Howdy, Stranger!

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


Faster downloads from kimsufi than dacentec
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.

Faster downloads from kimsufi than dacentec

So my downloads from dacentec are around 700Kb/s at most and normally around 300Kb/s. reinstalled and restarted the dedicated server several times.

My Kimsufi can saturate my Home Connection. Lol This has been this was for awhile.

Anyone else experience anything differently?

At first the first 3 weeks i could saturate my home connection but now the dacentec server gives above results.

Atleast my Kimsufi server is still running and can download the content from it and backup to it still lol.

Comments

  • comXyzcomXyz Member

    I'm not sure what you mean.

    So you get around 700 kbs when download from dacentec server. And what is the speed when you download from Kimsufi server?

  • Where are you located? This can depend on that.

  • Where are you and where is your kimsufi?

  • Im sorry but im so waiting for "i live in france"

    Thanked by 3Lee ATHK howardsl2
  • TheLinuxBugTheLinuxBug Member
    edited July 2015

    @cmsjr123 This likely is more related to your home ISP changing their routes to the destination rather than the data canter. Also, peering and routing vary between data centers. Many providers change routing across their network constantly to try and offer the best possible services to their network (at the best possible price) both residential providers and data centers, which can also cause your routing and throughput to vary.

    As an example of this in the US, Comcast (a cable internet provider) often switches from Tier 1 providers (Level3, telia, etc) during the day to pure Cogent at night because the routes are cheaper for them. This will cause you to see different routing and throughput at night than you would see during the day as the Cogent routes are often oversold. This is only one example, but many ISPs (cable companies, DSL companies) do the same thing constantly to save them selves money. The result of this could be better throughput and latency during business hours, and slower higher latency during off business hours.

    TL; DR:

    Routing/Peering is not a constant thing, it is always changing. There is no guarantee that you will have the same routes tomorrow as you do today and this can in some cases drastically effect throughput and latency to your destination. If it is bad enough, your best recourse is to contact your local provider (Cable company, DSL provider, etc) and let them know that you are having problems with getting good throughput or latency to a specific network. If you do this, be ready to provide specific information (traceroutes, mtr, transfers) to your provider so they are able to track down and correct the routing issue for you. While some may be able to offer you a solution and fix the routing, also be prepared to see no change if the route provided is the best they are willing to offer.

    I hope this helps.

    Cheers!

    Thanked by 2alexvolk Issam2204
  • This is what affects speeds:

    You have a residential ISP providing service to you.

    When you send traffic to your server, the shortest path is taken to transmit your data. It goes through your ISP's peers and through their peers until it gets to your destination (your server).

    If it takes a very long route, your speeds will drop.

    If there's a very direct peering agreement between your ISP and the peers, then your speeds will be higher.

    And again, it could just be someone sending layer 4 DOS attacks.

  • I live in california. FAR farther from France XD

    The longest route is to the Kimsufi

    only a few hops to dacentec about 30 to kimsufi.

  • @TheLinuxBug said:
    cmsjr123 This likely is more related to your home ISP changing their routes to the destination rather than the data canter. Also, peering and routing vary between data centers. Many providers change routing across their network constantly to try and offer the best possible services to their network (at the best possible price) both residential providers and data centers, which can also cause your routing and throughput to vary.

    As an example of this in the US, Comcast (a cable internet provider) often switches from Tier 1 providers (Level3, telia, etc) during the day to pure Cogent at night because the routes are cheaper for them. This will cause you to see different routing and throughput at night than you would see during the day as the Cogent routes are often oversold. This is only one example, but many ISPs (cable companies, DSL companies) do the same thing constantly to save them selves money. The result of this could be better throughput and latency during business hours, and slower higher latency during off business hours.

    TL; DR:

    Routing/Peering is not a constant thing, it is always changing. There is no guarantee that you will have the same routes tomorrow as you do today and this can in some cases drastically effect throughput and latency to your destination. If it is bad enough, your best recourse is to contact your local provider (Cable company, DSL provider, etc) and let them know that you are having problems with getting good throughput or latency to a specific network. If you do this, be ready to provide specific information (traceroutes, mtr, transfers) to your provider so they are able to track down and correct the routing issue for you. While some may be able to offer you a solution and fix the routing, also be prepared to see no change if the route provided is the best they are willing to offer.

    I hope this helps.

    Cheers!

    I did not know this to be honest. I was not aware they could change routes during different time but this is a consistent thing regardless of time.

    regardless the speeds to the server in dacentec remain fine. So Could be My ISP..

    Just baffled at the 30 hops to kimsufi maxing my home 100Mbit out and dacentec which is only a few hops to getitng 700Kb/s..

    Oh well. Will run traceroutes at regular intervals and measure them to see if routes are changing.

  • TheLinuxBugTheLinuxBug Member
    edited July 2015

    @cmsjr123 I should add that some data centers (possibly Dacentec) also have the ability to manipulate routes from their side to the destination network as well. You may consider opening a ticket to them as well and documenting the issue in the same way (providing traceroutes, mtr, transfer examples) so they can review and see if it is possible to use a different route to your ISP. I will mention, it is usually more productive to contact your internet provider directly though as often the routing issues can also stem from inside their own network, where changes on the data center side may or may not have the desired outcome.

    If the speed change has been as drastic as you stated, opening a ticket to Dacentec might actually help. If they are unable to fix the issue by manipulating routes on their side, then you will need to contact your local ISP directly and let them know of the issue.

    I hope this helps.

    Cheers!

  • Ticket them with mtr's in both directions + download results. They've fixed a few routes for me before, their support is great.

  • edited July 2015

    Sent a ticket in. waiting for a reply now.

    Wondering what it will be. Tracert seems to show charter all the way to the server from me to the dacentec network. From them to me seems to be 5 hops then out of dacentec then 1 hop to a 10ge core then to 10ge charter core then to cahrter network which goes to me.. far more hops out of the dacentec network to my home then the hop into dacentec directly from the charter network instead of having the 10ge core from another network.

    O.o wondering what they find. :}

    edited for spelling error

  • Connecting to my VPN that uses several different networks all around the US including LA which is only 120-150miles from me Yields not only less hops returning to me but sustained download speeds of maxing out my connecting to my home. -.-

    Well now we know the route to my home is the issue....

  • @cmsjr123 said:
    Connecting to my VPN that uses several different networks all around the US including LA which is only 120-150miles from me Yields not only less hops returning to me but sustained download speeds of maxing out my connecting to my home. -.-

    Well now we know the route to my home is the issue....

    A couple users of Dacentec found that their outbound traffic from their box was being shaped... harshly. (We're talking 300Mbps throttled down to 5Mbps and less)

    Keep an eye on it.

  • tomletomle Member, LIR

    Question is why, traffic is metered and should be "free" until the limit is reached.

  • tomle said: Question is why, traffic is metered and should be "free" until the limit is reached.

    Some networks do this to keep their overall traffic under their committed rate with their upstream. So if a ground of users is using too much of 'Expensive Transit Provider' then they rate limit it to avoid paying excess bandwidth fees.

  • must be that charter is to expensive for them then. It is charter to me nearly all the ay except for one core -.- This is bullshit. I also figured out that after running a download to my home using VPN that when ran a second time it capped at around 700 kbps to 1.1MB/s.

    Same thing with all destination for the VPN regardless of what transit providers the VPN uses as well.

    thats some funky smelling stuff right there.

  • @cmsjr123 said:
    must be that charter is to expensive for them then. It is charter to me nearly all the ay except for one core -.- This is bullshit. I also figured out that after running a download to my home using VPN that when ran a second time it capped at around 700 kbps to 1.1MB/s.

    Same thing with all destination for the VPN regardless of what transit providers the VPN uses as well.

    thats some funky smelling stuff right there.

    As I said they seem to shape outbound traffic. Message their support saying it's clearly being shaped and you might have the shaping removed. I know of one user who had his shaping removed and has no issues now. The other friend who was using them dropped their server because the shaping was bad.

  • Thanks. I wonder what will happen lol. I should have went with delimiter butttttt that 5tb of bandwidth sucks. Decantec has better bandwidth pricing :/ I guess my fault....

  • @cmsjr123 said:
    Thanks. I wonder what will happen lol. I should have went with delimiter butttttt that 5tb of bandwidth sucks. Decantec has better bandwidth pricing :/ I guess my fault....

    Just contact them. As I said somebody got it all sorted out. Took a bit but was resolved.

    Couldn't hurt.

  • @IThinkUFailed said:
    A couple users of Dacentec found that their outbound traffic from their box was being shaped... harshly. (We're talking 300Mbps throttled down to 5Mbps and less)

    I've been wondering if my Dacentec server is being shaped too - or where it's being shaped. Since setting it up two months ago I've been running a restore (https from Crashplan) to the server from Sydney Australia and seen that restore speed constantly sit at 2Mb/s. Never higher or lower.

    Example:

    image

    Green inbound is the restore traffic. Locked @ 2Mb/s. The few blips above and below that line would be other things I was downloading at the time.

    The Blue outbound is a rsync copy I performed to another server, located in France. That session showed signs of shaping too - originally sitting on 3Mb/s for a day, then hopping up to 6Mb/s. Then some network condition clearly changed and the copy started running at a more random speed.

    I raised this with Dacentec and they assure me it's not their network controls at play. I also raised a ticket with Crashplan about the restore speeds and they gave me the same "not us" answer.

    Looks like a Dacentec control to me. I'd like to know for sure either way.

  • edited August 2015

    @poolroom said:

    Looks like a Dacentec control to me. I'd like to know for sure either way.

    Dacentec did the same thing and blamed charter 100%..

    Needless to say where I move I might just get fios 300/300 In the next month and just use it for my Shit...... Sigh...

  • cmsjr123 said: I should have went with delimiter butttttt that 5tb of bandwidth sucks

    But if you believe Dacentec are shaping you then it doesn't really matter if they include 100000TB, you can't consume it.

    Delimiter has NO shaping/rate limiting so you can burst the full gigabit/10 gigabit as needed until you run out of included bandwidth

    Thanked by 1netomx
  • poolroom said: I've been running a restore (https from Crashplan)

    Crashplan is known to shape traffic, this could be Crashplan!

    You can easily test for general shaping or specific route shaping.

  • poolroom said: Green inbound is the restore traffic. Locked @ 2Mb/s. The few blips above and below that line would be other things I was downloading at the time.

    Agree with @MarkTurner - I bet this is Crashplan. I've always had that issue with them.

    Thanked by 1netomx
  • @poolroom said:

    Looks like a Dacentec control to me. I'd like to know for sure either way.

    Dacentec only shape outbound it seems. (Uploading to other boxes) would definitely say it's not them if its inbound traffic.

Sign In or Register to comment.