I rent servers from atjeu.com which are used for game servers. Up until 2 weeks ago, I have never had any issues with network latency. Then they tell me they have a problem with Global Crossing. Finaly a few days ago I get a email that it is fixed.

However I am still having problem in that datacenter. I lost every one of my gaming customers in this data center because of it. They ALL left saying it the network was horrible.

I have all my customers do tracerts and It show me that there is a problem with the network. Atjeu emailed me saying:

"the only think not normal on the tracrts you sent was on a hop on each one the times were a little high usually on the first try of those respective hops. whatever that was didnt last long and was a very short intermittent thing and appeared on all lines coming in and going out, not just glbx - there has not been any latency on the glbx line since the end of last week when a router port was changed out"

anyway, atjeu has SUPER customer service but I want your opinion on these trace routes. so here they are:

http://smart-hosting.com/atjeu/trace1.txt
http://smart-hosting.com/atjeu/trace2.txt
http://smart-hosting.com/atjeu/trace3.txt

today I pinged one of the routers that the traceroute shows high ping times and here is a ping result:


Reply from 162.42.149.13: bytes=32 time=72ms TTL=238
Reply from 162.42.149.13: bytes=32 time=96ms TTL=238
Reply from 162.42.149.13: bytes=32 time=79ms TTL=238
Reply from 162.42.149.13: bytes=32 time=98ms TTL=238
Reply from 162.42.149.13: bytes=32 time=85ms TTL=238
Reply from 162.42.149.13: bytes=32 time=77ms TTL=238
Reply from 162.42.149.13: bytes=32 time=78ms TTL=238
Reply from 162.42.149.13: bytes=32 time=76ms TTL=238
Reply from 162.42.149.13: bytes=32 time=112ms TTL=238
Reply from 162.42.149.13: bytes=32 time=77ms TTL=238
Reply from 162.42.149.13: bytes=32 time=77ms TTL=238
Reply from 162.42.149.13: bytes=32 time=266ms TTL=238
Reply from 162.42.149.13: bytes=32 time=268ms TTL=238
Reply from 162.42.149.13: bytes=32 time=251ms TTL=238
Reply from 162.42.149.13: bytes=32 time=234ms TTL=238
Reply from 162.42.149.13: bytes=32 time=162ms TTL=238
Reply from 162.42.149.13: bytes=32 time=74ms TTL=238
Reply from 162.42.149.13: bytes=32 time=74ms TTL=238
Reply from 162.42.149.13: bytes=32 time=100ms TTL=238
Reply from 162.42.149.13: bytes=32 time=78ms TTL=238
Reply from 162.42.149.13: bytes=32 time=97ms TTL=238
Reply from 162.42.149.13: bytes=32 time=96ms TTL=238

I get anywhere from 70ms ( which was normal before the problems) to 300ms to this router.

do any of you think this is normal or do you see a problem too?

Brad

Sounds like their network should be where the problem is. I would presume it is a routing error, but from the way pings sometimes goes up and down, it is more of a latency from either physical layer failure or hardware failure in their routers.

That's what I have been trying to tell them. I have sent in traceroutes and opened up tickets, but they keep telling me its normal. Guess I need to find an ISP that knows how to keep a network running smooth =/

BRad

Originally posted by tbnguyen
Sounds like their network should be where the problem is. I would presume it is a routing error, but from the way pings sometimes goes up and down, it is more of a latency from either physical layer failure or hardware failure in their routers.

If there were a physical layer failure then you wouldn't be able to ping them. But I think your right about the router hardware... either that or the router is getting some really whacky traffic, but that's pretty unlikely. Since you haven't had that many problems there, I would send them a ticket asking that either the router be replaced or you'll find a new company to deal with. That may get them working on it.

now they are saying I cannot post what they say to me in theier ticket system here at webhostingtalk.com and taht I could get sued?

Is this normal behaviour for a company?

I read all the time about problems people are having with ISP's here at webhosting talk. I think I have a right to post that kind of information.

Brad

what we said is that we do not give this customer or any of our customers permission to post our private comments verbatum from our private ticketing system to a public forum - we also said it is ok to summarize what is being said.

we also said that we would never post private comments from a customer verbatum to a public forum such at this.

in fact, here is our post "do not ever post private comments from a private ticketing system to a public forum, you could get sued big time for that from anyone - we would never do that to you - its ok to summarize what is said but not to post verbatum - we would never do that to you or any customers."

Originally posted by atjeu
what we said is that we do not give this customer or any of our customers permission to post our private comments verbatum from our private ticketing system to a public forum - we also said it is ok to summarize what is being said.

we also said that we would never post private comments from a customer verbatum to a public forum such at this.

in fact, here is our post "do not ever post private comments from a private ticketing system to a public forum, you could get sued big time for that from anyone - we would never do that to you - its ok to summarize what is said but not to post verbatum - we would never do that to you or any customers."

No you can't sue them, and furthermore, the reason your pissed is because your tech didn't have a clue what he/she was talking about. Quit whining and replace the guys router.

Brad - Find a new company.

wrong and wrong again - and we are not whining and never threatened to sue - just said it could happen - there was nothing wrong in particular with the comments brad posted - and the comments he posted were NOT in response to the pings he posted on this forum either - that was from a completely different ticket so you dont know what you are talking about.

everyone join in now... lets have typical wht bash fest... maybe we can get this thread up to 100 pages and have a record... yippie...

Heh yeah... drop it guys and keep on topic. Has the router been serviced/replaced atjeu?

there is nothing at all wrong with the router, it is in perfect shape - its (actually its 2 routers running in tandemn) triple homed and runs bgp4 so when the bgp scanner runs every minute its normal for ping times to elevate and does not indicate network latency, just router cpu load going up for a few secs - all ciscos do that at both our data centers and it does not effect ping times to individual servers etc... please dont ping our router interfaces on mulit-homed routers to test latency because its meaningless data. we prepend as paths and a lot of other things to load balance traffic which means a lot of traffic comes in one interface and goes out another which really blows a tracert but does not indicated latency either... on the other hand, brad is correct that global crossing has really been sucking the last 2 weeks and needs a good kick in the pants because of the atrocious level of service and time it has taken them to diagnose the occassional latency that has occured on the global crossing line - at one point we had them off all together and at another we had it padded down to only a meg or 2 of traffic on their line - they are suposed to be a top level tier 1 backbone but they have been dropping the ball lately and brad is definitely right about that.

Makes sense. You also have to remember Brad that on most routers, ICMP packets are not as important.

Originally posted by atjeu
there is nothing at all wrong with the router, it is in perfect shape - its (actually its 2 routers running in tandemn) triple homed and runs bgp4 so when the bgp scanner runs every minute its normal for ping times to elevate and does not indicate network latency, just router cpu load going up for a few secs - all ciscos do that at both our data centers and it does not effect ping times to individual servers etc... please dont ping our router interfaces on mulit-homed routers to test latency because its meaningless data. we prepend as paths and a lot of other things to load balance traffic which means a lot of traffic comes in one interface and goes out another which really blows a tracert but does not indicated latency either... on the other hand, brad is correct that global crossing has really been sucking the last 2 weeks and needs a good kick in the pants because of the atrocious level of service and time it has taken them to diagnose the occassional latency that has occured on the global crossing line - at one point we had them off all together and at another we had it padded down to only a meg or 2 of traffic on their line - they are suposed to be a top level tier 1 backbone but they have been dropping the ball lately and brad is definitely right about that.

If pinging or trace routing gives a false evaluation of pings, the fact remains that his game servers are having latency issues. Whether it's a hardware or network issue, it's something you guys need to fix. If it was just him having high latency, then yes it probably is a problem on his end. But if hundreds of other people are having the same problem, the problem has to be on your end.

Originally posted by dbbrock1
If pinging or trace routing gives a false evaluation of pings, the fact remains that his game servers are having latency issues.

Exactly, that is my point also

They are moving me to another datacenter. We will see how that does.

Brad

Originally posted by dbbrock1
If pinging or trace routing gives a false evaluation of pings, the fact remains that his game servers are having latency issues. Whether it's a hardware or network issue, it's something you guys need to fix. If it was just him having high latency, then yes it probably is a problem on his end. But if hundreds of other people are having the same problem, the problem has to be on your end.

that is true but you are missing the point about the routers - pinging a multi-homed router interface means nothing about latency on the network. when a cisco's cpu load goes up or it gets busy doing anything, it will respond to pings only if it has time in its clock cycle - otherwise it wont even respond at all. icmp is the absolute last priority by default in any router (unlike a server). we can turn off icmp on the ciscos all together and it will not effect or add latency to the network - then people who tried to ping the router interfaces would saying the network was down - would that be true? Not at all. that is what was in question here. if you having continual latency in pinging or tracing a server, that is a different story.

Ever seen a tracert that gets through to the end server but has an entire hop that times out? thats normal and does not mean anything is down or that there is any additional latency. it means the network is multi-homed and your packets are going in one interface and out another. it slows down your tracert but it does not add any latency for real internet use like game playing.

certainly if someones server is experiencing latency it needs to be addressed and we have over 100 hours into the global crossing issue these past 2 weeks -

the router pings and "high times" coming off a router interface are completely unrelated though, and are normal for a triple homed (and triple physical location - our network makes a triangle around the phoenix) data center.

anyone that offers in their 'linux dedicated servers' the following:

Your Choice of Linux OS - Redhat 9.0 (default), Slackware, FreeBSD etc...

worries me...

you do know theres a difference between bsd and linux dont you ?

No I'm not getting paid for this.

I have a server at Atjeu, and am very happy with it.

Sorry to hear that Bras has a hard time, but I would recommend Atjeu to anybody looking for a dedicated server.

The best customer support for any non managed server package.

If the problem is with Global crossing what do you want Atjeu to do???

:rolleyes: