Results 1 to 12 of 12

Thread: Limelight

  1. #1
    Join Date
    Aug 2004
    Location
    Southern NYS
    Posts
    533

    Limelight

    I've just started using Limelight for SHOUTcast hosting, and it is constantly buffering. Any idea?

  2. #2
    Join Date
    Aug 2002
    Location
    Seattle
    Posts
    5,512
    Go ahead and post some pings/traces from the clients you're having trouble with to their network.

    I recall having customers complaining of Limelight's Phoenix network prior to leaving them, not sure if the issues continue to persist.

  3. #3
    Join Date
    Aug 2004
    Location
    Southern NYS
    Posts
    533
    We're routed through Phoenix apparently...

    C:\>tracert 65.23.156.*

    Tracing route to 65.23.156.* over a maximum of 30 hops

    1 <1 ms <1 ms <1 ms 10.99.10.1
    2 6 ms 9 ms 8 ms 10.96.64.1
    3 11 ms 14 ms 10 ms dstswr2-vlan-2.rh.nantny.cv.net [67.83.252.162]

    4 10 ms 10 ms 12 ms r2-ge11-0.mhe.prnynj.cv.net [67.83.252.133]
    5 8 ms 7 ms 12 ms r2-srp3-0.wan.prnynj.cv.net [65.19.112.130]
    6 11 ms 10 ms 9 ms r2-srp1-0.in.nycmny83.cv.net [65.19.96.38]
    7 21 ms 16 ms 14 ms 65.19.101.130
    8 * * * Request timed out.
    9 53 ms 50 ms 55 ms ge1-11.fr1.dal.llnw.net [69.28.172.209]
    10 79 ms 79 ms 82 ms ge2-0-0-27.jr1.phx1.llnw.net [69.28.172.205]
    11 79 ms 139 ms 79 ms ge1-2-0.jr1.phx3.llnw.net [69.28.172.146]
    12 82 ms 79 ms 78 ms serve-rtr.phx3.llnw.net [69.28.128.206]
    13 84 ms 82 ms 85 ms 65.23.156.*

    Trace complete.
    Last edited by BaselineAce; 06-12-2005 at 09:42 PM.

  4. #4
    Join Date
    Aug 2002
    Location
    Seattle
    Posts
    5,512
    Phoenix is LLNW's only real POP. The rest of their POP's are small footprints (I would guess no more than a single cabinet of gear) in places such as Equinix.

    The traceroute you've shown is using Limelight's WVFiber transport and peering they have with CV. It's a decent route when you consider the distance, but it's still too far for media content. The latency is going to hurt.

  5. #5
    Join Date
    Aug 2004
    Location
    Southern NYS
    Posts
    533
    ...and now I'm pinging at 127ms, opposed to 80ms yesterday, as well as evidence of packet loss.
    Last edited by BaselineAce; 06-13-2005 at 08:32 AM.

  6. #6
    Join Date
    Aug 2004
    Location
    Southern NYS
    Posts
    533
    I should also mention that we're using Serve.com.

    The ping went back to normal, and then quickly jumped to 125ms.

  7. #7
    Join Date
    Oct 2000
    Posts
    1,654
    Ouch. Packet loss is never a good thing.

    Does latency have that great of an effect on Shoutcast hosting? I thought it was mostly driven by throughput...
    [QuickPacket™] [AS46261]
    Located in Atlanta, GA and Los Angeles, CA
    Dedicated Servers, KVM, Xen & OpenVZ VPS, Co-location, R1Soft Data Backup, Shared & Reseller Hosting

  8. #8
    Join Date
    Aug 2004
    Location
    Southern NYS
    Posts
    533
    Originally posted by QuickPacket
    Ouch. Packet loss is never a good thing.

    Does latency have that great of an effect on Shoutcast hosting? I thought it was mostly driven by throughput...
    No, but packet loss surely does.

  9. #9
    Join Date
    Oct 2004
    Posts
    300
    Yeah, packet loss is always bad, especially for media streaming. Better complain to Limelight and just leave them if they don't fix it immediately.

  10. #10
    Join Date
    Aug 2004
    Location
    Southern NYS
    Posts
    533
    Originally posted by blueface
    Yeah, packet loss is always bad, especially for media streaming. Better complain to Limelight and just leave them if they don't fix it immediately.
    Well the problem is actually on the Serve.com/Datarealm/Rackmounted.com end; it seems to be better right now. I'll be monitoring it all day to see...

  11. #11
    Join Date
    Aug 2004
    Location
    Southern NYS
    Posts
    533
    Just to update everyone, Datarealm seems to have shaped up its act. We average about one packet lost per day.
    PacketAce
    Because packets were meant to be delivered.
    Premium Mzima Bandwith at Equinix - Secaucus, NJ

  12. #12
    Join Date
    Feb 2004
    Location
    Louisville, Kentucky
    Posts
    1,083
    Originally posted by BaselineAce
    Just to update everyone, Datarealm seems to have shaped up its act. We average about one packet lost per day.
    That's not actually true. By design, TCP induces packet loss as the window size varies throughout the life of a particular TCP connection. For low-speed end-users or large file transfers to even high-speed end-users, there will always be some packet loss. It's not detremental, but your assertation that you average one packet lost per day is quite broad and incorrect.

    Perhaps you mean that, when using a monitoring application, your tool records one packet lost per day? That statement could easily be correct.
    Jeff at Innovative Network Concepts / 212-981-0607 x8579 / AIM: jeffsw6
    Expert IP network consultation and operation at affordable rates
    95th Percentile Explained Rate-Limiting on Cisco IOS switches

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •