Email from Hopone about lag

[HateTheFake]

Ive have some horrid lag.

PING Current:170 ms Average:240 ms Low:40 ms High:13120 ms S2C: 5.8% C2S: 0.7%

I didnt start having this lag til like 4 weeks ago, i also did a speed test and got 2400 mbps with a 430 upload speed. Just seems odd this is happening my normal high used to be 300 or 400 ms with a .1 s2c c2s.

I used http://www.speakeasy.net/speedtest/ for the speed test and even doubled my current package to 3000 mbps, same lag same problems

[Money]

way to waste money trying to fix something u have no control over!

[gdugwyler]

Here's a follow-up email from HopOne. I first wrote that I realized any tracert done to the TW server will find that HopOne gives massive packetloss (it's unpingable as he said) but asked if when making the recent router replacement any config was changed, or if changes were made to throttle or limit UDP packets. If you read the first email, it's in complete opposition to what he originally said. (First: 'this is due to core2.dca2 router being down'; second: 'There was no "router going down"') Nevertheless:


----

"Greetings,

My apologies for the tardy response - I was out of the office and
offline until today.

There was no "router going down" and no "router replacement" (such a
thing is unheard of: a router should never, ever be replaced - if so
is needed, then something is seriously wrong with the network in
question, as proper routers such as Cisco GSRs are built to last
many decades and do not need to be ever shut down, save for an IOS
update - any new hardware can be added or removed or changed live
while the router is running, only shut down is ever for IOS (OS)
update/reload). We did update IOS on most of the routers on our
network, but that was just a routine upgrade, and caused no
noticeable impact for any customers or end users as it was done one
router at a time and only some routes would be different then for a
few minutes, with no real performance or other noticeable impact.

In terms of UDP packets, we have a lot of UDP packets on the network
- as we are one of the most popular networks for game servers that
rely heavily on UDP (we are usually chosen for game servers due to
our superior network performance when compared to others) - and we do
not restrict or otherwise control UDP packet flow in any way.

We have many 10,000s (if not even 100,000 by now - I haven't checked
the counts recently) performance-sensitive game servers at our DCA2
data centre alone. If there were any UDP related problems, we'd be
hearing from many 1000s of cutsomers - yet as we are not.

I am 100% sure that this is an issue isolated only to your server.
You may want to check your port duplex settings and your NIC port
logs to ensure that there is no duplex mismatch, duplex renogiations
or such issues. NIC misconfig/duplex mismatch can often cause
intermittent and hard to trace down performance issues.
I am not saying this is the issue here, but if you haven't checked
into that, you should.

Sorry that I can't be of more help, I can just say that 100% surely
there is not and has not been an issue on our side. We have many
10,000s game servers on our network, not to mention government
institutions, financial institutions, etc. - so even upon any
slightest network issue we are immediately aware.

Best Regards,
Haralds Jass"

[NaiLed]

yeah i am starting to get this lag too, from hopone thingers.

[Doc Flabby]

I'm inclined to think its TW's Server. As EG is unaffected and they use the same routers as us...They run on a different server however.

[gdugwyler]

A reliable source believes it's caused by a router "close to our server" -- closer than HopOne at any rate. It looks like we're on Superb.net if you do a whois and look at the extended info. These people might subcontract with HopOne or the like, if you look at the IP ranges ... not sure if this has any relevance. The problem is being further isolated, however.

[Doc Flabby]

A bit of a tip for everyone

This page shows the status of the links to tw's datacenter.

TW is based in this Network:

DCA2 CORE & DIST

http://nsssc.superb.net/status/corenet.php

[gdugwyler]

Annnd -- ah, what do you know! CORE1.IAD1 in the DCA2 CORE & DIST network presently has 700ms ping and 44% packetloss. Sounds familiar.

[Doc Flabby]

Ya and TW is laggy. Its a router inside Superb Servers' datacenter...like your post said.

[Stormwick]

i'm not lagging.

[D1st0rt]

i'm not lagging.

congratulations

[gdugwyler]

IAD1 - network presence for peering with major networks/backbones, ISPs, cable and DSL providers; also for transit sales and direct commercial transit connections; located at Equinix in Ashburn, Virginia


I wrote the hostmaster an email asking if he/she has recognized any issues with CORE1.IAD1, and if there is a timetable as to when it may be fixed. Will post any response here.

Edit: Just refreshed the page, and CORE2.DCA2 also showed "DOWN" for a second. Something fishy is going on here.

If anyone is die-hard about getting this figured out they have a list to which you can subscribe for manual network updates. I'm just that nerd.

[Thorkell]

Way to go dug! Your my new favorite person.. well, at least top 10 anyways. :groovy: