-1

For a remote Windows desktop solution, accessed from Las Palmas de Gran Canaria, I want to evaluate various options. What currently comes into my mind:

  • EC2, Ireland

  • Azure Virtual Machine, Ireland

  • Azure Virtual Machine, Amsterdam

How do I best compare RDP latency? Just boot up machines and ping them?

feklee
  • 505
  • 5
  • 19

3 Answers3

1

Honestly yes, testing is going to give you the best information. Anything else will just be a guess. Spin up a machine at each site running Windows and RDP into it and see how the experience is. Try it at high color and low color (colour if you prefer) and see how it goes. All told the testing should cost you $10 at the most.

mrdenny
  • 27,174
  • 4
  • 41
  • 69
1

I would boot an instance in each location and test the RDP experience. Choose the connection with the lowest latency.

If RDP proves to be a challenge, or isn't as responsive as you wish, you can try an RDP accelerator. The one I use for my own remote needs is Ericom Blaze, which runs on port 3399 instead of RDP's 3389. There's a free trial version. Try that as well, since that can absorb the difference between the various locations you're trying to test.

ewwhite
  • 197,159
  • 92
  • 443
  • 809
  • How does the accelerator work in order to reduce latency? Do they make use of some high speed network for routing traffic? – feklee Mar 05 '13 at 12:25
  • It compresses the RDP stream and makes some other enhancements. It's all done in software. – ewwhite Mar 05 '13 at 12:53
  • 1
    RDP has compression [built in](http://www.microsoft.com/resources/documentation/windows/xp/all/proddocs/en-us/mstsc_data_compression.mspx?mfr=true), and I would not expect compression to have an effect on latency anyhow. The only way to really cut down latency, as far as I can see, is to use different routing, with less hops and/or shorter distances and/or faster switches/routers. Still, Ericom Blaze does look interesting. So thanks for the suggestion! – feklee Mar 05 '13 at 15:26
  • Ericom Blaze isn't RDP... – ewwhite Mar 05 '13 at 15:36
  • Ericom Blaze is built on top of RDP, though. That's quite obvious from the statements on Ericom's website. In boldface, they write [Ericom Blaze RDP compression and acceleration software](http://www.ericom.com/ericom_blaze.asp). I am really curious to understand what technology they use, but that would be something for another question. – feklee Mar 05 '13 at 23:31
1

I found an article by Rob Blackwell where he recommends using tcping for evaluating latency, which I prefer to comparing experience. I ran:

tcping -i 5 -n 12 %AZURE_1% 80 >azure_europe-north.txt
tcping -i 5 -n 12 %AZURE_2% 80 >azure_europe-west.txt
tcping -i 5 -n 12 %EC2% 80 >ec2_eu-west.txt

Measured average latency:

  • Azure, both locations: around 180ms (except once I got an average of 233ms for Europe North)

  • EC2, eu-west: between 80 and 90ms

Details about the tests:

  • My location: Las Palmas de Gran Canaria

  • I did the tests yesterday in the evening (Western European Time) and today in the morning.

  • I repeated each test several times.

  • Host OS for machines: Windows Server 2012

  • Client OS: Windows XP/SP3 (32bit)

  • ISP: ONO Canarias (DSL or Cable, I don't know for sure. I'm in a shared office.)

See also my corresponding thread in the MSDN forums.

feklee
  • 505
  • 5
  • 19
  • While other answers provide useful suggestions, for now, I decided to mark my own answer as accepted. The reason is that it shows how to measure (RDP) latency quantitatively (though not necessarily reliably: there may be fluctuations that I'm unaware of). – feklee Mar 10 '13 at 11:54