Howdy, Stranger!

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


How to Fix Network Bottleneck -- Slow Throughput on High Latency VPS
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.

How to Fix Network Bottleneck -- Slow Throughput on High Latency VPS

cirrus_cloudcirrus_cloud Member
edited March 2018 in Help

Machine: time4vps storage VPS

OS: Centos 7 (trying out the owncloud template)

Port: 100 mbps

Latency: like 130ms from my location

Home internet: 80mbps down, ~93 up

At home: I'm getting like 25 KB/s at times downloading via SSH (WinSCP) or Apache. It'll max at around 3 MB/s if I'm lucky. I've been trying various times throughout the past couple days getting these poor speeds. I don't expect to get 10 MB/s since the latency is higher, but I was wondering why it's so low. I don't think it's Time4VPS because I've hit the storage server with 3 different VPSes that all have 110-130ms latency and they can all hit 5-10 MB/s, although it seems like they can get throttled, too, maybe due to the latency. snipped in Miami can hit 10 MB/s and backs off to like 3-5 MB/s and then builds back up again.

Nethogs seems to show the Time4VPS storage node is trying to push a lot more data (like 4000+ KB/s) than I'm getting (like 2 MB/s) at home. Does that mean the VPS is trying to push the data but the TCP connection is being throttled? I know it's a single high latency connection, but why can I get way better speeds on VPSes with the same latency?

Halp

Comments

  • trewqtrewq Administrator, Patron Provider

    It may just be the route it's taking is congested. Can you post traceroutes for both directions? Remember to remove any private IPs.

  • cirrus_cloudcirrus_cloud Member
    edited January 2018

    home -> storage Time4VPS

    beginning is just Verizon

    6 16 ms 16 ms 20 ms 0.et-5-3-0.BR1.IAD8.ALTER.NET [140.222.239.79] 7 26 ms 25 ms 26 ms ae-4.r06.asbnva02.us.bb.gin.ntt.net [129.250.8.209] 8 24 ms 25 ms 24 ms ae-2.r23.asbnva02.us.bb.gin.ntt.net [129.250.2.132] 9 111 ms 115 ms 135 ms ae-2.r25.amstnl02.nl.bb.gin.ntt.net [129.250.6.163] 10 107 ms 119 ms 113 ms ae-3.r24.amstnl02.nl.bb.gin.ntt.net [129.250.4.68] 11 157 ms 143 ms 136 ms ae-2.r01.stocse01.se.bb.gin.ntt.net [129.250.7.13] 12 145 ms 143 ms 144 ms 83.231.187.190 13 150 ms 144 ms 145 ms 195.122.0.150 14 * * * Request timed out. 15 * * * Request timed out. 16 162 ms 162 ms 159 ms s2-a8-lt1.storage.serveriai.lt [31.14.177.114]

    17 is the last step, which is my VPS.

  • cirrus_cloudcirrus_cloud Member
    edited January 2018

    time4vps -> home

    3 195.13.220.73 (195.13.220.73) 0.731 ms 0.704 ms 0.787 ms 4 ix-xe-1-0-0-0.thar1.W1T-Warsaw.as6453.net (80.231.124.77) 10.604 ms 10.581 ms 10.539 ms 5 if-ae-21-2.tcore1.AV2-Amsterdam.as6453.net (195.219.188.27) 116.609 ms 116.588 ms 116.545 ms 6 if-ae-2-2.tcore2.AV2-Amsterdam.as6453.net (195.219.194.6) 106.578 ms 106.255 ms 106.156 ms 7 if-ae-14-2.tcore2.L78-London.as6453.net (80.231.131.160) 108.606 ms 109.941 ms 109.853 ms 8 if-ae-4-2.tcore2.N0V-New-York.as6453.net (80.231.131.158) 108.847 ms if-ae-15-2.tcore2.LDN-London.as6453.net (80.231.131.118) 110.109 ms if-ae-4-2.tcore2.N0V-New-York.as6453.net (80.231.131.158) 109.934 ms 9 if-ae-2-2.tcore1.N0V-New-York.as6453.net (216.6.90.21) 110.251 ms if-ae-32-2.tcore2.NTO-New-York.as6453.net (63.243.216.22) 113.870 ms 114.048 ms 10 if-ae-7-2.tcore1.NTO-New-York.as6453.net (63.243.128.141) 108.833 ms if-ae-8-2.tcore1.NTO-New-York.as6453.net (63.243.128.69) 109.336 ms if-ae-7-2.tcore1.NTO-New-York.as6453.net (63.243.128.141) 109.216 ms 11 0.ae20.BR2.NYC4.ALTER.NET (204.255.168.173) 136.079 ms 126.181 ms 209.58.26.94 (209.58.26.94) 121.602 ms

    end is just Verizon.

  • My bet is the port may be oversold.

    Wasn't able to actually see what the port is, but clearly most of these servers are used for backups, so that's definitely going to be using up the port.

    Still not terrible value for the pricing of the storage, I suppose.

  • cirrus_cloudcirrus_cloud Member
    edited January 2018

    I wanted to see if it was the OS, and I installed Ubuntu 16 again instead of Centos 7. I'm seeing 200-500 KB/s again downloading. I can get 5-10 MB/s upload, which I was getting before with Centos 7.

  • Oversold port, not something you can fix yourself. You may her some where with a ticket you may not.

  • @Crandolph said:
    My bet is the port may be oversold.

    Wasn't able to actually see what the port is, but clearly most of these servers are used for backups, so that's definitely going to be using up the port.

    Still not terrible value for the pricing of the storage, I suppose.

    I've seen really good speeds downloading, though. I'm just confused why I can download at great speeds (5-10 MB/s) from 3 different VPSes (Miami, Atlanta, and Montreal) in the US / Canada with the same exact latency but not from home.

  • @cirrus_cloud said:
    I wanted to see if it was the OS, and I installed Ubuntu 16 again instead of Centos 7. I'm seeing 200-500 KB/s again downloading. I can get 5-10 MB/s upload, which I was getting before with Centos 7.

    This post hurts.

  • BBR is the best.

  • cirrus_cloudcirrus_cloud Member
    edited January 2018

    @mksh said:

    @cirrus_cloud said:
    I wanted to see if it was the OS, and I installed Ubuntu 16 again instead of Centos 7. I'm seeing 200-500 KB/s again downloading. I can get 5-10 MB/s upload, which I was getting before with Centos 7.

    This post hurts.

    And I can get 10 MB/s from Miami.

  • @a86913179 said:
    BBR is the best.

    ?

  • @cirrus_cloud said:

    @mksh said:

    @cirrus_cloud said:
    I wanted to see if it was the OS, and I installed Ubuntu 16 again instead of Centos 7. I'm seeing 200-500 KB/s again downloading. I can get 5-10 MB/s upload, which I was getting before with Centos 7.

    This post hurts.

    And I can get 10 MB/s from Miami.

    Sure Ubuntu has always been great for connections to Miami.

  • @AuroraZ said:
    Oversold port, not something you can fix yourself. You may her some where with a ticket you may not.

    What was the last part of your post?

  • bsdguybsdguy Member
    edited January 2018

    @mksh said:

    @cirrus_cloud said:

    @mksh said:

    @cirrus_cloud said:
    I wanted to see if it was the OS, and I installed Ubuntu 16 again instead of Centos 7. I'm seeing 200-500 KB/s again downloading. I can get 5-10 MB/s upload, which I was getting before with Centos 7.

    This post hurts.

    And I can get 10 MB/s from Miami.

    Sure Ubuntu has always been great for connections to Miami.

    That's a bloody lie! slack is clearly better for downloading from Miami (though, granted, inferior to ubuntu when downloading from Dallas).

  • I was just trying to point out that Centos 7 vs. Ubuntu 16 might have different TCP settings :\

  • bsdguybsdguy Member
    edited January 2018

    @cirrus_cloud

    a) you are mixing up things (e.g. latency and throughput), b) if you get lower throughput from at home (than from other places) it may be due to either mere coincidence (rather unlikely) or to your provider (e.g. lousy routing/peering).

    But for the sake of clarity: what kernel do you have with ubuntu and centos?

    (P.S./Edit: The 2nd floor thing was a (poor attempt of) a joke. Just ignore it)

  • cirrus_cloudcirrus_cloud Member
    edited January 2018

    @bsdguy said:
    @cirrus_cloud

    Have you tried to download from the second (or any other) floor?

    I'm connected via ethernet.

    Yeah, I thought it was a joke. I'm not upset, though. :)

  • cirrus_cloudcirrus_cloud Member
    edited January 2018

    @bsdguy said:
    @cirrus_cloud

    a) you are mixing up things (e.g. latency and throughput), b) if you get lower throughput from at home (than from other places) it may be due to either mere coincidence (rather unlikely) or to your provider (e.g. lousy routing/peering).

    But for the sake of clarity: what kernel do you have with ubuntu and centos?

    uname -r 2.6.32-042stab127.2 #1 SMP Thu Jan 4 16:41:44 MSK 2018 x86_64 x86_64 x86_64 GNU/Linux

    I'm not confusing latency and throughput. Higher latency can cause issues with TCP since it can get throttled really easily on a connection with high latency (like 130-160ms in my case) because it's stateful and backs off if there's a bit of throttling. I'm not seeing a perfect connection on the various VPSes I've tried, but they're at least getting decent speed of 3, 5, or up to 10 MB/s, which is pretty close to the 100 mbps port speed. And they've all had higher latency (110-130), so I don't think Time4VPS is to blame, but I'm not a network expert.

  • trewqtrewq Administrator, Patron Provider

    I'd contact time4vps and try and get them to change to route to you to go over NTT and see if that fixes it.

Sign In or Register to comment.