Results 1 to 18 of 18
  1. #1
    Join Date
    May 2004
    Location
    San Francisco Bay Area
    Posts
    143

    Longest Network Uptime

    In a data center and/or ISP environment, what is the longest network uptime you have experienced?

    Please be honest with your answers. Try not to forget the times there were little problems.

  2. #2
    Join Date
    Feb 2004
    Location
    Georgia, USA
    Posts
    201
    thats a great question, would love to know that one myself. Something to compare our uptime too.

  3. #3
    Join Date
    Mar 2004
    Location
    USA
    Posts
    4,345
    1 month
    Testing 1.. Testing 1..2.. Testing 1..2..3...

  4. #4
    Join Date
    Oct 2003
    Location
    Australia
    Posts
    1,738
    Well with the rate at which security updates are required, 100% uptime over even a few months is a bit far-fetched.

    But my local ISP (DSL in Australia) goes down far more often than any datacentre I have ever had dealings with
    Brad Baker
    XYZulu Hosting

  5. #5
    Join Date
    Mar 2003
    Location
    Duluth MN
    Posts
    3,863
    Over the past year, I have had 100% network availablility/uptime with my datacenter. Although, I have had downtime, it was all due to faults on my end (i.e. rebooting a server etc)

    Of course, we are talking about network availablilty/uptime.

  6. #6
    Join Date
    Oct 2003
    Location
    Australia
    Posts
    1,738
    Of course, we are talking about network availablilty/uptime.
    'scuse my hastliy typed previous reply.

    DC network uptime can reach near 100%
    Brad Baker
    XYZulu Hosting

  7. #7
    Join Date
    Mar 2003
    Location
    Duluth MN
    Posts
    3,863
    Yeah, its almost hard for a DC to have less than 99.9/100% network uptime, especially since the SLA's they have with the backbone providers guarantee the same. And when you have multiple backbones, the chances of all of them failing for the 0.01% at the same time is very, very, very little.

    The only real way a DC can have network outages is to have massive hardware failure with core routers insided the DC, to be DDOSed, or to have massive power failure like the rolling blackouts in NY last summer.

  8. #8
    I don't believe Drexel's cogent line has gone down the entire 9 months I've been here.
    Dan Grossman - dan @ awio.com
    My Blog | Affiliate Program for Web Hosts

  9. #9
    Join Date
    Mar 2003
    Posts
    587
    I had almost 8 months of network and server uptime on few servers...i had ordered few things for our horses so the credit card was full so because of that the payment didn't go through for servers. when getting the servers i had used a yahoo email which i hardly ever checked after a while...so due to this i had abt 10 hours downtime and its been abt 2/3 months no downtime again..

  10. #10
    Join Date
    Mar 2003
    Posts
    587
    Originally posted by amish_geek
    The only real way a DC can have network outages is to have massive hardware failure with core routers insided the DC, to be DDOSed, or to have massive power failure like the rolling blackouts in NY last summer.
    actually me and a friend of mine shared server which was in the peer1 DC here in manhattan...went through the power outage without a single problem...i guess we just need more DC's like that and all of us can have long uptimes...reboots while updating hardly matter

  11. #11
    Join Date
    May 2004
    Location
    San Francisco Bay Area
    Posts
    143
    Ok, Ok. Some people here are saying that data centers hardly have problems. Let's be real here people. Human mistakes happen. This is what causes most of the problems.

    -The wrong cross-connect being pulled or moved.
    -Some genius who decided to do an IOS upgrade at 2pm on a production router
    -Circuits flapping
    -Someone pressed the EPO (emergency power off button) on accident
    -A power generator test that went bad
    -A power circuit tripped bringing down 10 racks.
    -A shared upstream device gets flooded
    Etc...

    You know what I'm talking about. Be honest here people...
    We've all herd a plenty stories. Check out the network/power outage forum on WHT for pete's sake...

  12. #12
    Join Date
    Oct 2003
    Location
    Tas, Australia
    Posts
    2,487
    Well...

    Rackspace.com claims to have 3 years of uptime...
    New site: www.talkutas.com

  13. #13
    Its possible if you have multiple Core Routers (EG: A cluster)

  14. #14
    Join Date
    May 2004
    Location
    San Francisco Bay Area
    Posts
    143
    Originally posted by HG-Owen
    Its possible if you have multiple Core Routers (EG: A cluster)
    Core router "clusters" help but are not the reason.

    Multiple cores may help keep the edge up but that does not help keep customers up.. dialup customers or data center customers.

  15. #15
    Join Date
    May 2004
    Location
    San Francisco Bay Area
    Posts
    143
    285 days +/- is the longest uptime I've ever experienced.
    My server never went down, the data center never had a facility problem (like power or something) and the ISP I use never had a network issue. Then one day they did network upgrades....

    Outage time. About 10 minutes. Not too bad.

  16. #16
    You can have %100 network uptime, see the following crappy paint.
    Attached Thumbnails Attached Thumbnails network1.jpg  

  17. #17
    Join Date
    May 2004
    Location
    San Francisco Bay Area
    Posts
    143
    Originally posted by dk2
    You can have %100 network uptime, see the following crappy paint.

    Nice pretty colors.
    Anyone can create a topology like that. Making it work right and supporting it are two different things all together.

    What about route redistribution?
    What about protocol metrics and timers?
    What about convergence?
    What about change management?
    What about filters so a network config mistake made is not populated into every adjacent device?
    What about network interface attacks?
    What about CPU and memory performance issues that may cause protocols to time out making a number of devices look as though they were down but really not?
    ETc, ETc,

    A picture of a standard full mesh topology does not make a network have 100% uptime especially when a full mesh at the core/edge causes more problems than it solves.

    That would be like me drawing a picture of a spaceship and saying "see, it's possible to build a spaceship that will fly to the MilkyWay".

    Your picture only represents about 1% of the work involved in designing and supporting a nework with 100% uptime. The same is with desiging a car, spaceship or building. Pictures and diagrams only help present a concept. But actually creating and supporting the concept is where the real work lies.


    Don't get me wrong. I'm not trashing your picture. It is a decent diagram. It's just that too many times people misinterpret diagrams for actual solutions.
    Last edited by appolo; 06-03-2004 at 02:01 PM.

  18. #18
    appolo I totally understand what your saying, I'm just saying it can be done with the right setup. A very expensive setup ( is it bad that I have fantasies about a setup like this?)

    You just gotta use good routers (juniper, cisco), good switches(cisco, 3 com, etc.), and have a kick butt network engineer/network archietect ( the one who you catch jerking off to a CDW catalog instead of a playboy).

Posting Permissions

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