Latency between GCI Anchorage and VZB in NY
All, I've been working diligently to improve performance of interactive applications (Citrix, terminal) that are run by users in our office located in Anchorage, and are served by a managed Internet connection provided by GCI. Our applications reside in the Buffalo, NY area. Via MTR, I've seen no or almost-no lost packets, whereas RTT averages around 124ms and at times is as high as 328ms. I'm looking for feedback from others regarding this RTT, hopefully from customers of GCI. 328ms RTT is high as far as I'm concerned, and it seems like this could be controlled a little better. If this is status quo for GCI->VZB/MCI connections, then I'd be interested in that feedback as well. Thanks in advance Note: The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer.
On Wed, 26 May 2010, Brad Beck wrote:
I've been working diligently to improve performance of interactive applications (Citrix, terminal) that are run by users in our office located in Anchorage, and are served by a managed Internet connection provided by GCI. Our applications reside in the Buffalo, NY area.
I don't know anything about your specific case... but with an 100% optimum light-path you're looking at ~55 ms RTT between ANC-NYC (1ms RTT per 100km of fiber): <http://gc.kls2.com/cgi-bin/gc?PATH=anc-nyc%0D%0A&RANGE=&PATH-COLOR=red&PATH-UNITS=km&PATH-MINIMUM=&SPEED-GROUND=&SPEED-UNITS=kts&RANGE-STYLE=best&RANGE-COLOR=navy&MAP-STYLE=> Of course fiber very seldom goes direct path, so 124ms sounds plausable and could most likely be improved by a little bit by choosing another provider, but would that really substantially improve your interactivity problem? -- Mikael Abrahamsson email: swmike@swm.pp.se
On 26 May 2010 16:38, Mikael Abrahamsson <swmike@swm.pp.se> wrote:
On Wed, 26 May 2010, Brad Beck wrote:
I've been working diligently to improve performance of interactive
applications (Citrix, terminal) that are run by users in our office located in Anchorage, and are served by a managed Internet connection provided by GCI. Our applications reside in the Buffalo, NY area.
I don't know anything about your specific case... but with an 100% optimum light-path you're looking at ~55 ms RTT between ANC-NYC (1ms RTT per 100km of fiber):
Of course fiber very seldom goes direct path, so 124ms sounds plausable and could most likely be improved by a little bit by choosing another provider, but would that really substantially improve your interactivity problem?
-- Mikael Abrahamsson email: swmike@swm.pp.se
Brad
I'd be looking at WAN optimisers like Riverbed, Juniper, Silverpeak etc. Even if you move providers and it works better today there's no saying they won't move their paths about and you'll be back on the same boat tomorrow. -- Martin Hepworth Oxford, UK
Hi, On Wed, 2010-05-26 at 11:27 -0400, Brad Beck wrote:
All,
I've been working diligently to improve performance of interactive applications (Citrix, terminal) that are run by users in our office located in Anchorage, and are served by a managed Internet connection provided by GCI. Our applications reside in the Buffalo, NY area.
The interactivity problem probably has more to do with your Citrix setup. What specs are your Citrix server(s)? I doubt it is virtualized, but just in case, is it? I ask because it sounds more like iowait. Process Explorer (sysinternals) can be useful for listing CPU and I/O hungry applications on the Citrix server(s).
Via MTR, I've seen no or almost-no lost packets, whereas RTT averages around 124ms and at times is as high as 328ms.
I'm looking for feedback from others regarding this RTT, hopefully from customers of GCI. 328ms RTT is high as far as I'm concerned, and it seems like this could be controlled a little better.
Spikes in RTT are likely normal. Your managed internet connection is probably provided over an MPLS private network, which means that it still has to share packet queues with other customers. William
participants (4)
-
Brad Beck -
Martin Hepworth -
Mikael Abrahamsson -
William Pitcock