Quantifying Availability

Collapse
This topic is closed.
X
X
 
  • Time
  • Show
Clear All
new posts
  • Jeremy

    Quantifying Availability

    How can I prove uptime (or quantify downtime) on an ASP.NET 1.1 Web
    application (running on an IIS6 Windows Server Standard server)?

    What's going on here is that I'm considering the feasability of claiming or
    at least establishing some percentage of availability for our production Web
    site (99.9%, or 99.99%, or 99.999% "five nines). Of course I'll only
    advertise the fact if I can establish it objectively (and it's at least 3
    nines :) For those interested, here's a link to an article that describes
    what each of those percentages actually means in terms of downtime:


    So, in IIS6 or perhaps some other feature I could build into our ASP.NET Web
    application, is there some way to objectively and accurately measure
    downtime or uptime? I'm specifically interested in quantifying the
    availability of the ASP.NET Web application.

    Suggestions? Thoughts? Perspective?

    Thanks!


  • Chris Crowe [MVP 2005/2006]

    #2
    Re: Quantifying Availability

    First thoughts - you can not use your ASP.NET application to measure
    availability.

    You need to test from a number of regions of the world - the reason is that
    if you test from say New Zealand and your server is in USA then maybe the
    New Zealand gateway is down, maybe the Pacific Gateway is down, maybe the US
    gatway is down etc...

    is Testing from multiple locations in the world will give you a number of
    locations with a number of gateways.

    So what are you going to test, a simple ping to the server?, a HTTP Request
    checking for a 200 status?, Check for some specific content in the returned
    data?

    This is all up to you to determine what you consider is up!

    If you have 5 sites worldwide, then each site contributes 20% of the total,
    now 5 9's means a total downtime of just over 5 minutes.

    5 minutes in my mind means - multiple servers, redundant hardware, redundant
    links...

    --

    Cheers

    Chris

    Chris Crowe [MVP 2005/2006]


    ------------------------------------------------
    "Jeremy" <a@b.com> wrote in message
    news:%23XFtBgM7 FHA.3416@TK2MSF TNGP15.phx.gbl. ..[color=blue]
    > How can I prove uptime (or quantify downtime) on an ASP.NET 1.1 Web
    > application (running on an IIS6 Windows Server Standard server)?
    >
    > What's going on here is that I'm considering the feasability of claiming
    > or at least establishing some percentage of availability for our
    > production Web site (99.9%, or 99.99%, or 99.999% "five nines). Of course
    > I'll only advertise the fact if I can establish it objectively (and it's
    > at least 3 nines :) For those interested, here's a link to an article that
    > describes what each of those percentages actually means in terms of
    > downtime:
    > http://searchnetworking.techtarget.c...064318,00.html
    >
    > So, in IIS6 or perhaps some other feature I could build into our ASP.NET
    > Web application, is there some way to objectively and accurately measure
    > downtime or uptime? I'm specifically interested in quantifying the
    > availability of the ASP.NET Web application.
    >
    > Suggestions? Thoughts? Perspective?
    >
    > Thanks!
    >[/color]


    Comment

    • Jeremy

      #3
      Re: Quantifying Availability

      Well, not so fast! : )

      The people who established this whole "5 nines" availability concept are
      from the telephone industry and, for years, they've been measuring the
      availability of their various systems (e.g., PBX) and they explicitly
      exclude individual telephones and wiring to get TO the PBX board. Yes, those
      things are "part of the system" but are explicitly *excluded* from their
      availability measures. It would simply be unrealistic to attempt to include
      every telephone in the entire world in their availability measures.
      Additionally, they are measuring the availabilty of the PBX (or other
      switching unit), and *not* measuring the availability of wires getting to it
      or of individual telephones at the other end of the wire.

      So, I think the same rationalle could be easily generalized over to an
      ASP.NET Web site. It can be measured as available - even if some people in
      the world cannot get to it.

      Put another way, the bank may be open for business, but you cannot get to
      the bank because your car has a flat tire. Your flat tire does not in any
      way make the bank any less available. It's still open for business.

      Any other ideas?






      "Chris Crowe [MVP 2005/2006]" <IISMVP2005@iis faq.homeip.net> wrote in
      message news:%23fzbDWO7 FHA.4084@TK2MSF TNGP10.phx.gbl. ..[color=blue]
      > First thoughts - you can not use your ASP.NET application to measure
      > availability.
      >
      > You need to test from a number of regions of the world - the reason is
      > that if you test from say New Zealand and your server is in USA then maybe
      > the New Zealand gateway is down, maybe the Pacific Gateway is down, maybe
      > the US gatway is down etc...
      >
      > is Testing from multiple locations in the world will give you a number of
      > locations with a number of gateways.
      >
      > So what are you going to test, a simple ping to the server?, a HTTP
      > Request checking for a 200 status?, Check for some specific content in the
      > returned data?
      >
      > This is all up to you to determine what you consider is up!
      >
      > If you have 5 sites worldwide, then each site contributes 20% of the
      > total, now 5 9's means a total downtime of just over 5 minutes.
      >
      > 5 minutes in my mind means - multiple servers, redundant hardware,
      > redundant links...
      >
      > --
      >
      > Cheers
      >
      > Chris
      >
      > Chris Crowe [MVP 2005/2006]
      > http://blog.crowe.co.nz
      >
      > ------------------------------------------------
      > "Jeremy" <a@b.com> wrote in message
      > news:%23XFtBgM7 FHA.3416@TK2MSF TNGP15.phx.gbl. ..[color=green]
      >> How can I prove uptime (or quantify downtime) on an ASP.NET 1.1 Web
      >> application (running on an IIS6 Windows Server Standard server)?
      >>
      >> What's going on here is that I'm considering the feasability of claiming
      >> or at least establishing some percentage of availability for our
      >> production Web site (99.9%, or 99.99%, or 99.999% "five nines). Of course
      >> I'll only advertise the fact if I can establish it objectively (and it's
      >> at least 3 nines :) For those interested, here's a link to an article
      >> that describes what each of those percentages actually means in terms of
      >> downtime:
      >> http://searchnetworking.techtarget.c...064318,00.html
      >>
      >> So, in IIS6 or perhaps some other feature I could build into our ASP.NET
      >> Web application, is there some way to objectively and accurately measure
      >> downtime or uptime? I'm specifically interested in quantifying the
      >> availability of the ASP.NET Web application.
      >>
      >> Suggestions? Thoughts? Perspective?
      >>
      >> Thanks!
      >>[/color]
      >
      >[/color]


      Comment

      • Ken Schaefer

        #4
        Re: Quantifying Availability

        Hi,

        As Chris has mentioned - you need to decide what "available" means. The
        server returns a particular page? That certain functionality is still
        working? That no requests are timing out (because site or backend is
        overloaded). There are a number of 3rd party independant companies that
        measure "uptime" (which the major sites use to verify uptime). You might
        want to have a look at what those monitoring companies define as "available"

        If you want to do the monitoring yourself, you could use something like MOM
        (MS Operations Manager) plus the IIS Management Pack to get some idea of
        what the site's currently responding to.

        But site available is more than just the web components. As you way, you'll
        want to exclude the stuff outside your control, but your edge routers, load
        balancing equipment etc will need to be monitored (there's no point saying
        your website's up for 99.999% of the time if your load balancer is down for
        half the day!). All the really big websites have load balancing between
        disparate data centers to get around this (e.g. microsoft.com is
        geographically load balanced between 3 data centers)

        Cheers
        Ken

        "Jeremy" <a@b.com> wrote in message
        news:eBfCriS7FH A.3864@TK2MSFTN GP10.phx.gbl...
        : Well, not so fast! : )
        :
        : The people who established this whole "5 nines" availability concept are
        : from the telephone industry and, for years, they've been measuring the
        : availability of their various systems (e.g., PBX) and they explicitly
        : exclude individual telephones and wiring to get TO the PBX board. Yes,
        those
        : things are "part of the system" but are explicitly *excluded* from their
        : availability measures. It would simply be unrealistic to attempt to
        include
        : every telephone in the entire world in their availability measures.
        : Additionally, they are measuring the availabilty of the PBX (or other
        : switching unit), and *not* measuring the availability of wires getting to
        it
        : or of individual telephones at the other end of the wire.
        :
        : So, I think the same rationalle could be easily generalized over to an
        : ASP.NET Web site. It can be measured as available - even if some people in
        : the world cannot get to it.
        :
        : Put another way, the bank may be open for business, but you cannot get to
        : the bank because your car has a flat tire. Your flat tire does not in any
        : way make the bank any less available. It's still open for business.
        :
        : Any other ideas?
        :
        :
        :
        :
        :
        :
        : "Chris Crowe [MVP 2005/2006]" <IISMVP2005@iis faq.homeip.net> wrote in
        : message news:%23fzbDWO7 FHA.4084@TK2MSF TNGP10.phx.gbl. ..
        : > First thoughts - you can not use your ASP.NET application to measure
        : > availability.
        : >
        : > You need to test from a number of regions of the world - the reason is
        : > that if you test from say New Zealand and your server is in USA then
        maybe
        : > the New Zealand gateway is down, maybe the Pacific Gateway is down,
        maybe
        : > the US gatway is down etc...
        : >
        : > is Testing from multiple locations in the world will give you a number
        of
        : > locations with a number of gateways.
        : >
        : > So what are you going to test, a simple ping to the server?, a HTTP
        : > Request checking for a 200 status?, Check for some specific content in
        the
        : > returned data?
        : >
        : > This is all up to you to determine what you consider is up!
        : >
        : > If you have 5 sites worldwide, then each site contributes 20% of the
        : > total, now 5 9's means a total downtime of just over 5 minutes.
        : >
        : > 5 minutes in my mind means - multiple servers, redundant hardware,
        : > redundant links...
        : >
        : > --
        : >
        : > Cheers
        : >
        : > Chris
        : >
        : > Chris Crowe [MVP 2005/2006]
        : > http://blog.crowe.co.nz
        : >
        : > ------------------------------------------------
        : > "Jeremy" <a@b.com> wrote in message
        : > news:%23XFtBgM7 FHA.3416@TK2MSF TNGP15.phx.gbl. ..
        : >> How can I prove uptime (or quantify downtime) on an ASP.NET 1.1 Web
        : >> application (running on an IIS6 Windows Server Standard server)?
        : >>
        : >> What's going on here is that I'm considering the feasability of
        claiming
        : >> or at least establishing some percentage of availability for our
        : >> production Web site (99.9%, or 99.99%, or 99.999% "five nines). Of
        course
        : >> I'll only advertise the fact if I can establish it objectively (and
        it's
        : >> at least 3 nines :) For those interested, here's a link to an article
        : >> that describes what each of those percentages actually means in terms
        of
        : >> downtime:
        : >>

        : >>
        : >> So, in IIS6 or perhaps some other feature I could build into our
        ASP.NET
        : >> Web application, is there some way to objectively and accurately
        measure
        : >> downtime or uptime? I'm specifically interested in quantifying the
        : >> availability of the ASP.NET Web application.
        : >>
        : >> Suggestions? Thoughts? Perspective?
        : >>
        : >> Thanks!
        : >>
        : >
        : >
        :
        :


        Comment

        • Parvez

          #5
          RE: Quantifying Availability

          to cut the chase and provide a direct answer , try using bigbrother.com. au a
          service that measures site availibility/uptime , you would need to install
          the client on your web server.
          The BB service will monitor IIS for uptime, and check if the IIS Service is
          running, else it would mean the site is down and not working, it would not
          account for any additional service checks, and definetely not if the gateways
          are up, or bandwidth/network is having issues


          "Jeremy" wrote:
          [color=blue]
          > How can I prove uptime (or quantify downtime) on an ASP.NET 1.1 Web
          > application (running on an IIS6 Windows Server Standard server)?
          >
          > What's going on here is that I'm considering the feasability of claiming or
          > at least establishing some percentage of availability for our production Web
          > site (99.9%, or 99.99%, or 99.999% "five nines). Of course I'll only
          > advertise the fact if I can establish it objectively (and it's at least 3
          > nines :) For those interested, here's a link to an article that describes
          > what each of those percentages actually means in terms of downtime:
          > http://searchnetworking.techtarget.c...064318,00.html
          >
          > So, in IIS6 or perhaps some other feature I could build into our ASP.NET Web
          > application, is there some way to objectively and accurately measure
          > downtime or uptime? I'm specifically interested in quantifying the
          > availability of the ASP.NET Web application.
          >
          > Suggestions? Thoughts? Perspective?
          >
          > Thanks!
          >
          >
          >[/color]

          Comment

          Working...