Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


Optimal POP between AUS and PL
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.

Optimal POP between AUS and PL

Here is the challenge: find a location that offers best latency for both Poland and Australia.

  1. Minimum sum(PL+AUS) ICMP.
  2. Most equal distribution between the two.

Thoughts?

Comments

  • Los Angeles probably comes close

  • bdlbdl Member

    @IziD said:
    Here is the challenge: find a location that offers best latency for both Poland and Australia.

    1. Minimum sum(PL+AUS) ICMP.
    2. Most equal distribution between the two.

    Thoughts?

    Why? :)

    But practically, Australia is very connection dependent, some connections go via the US, some via Singapore... Do you have an existing service in Australia that you can test routing from?

    Thanked by 1yoursunny
  • Singapore, with slightly less than 200ms from Poland and (ideally) less than 100ms from Australia.

  • bdlbdl Member

    @Kousaka said:
    Singapore, with slightly less than 200ms from Poland and (ideally) less than 100ms from Australia.

    Except when your crappy connection (I'm looking at you, AAPT), routes you via the west coast US and/or Japan to Singapore, so you end up getting 200ms pings to .sg rather than ~95ish...

  • IziDIziD Member

    @bdl said: Do you have an existing service in Australia that you can test routing from?

    I do.

    @bdl said: routes you via the west coast US and/or Japan to Singapore, so you end up getting 200ms pings to .sg rather than ~95ish...

    Can confirm, not happy with DO in Singapore due to the above.

    @Kousaka said: and (ideally) less than 100ms from Australia.

    Unfortunately, not the case. Closer to ~160ms.

    @dahartigan said: Los Angeles probably comes close

    Interesting pick, I am seeing ~190-200ms to LA.

  • Chicago or Dallas

    Thanked by 1bdl
  • IziDIziD Member

    @corbpie said: Chicago or Dallas

    Courtesy of Psychz:

    • Dallas, ~230ms average
    • Chicago, ~260ms
  • It all depends on routing. I live in Poland and:
    RackNerd New York = ~100ms (around 6632km/4121mi)
    Hetzner Nuremberg = ~50ms (around 533km/331mi)
    Oracle Frankfurt = ~28ms (around 646km/400mi)
    RackNerd AMS = ~24ms (around 838km/520mi)

    Even in Europe routing can be mess.

    I don't have experience with AUS. But based on pure geography Singapore should be the best location. But it all depends on routing. If you wont find anything good in SG then I think you should check Dallas.

  • New York will be the best for PL (~ 95-100ms twelve99)

  • bdlbdl Member
    edited August 2021

    @AXYZE said:
    It all depends on routing. I live in Poland and:
    RackNerd New York = ~100ms (around 6632km/4121mi)
    Hetzner Nuremberg = ~50ms (around 533km/331mi)
    Oracle Frankfurt = ~28ms (around 646km/400mi)
    RackNerd AMS = ~24ms (around 838km/520mi)

    Even in Europe routing can be mess.

    I don't have experience with AUS. But based on pure geography Singapore should be the best location. But it all depends on routing.

    Unfortunately due to historical shenanigans by Australia's monopolistic incumbent telco, Singapore should be the best location but routing often means that the US is better... I would suggest that OP tests potential VPS providers between their sites with relevant looking glasses..

    Based on my VPSdime (Dallas) and Hosthatch (Chicago) service, I would recommend Chicago over Dallas, but again that's just due to my current providers' network mixes...

    ps: for those that are interested, this book provides some good examples the kind of shenanigans Australia has had to deal with https://www.goodreads.com/book/show/12177695-wired-brown-land

    Thanked by 1hosthatch
  • For the same two endpoints...

    My connection in Australia routes via Perth and SG (Equinix) and is about 250 msec to Warsaw.

    My connection in Warsaw routes via New York (Telstra) and is around 310 msec to Australia.

  • jenkkijenkki Member
    edited August 2021

    Moscow, Russia >32Ms and >280Ms

  • NeoonNeoon Community Contributor, Veteran
    edited August 2021

    Well, Singapore, from Poland 258ms here.
    Frankfurt has 249ms via Singapore.

  • DwayneDwayne Member
    edited August 2021

    Minimum sum(PL+AUS) ICMP: Singapore (<300ms)

    Most equal distribution between the two: Los Angeles (~155-160ms each)

    Tested from Hosthatch VPS in Warsaw and Sydney to Webhosting24 Singapore and Hosthatch LAX

  • yoursunnyyoursunny Member, IPv6 Advocate

    I did some tests with https://ping.sx , using these destinations:

    • waw.lg.webhorizon.in
    • lg.au.webhosting24.com

    To minimize sum of latency to Poland and Australia: you should select either Sydney or Singapore.

    To minimize difference of latency to Poland and Australia: you should select USA West Coast.

    raw data in TSV format

    Thanked by 1cybertech
  • @yoursunny said:
    I did some tests with https://ping.sx , using these destinations:

    • waw.lg.webhorizon.in
    • lg.au.webhosting24.com

    To minimize sum of latency to Poland and Australia: you should select either Sydney or Singapore.

    To minimize difference of latency to Poland and Australia: you should select USA West Coast.

    raw data in TSV format

    which is better, lowest sum of latency or least difference between both ends?

  • I'd lean towards LAX, you have far more options there, it's cheaper and it's more likely to deliver consistent results.

    Singapore is a decent option only in theory, you have a high chance of encountering crappy routes, both from Poland and Australia (via Japan or West Coast).

  • bdlbdl Member

    @salakis said:
    I'd lean towards LAX, you have far more options there, it's cheaper and it's more likely to deliver consistent results.

    Singapore is a decent option only in theory, you have a high chance of encountering crappy routes, both from Poland and Australia (via Japan or West Coast).

    this. 100%

  • yoursunnyyoursunny Member, IPv6 Advocate

    @cybertech said:

    @yoursunny said:
    I did some tests with https://ping.sx , using these destinations:

    • waw.lg.webhorizon.in
    • lg.au.webhosting24.com

    To minimize sum of latency to Poland and Australia: you should select either Sydney or Singapore.

    To minimize difference of latency to Poland and Australia: you should select USA West Coast.

    raw data in TSV format

    which is better, lowest sum of latency or least difference between both ends?

    Latency sum and difference are objective, and can be measured with tools then aggregated in a spreadsheet.
    Better is subjective, and highly dependent on the application:

    • If you are deploying a TURN server to relay telephone traffic, you want lower sum for lower voice latency.
    • If you are deploying a Jeopardy game, you want lower difference for fairness.

    The push-up delivery network has POPs in Los Angeles, Dallas, Buffalo, Miami, Roubaix, Munich, Warsaw, Singapore, Tokyo.
    Viewer would connect to four nearest locations, test their speeds, then keep the fastest connection.

  • NeoonNeoon Community Contributor, Veteran
    edited August 2021

    @salakis said:
    Singapore is a decent option only in theory, you have a high chance of encountering crappy routes, both from Poland and Australia (via Japan or West Coast).

    Can't agree with that.

    95% of the Time, I does stick fine to 154ms from Frankfurt.
    However: My Network does reroute a few times per day, based on latency, so could be that its swapping between my Frankfurt Servers and others to jump down to Singapore.

    edit:

    But when I look at Tokyo and Hong Kong, it seems rather unstable.
    Here For comparison:


    edit: 2

    But Sydney looks stable so

  • Hey, I haven't been around for some time and missed many updates to this thread.

    Kudos to @Neoon and @yoursunny (we all need to do more pushups!) for their comprehensive replies.

    Ended-up using SG POP.

    Thanked by 1commercial
  • @AXYZE said:
    Hetzner Nuremberg = ~50ms (around 533km/331mi)
    Oracle Frankfurt = ~28ms (around 646km/400mi)

    Damn, man, you have some incredibe crappy routing from Poland. I can't imagine to have more than 15 - 22ms max from Ljubljana to any of those locations.

  • most providers in australia do route via perth to singapore however if in LA the cheaper providers will route you via hong kong and the more expensive providers will route you to LA with around a 140-160ms

  • MumblyMumbly Member
    edited November 2022

    @AXYZE said:
    Hetzner Nuremberg = ~50ms (around 533km/331mi)
    Oracle Frankfurt = ~28ms (around 646km/400mi)
    RackNerd AMS = ~24ms (around 838km/520mi)

    Even in Europe routing can be mess.

    As we discused in one of the previous threads already, it's really messed up how you get better latency from Poland to the Netherlands than to the Germany. I mean 50ms... this latency is half way to the other continent (US, east coast) and not normal for the neighbour country.
    I checked from several random locations and I wouldn't say that's a norm in the Europe. At least not so drastic. You should blame your ISP(s) peering not European routing.

    I coudn't resist urge to check if also from my side (similar / in some cases larger distances with Austria in between).

    Oracle-Frankfurt - Ljubljana

    Pinging 144.24.173.xxx with 32 bytes of data:
    Reply from 144.24.173.xxx: bytes=32 time=16ms TTL=55
    Reply from 144.24.173.xxx: bytes=32 time=16ms TTL=55
    Reply from 144.24.173.xxx: bytes=32 time=16ms TTL=55
    Reply from 144.24.173.xxx: bytes=32 time=17ms TTL=55

    Ping statistics for 144.24.173.xxx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 16ms, Maximum = 17ms, Average = 16ms

    Hetzner-Nuremberg - Ljubljana

    Pinging 49.12.228.xx with 32 bytes of data:
    Reply from 49.12.228.xx: bytes=32 time=22ms TTL=53
    Reply from 49.12.228.xx: bytes=32 time=22ms TTL=53
    Reply from 49.12.228.xx: bytes=32 time=22ms TTL=53
    Reply from 49.12.228.xx: bytes=32 time=22ms TTL=53

    Ping statistics for 49.12.228.xx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 22ms, Maximum = 22ms, Average = 22ms

    Inceptionhosting DE VPS

    Pinging 193.107.xx.xxx with 32 bytes of data:
    Reply from 193.107.xx.xxx: bytes=32 time=20ms TTL=54
    Reply from 193.107.xx.xxx: bytes=32 time=20ms TTL=54
    Reply from 193.107.xx.xxx: bytes=32 time=20ms TTL=54
    Reply from 193.107.xx.xxx: bytes=32 time=20ms TTL=54

    Ping statistics for 193.107.xx.xxx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 20ms, Maximum = 20ms, Average = 20ms

    But absolute winner is my webhosting24 DE VPS

    Pinging 194.39.xxx.xx with 32 bytes of data:
    Reply from 194.39.xxx.xx: bytes=32 time=11ms TTL=53
    Reply from 194.39.xxx.xx: bytes=32 time=11ms TTL=53
    Reply from 194.39.xxx.xx: bytes=32 time=11ms TTL=53
    Reply from 194.39.xxx.xx: bytes=32 time=11ms TTL=53

    Ping statistics for 194.39.xxx.xx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 11ms, Maximum = 11ms, Average = 11ms

    To the Netherlands is average around 28 - 32ms, which I would say it's normal from my place (Air line: 614.77 mi / 989.38 km)

  • AXYZEAXYZE Member
    edited November 2022

    @Mumbly said:

    @AXYZE said:
    Hetzner Nuremberg = ~50ms (around 533km/331mi)
    Oracle Frankfurt = ~28ms (around 646km/400mi)
    RackNerd AMS = ~24ms (around 838km/520mi)

    Even in Europe routing can be mess.

    As we discused in one of the previous threads already, it's really messed up how you get better latency from Poland to the Netherlands than to the Germany. I mean 50ms... this latency is half way to the other continent (US, east coast) and not normal for the neighbour country.
    I checked from several random locations and I wouldn't say that's a norm in the Europe. At least not so drastic. You should blame your ISP(s) peering not European routing.

    I coudn't resist urge to check if also from my side (similar / in some cases larger distances with Austria in between).

    Oracle-Frankfurt - Ljubljana

    Pinging 144.24.173.xxx with 32 bytes of data:
    Reply from 144.24.173.xxx: bytes=32 time=16ms TTL=55
    Reply from 144.24.173.xxx: bytes=32 time=16ms TTL=55
    Reply from 144.24.173.xxx: bytes=32 time=16ms TTL=55
    Reply from 144.24.173.xxx: bytes=32 time=17ms TTL=55

    Ping statistics for 144.24.173.xxx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 16ms, Maximum = 17ms, Average = 16ms

    Hetzner-Nuremberg - Ljubljana

    Pinging 49.12.228.xx with 32 bytes of data:
    Reply from 49.12.228.xx: bytes=32 time=22ms TTL=53
    Reply from 49.12.228.xx: bytes=32 time=22ms TTL=53
    Reply from 49.12.228.xx: bytes=32 time=22ms TTL=53
    Reply from 49.12.228.xx: bytes=32 time=22ms TTL=53

    Ping statistics for 49.12.228.xx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 22ms, Maximum = 22ms, Average = 22ms

    Inceptionhosting DE VPS

    Pinging 193.107.xx.xxx with 32 bytes of data:
    Reply from 193.107.xx.xxx: bytes=32 time=20ms TTL=54
    Reply from 193.107.xx.xxx: bytes=32 time=20ms TTL=54
    Reply from 193.107.xx.xxx: bytes=32 time=20ms TTL=54
    Reply from 193.107.xx.xxx: bytes=32 time=20ms TTL=54

    Ping statistics for 193.107.xx.xxx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 20ms, Maximum = 20ms, Average = 20ms

    But absolute winner is my webhosting24 DE VPS

    Pinging 194.39.xxx.xx with 32 bytes of data:
    Reply from 194.39.xxx.xx: bytes=32 time=11ms TTL=53
    Reply from 194.39.xxx.xx: bytes=32 time=11ms TTL=53
    Reply from 194.39.xxx.xx: bytes=32 time=11ms TTL=53
    Reply from 194.39.xxx.xx: bytes=32 time=11ms TTL=53

    Ping statistics for 194.39.xxx.xx:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 11ms, Maximum = 11ms, Average = 11ms

    To the Netherlands is average around 28 - 32ms, which I would say it's normal from my place (Air line: 614.77 mi / 989.38 km)

    Bro, this thread any my reply is from more than year ago xD

    "I checked from several random locations and I wouldn't say that's a norm in the Europe."
    You are checking from datacenter IPs with direct peering to another datacenter IPs.

    Residential peering is completely different.

    Orange needs direct connection.
    Deutsche Telecom needs direct connection (DTAG).
    China needs direct connection (CN2/connection with CM/CU).
    etc. etc.

    Otherwise routing & latency is a mess. Thats why all German providers are making huge point of having DTAG.

    It's a norm in Europe or I should say across the world.
    Indians also can have bad experience with peering to Philippines, Singapore. Its close, ping should be very low, but not with every provider.

    Only USA maybe better (I honestly dont know how Comcast etc. work there in terms of routing)

    PHP-Friends year ago got DTAG. Its such big deal that they announced it ;p
    "Already in the summer we started negotiations with Deutsche Telekom, since a direct connection to them was repeatedly mentioned by customers as a wish, but unfortunately DTAG does not peer at DE-CIX (to which we have been connected since March). Since 03.12.2021 our direct connection to DTAG is now active and ensures even better latencies and stability. And not only to DTAG itself: We also actively route some selected networks such as T-Mobile Austria via DTAG. As a matter of principle, we always make our routing decisions based on the best route and not on cost, since our goal is always the best possible traffic mix. As a rule, the best way to reach direct customers and partners of Deutsche Telekom is via Telekom itself, which is why we prefer this route in such cases. By connecting to Deutsche Telekom, we are further improving our network quality and meeting a popular cust omer request."

  • MumblyMumbly Member
    edited November 2022

    @AXYZE said: Bro, this thread any my reply is from more than year ago xD

    Ah, crap... I overlooked this. I am sorry :)

    @AXYZE said: You are checking from datacenter IPs with direct peering to another datacenter IPs.

    No, I did not.
    I checked it from my home connection. Also examples above are from my ISP to the locations you mentioned. Nevertheless, it seems like your ISP have extremely terrible connectivity to the neighbour Germany - as you posted. 50ms? wtf. That's not a norm here (I don't think there's a single part of the EU I would have 50ms to) neither relevant to compare connectivity between neigbour EU countries with connectivity between India and Philippines.

Sign In or Register to comment.