New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Ok ok, this is probably cheating because it is a leased line, but it is in my home
[email protected] :: ~# ping ns1.edis.at
PING ns1.edis.at (91.227.204.227) 56(84) bytes of data.
64 bytes from ns1.edis.at (91.227.204.227): icmp_req=1 ttl=63 time=0.585 ms
PS.:
[email protected] :: ~# wget at.edis.at/100mb.bin
--2012-04-11 21:18:07-- http://at.edis.at/100mb.bin
Auflösen des Hostnamen at.edis.at... 149.154.152.57
Verbindungsaufbau zu at.edis.at|149.154.152.57|:80... verbunden.
HTTP-Anforderung gesendet, warte auf Antwort... 200 OK
Länge: 100000000 (95M) [application/octet-stream]
In »100mb.bin« speichern.
100%[===================================================================================================================================================================================================>] 100.000.000 20,4M/s in 4,4s
2012-04-11 21:18:12 (21,9 MB/s) - »100mb.bin« gespeichert [100000000/100000000]
@francisco Why did Egihosting drop Highwinds?
I get around 47 ms pings to my TinyVZ from my home in NJ through Cogent.
I don't know, I'm assuming the drops they were getting every few weeks pissed them off to the point of termination =\
I know I reported a good half dozen random routing drop outs with highwinds.
Cogent has been working to improve their pings/speeds. I always liked working with their support, really helpful people.
Francisco
Yep, Cogent support is good, better than level3 for example...
From Central California to Amerinoc in San Diego, about 235 mi/380 km:
BuyVM in San Jose, pretty much the same distance, a close second:
It should be 1000 actually, in reality 300 - so that speed is "bad"
I pay 59eur with VAT for it per month.
Lucky man!
Well, it is not like someone would normally get that - Usualy it is "just" 100/10Mbit for 59EUR.
But when you also commit a lot of transit at UPC and know a few people.... you know, networking is important
Touche!
=-=-=-=-=
Kansas City, DataShack
=-=-=-=-=
64 bytes from localhost.localdomain (127.0.0.1): icmp_seq=4 ttl=64 time=0.030 ms
64 bytes from localhost.localdomain (127.0.0.1): icmp_seq=5 ttl=64 time=0.019 ms
From my UK easevps vps.
Approximate round trip times in milli-seconds:
Minimum = 27ms, Maximum = 52ms, Average = 33ms
@francisco Huh. And here I thought Highwinds offered quality transit... I guess I was wrong.
My ISP's routing is all screwed up right now so these results are equally screwed up:
From my home to SD's router:
Approximate round trip times in milli-seconds:
Minimum = 10ms, Maximum = 32ms, Average = 20ms
If I do a traceroute to the router IP it's 9 hops, if I do a traceroute to one of my node IPs it's 15 hops (my ISP routes it from my home, to Tampa, to Miami, back to Tampa). Working on getting the routes fixed.
Yup, Highwinds is crap. Much crappier than budget carriers like HE.
@ramnet Hmm... I'll take your word on that.
Regarding HE... my ISP routes a majority of their traffic over HE and haven't had any problems with latency or downtime at all. I'm getting routed over cogent to my TinyVZ though, which is strange because my ISP barely pushes anything over Cogent.
@flam316
HE network quality varies widely. New York and FMT/SJC areas are particularly bad parts of HE's network. As long as your traffic doesn't pass through one of their routers there, or hands off to one of HE's transit carriers relatively quickly (GBLX or Telia), their network is quite good.
Cogent has gotten quite a bit better lately actually. Our Kansas City location temporarily dropped Cogent and went HE-only for about a month last year while IP transit contracts were re-negotiated, and you wouldn't believe how many complaints of network issues we got (particularly from Asia and to a lesser extent Europe). As soon as Cogent was put back into the mix all those complaints disappeared.
We primarily use HE for reaching networks directly connected to HE as well as for cheap access to their transit carriers as a Tier 2 network (since Cogent is a Tier 1 it tends to have dramatically different routing than HE as a Tier 2 backed with solid GLBX and Telia transit, so they compliment each other quite nicely), as well as access to IPv6 routes since Cogent's IPv6 route coverage is severely lacking.
@ramnet A lot of things I wasn't expecting in your response...
All of my ISPs traffic gets routed through NY because I'm about 30 miles away from Manhattan. It's interesting that you say that NY and FMT/SJC are there two weak spots... I don't notice it. It seems like they've picked up a lot more fiber routes recently (as you can see by their network map); could that have made a difference?
It's also interesting how people complained of network issues from Asia when Cogent was dropped because HE has more PoPs in Asia and HE peers with China Telecom, Unicom, and Railway as well as many other Asian peers. Cogent doesn't peer with China Telecom, Unicom or Railway.
Cogent also has the third largest IPv6 network today with 640 IPv6 peers. Of course, it's not even close to HE, but that's a large IPv6 network. Even with this large of a network, their IPv6 route coverage is still lacking?
[root@serverwithattitude ~] ping ping.symantec.com
64 bytes from 127.0.0.1: icmp_seq=1 ttl=64 time=HAHA ms
64 bytes from 127.0.0.1: icmp_seq=2 ttl=64 time=YOUWISH ms
64 bytes from 127.0.0.1: icmp_seq=3 ttl=64 time=GOTSPD? ms
64 bytes from 127.0.0.1: icmp_seq=4 ttl=64 time=PWN3D! ms
64 bytes from 127.0.0.1: icmp_seq=5 ttl=64 time=GIVEUP! ms
It is well known that HE NY used to be dropping packets on a regular basis.
It does appear they have improved their network a bit - particularly improving routing in the south-central which will have provided an alternative/better path vs the northern route via NY (thus the load on HE's router in NY will have dropped considerably from a lot of traffic from the south east and south america taking a southern-central route instead of a northern route to get to places like chicago/denver/seattle/etc).
It is well known that these peering ports between HE and Asian providers (particularly to China Telcom/Unicom) are overloaded and have long needed increasing in capacity. There is routinely heavy packet loss at peak times from the Asian providers you mention to such an extent that the network is nearly unusable. This is why those routes come to us over a carrier such as Cogent that don't have this overloaded peering problem.
I'm sure Fran will confirm this too - as I recall it was a large part of why Frantech moved out of HE Fremont - HE's network has lots of issues to Asia and was a large part of their decision to move so they could pick up nLayer at Coresite, which has all those same peerings but doesn't have their network ports overloaded, or in our case via Cogent which also has all those same peerings but isn't overloaded either.
From my home in Norway to my EDIS KVM VPS in Sweden:
rtt min/avg/max/mdev = 22.463/25.467/29.175/1.786 ms
From work it's even a few ms better. I'm very satisfied :-)