dual Quad Xeon E5504 12GB
  -  8x500GB SATA 7200RPM 32MB Cache 2xRAID5
  -  Storage via iSCSI LUNs
  -  ESXi 4.1.0 800380

Storage Info
  -  Synology DS411+ii
  -  4x3TB SATA 7200RPM 64MB Cache RAID5

Local Network Connectivity
  -  FiOS 25/25
  -  Gig connectivity all around
  -  No firewalls between host and NAS

  -  No complex network configuration on switches/routers (other than VLAN tagging)
  -  No complex VMWare network configuration (other than VLAN tagging)

0 Kudos
1 Reply
Phylum
Enthusiast
Enthusiast

For what its worth: I'm scp'ing a 60+GB tar'd.gz'pd file to Amazon and it seems capped at roughly 1MB/s but it shoudl be closer to 3MB/s.

I can't figure out if its the VM, the host, FiOS or Amazon and I'm not sure what the best testing methodology is for ruling out one or the other.

0 Kudos