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.
Best latency between EU and NA
I am trying to figure out what is the best latency achievable between any two datacenters in Europe and North America (East Coast). So far, the best I've found is 70ms between Rapidswitch (Maidenhead, UK) and Choopa (Piscataway, NJ, US). Any better combination ?
Comments
Choopa will most likely be the best. Not sure of the Europe side
For comparison... if it was a straight shot from Maidenhead to Piscataway, the minimum theoretical latency would be about 18ms, according to Wolfram Alpha.
@netadmin I would pick the same combination.
@NickM Actually this would be 26.1ms in fiber (18ms, more precisely 18.6ms, would be in vacuum) ;-). Here is a mtr from Maidenhead to Piscataway.
1. 1.111.255.149.in-addr.arpa 0.0% 46 0.8 4.9 0.6 69.9 11.6
2. 87.117.212.41 0.0% 46 0.4 0.9 0.3 9.3 1.6
3. 610.core1.hex.as20860.net 0.0% 46 24.4 31.3 1.3 318.2 55.2
4. TenGigabitEthernet7-1.ar4.LON3.gblx.net 0.0% 46 1.4 5.7 1.4 105.4 17.9
5. xe-3-0-6.ar2.nyc3.us.nlayer.net 0.0% 46 73.7 72.2 70.3 84.4 2.3
6. as20473.ae7.ar2.nyc3.us.nlayer.net 0.0% 46 73.8 74.2 70.0 100.8 5.7
7. ethernet1-2-1-c2-20-b2-1.pnj1.choopa.net 0.0% 46 72.1 72.9 71.5 83.9 2.8
8. 108.61.51.245 0.0% 46 73.8 74.1 72.6 90.7 3.8
9. 108.61.68.* 0.0% 45 70.5 70.4 69.7 81.6 1.8
Given the best conditions and the lowest number of hops (cant go below 7 I think, certainly not below 5), it should be about 50 ms theoretically.
@Maounique number of hops means nothing except transparency of transit providers. With some tinet you will see one hop between (central) Europe and various US locations which is imho bad business practise. Example:
from EU to E-Solutions, Tampa
8 18 ms 15 ms 15 ms xe-0-3-0.vie20.ip4.tinet.net [77.67.75.201]
9 143 ms 159 ms 142 ms xe-10-2-0.mia10.ip4.tinet.net [141.136.109.117]
from EU to Internap, Dallas
8 15 ms 22 ms 14 ms xe-0-3-0.vie20.ip4.tinet.net [77.67.75.201]
9 151 ms 154 ms 152 ms xe-1-0-0.dal40.ip4.tinet.net [89.149.180.25]
70ms? Count me impressed. That's the lowest I've seen for a UK to US route end to end. Much slower/higher latency than theoretical speed of light, as always.
Get about 81ms to your UK IP from Buffalo.
90ms " from Chicago.
Good connectivity in that facility in Maidenhead.
@Spirit,
Right on. Tinet does that big silent hop regularly. Telia does it too in places.
Around 70ms is the best I've seen. Global Crossing between the NYC area and London is good.
hops 5-6 from traceroute
5 te6-2.ar7.NYC1.gblx.net (69.31.94.73) 0.972 ms 0.956 ms 0.946 ms
6 IOMART.TenGigabitEthernet7-1.ar4.LON3.gblx.net (64.211.1.190) 70.230 ms 69.958 ms 69.930 ms
Link between Sirius II DC (AllSimple vps) to Piscataway, NJ
Also link between Sirius II DC to webrulon/alienvps NYC over cogent isn't that bad
That has nothing to do with business practises (except saving money). In reality it's called MPLS network and this behaviour will become more and more popular.
I was thinking of real hops, not if they are masked or not.
It is your computer, the bgp router, the main router at one end of sub-atlantic fiber, the one a the other end, the bgp router in US and the computer there.
I dont see any way to shorten that except if you do the measuring end-to-end at the cable. In reality, it will be at least 2 times that for most regular colo/vps providers.