sql server sizing planning - architecture help

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

    sql server sizing planning - architecture help

    Hi

    I joined a project where 100,000 rows were added everyday. Now due to
    additional customers the expectation is 2 million reocrds/day ie 10 GB worth
    of textfiles. We have to estimate the hard disk, memory, # of CPUs etc.We
    will have one yearworth of data in the db. Rest will be in tapes etc.

    We will be using WIN2000, SQL Server2000.- Any comparable server sizing will
    be appreciated.

    1. Tohandle every day load, I thought that we will have a table for each day
    (pre created in the database )and have a view with union all selecting
    fromall these 365 tables. (This is the only way to partition in MSSQL Server
    right?).

    2. The requirement is to populate datawarehouse tables with all the data.
    However there will be only inserts mostly but there can be updates too which
    happenned in the past 12 days.Hence we have to use the data from the last
    12 days and massage it etc and populate into datawarehouse tables.

    How can I do this so that I will have the datawarehouse tables with n-12
    days of data and I will alwys add the last 12 days data to it.

    Do you have any suggestions?

    Ragu



  • John Bell

    #2
    Re: sql server sizing planning - architecture help

    Hi

    Check out "Estimating the Size of a Database" in Books online:
    mk:@MSITStore:C :\Program%20Fil es\Microsoft%20 SQL%20Server\80 \Tools\Books\cr e
    atedb.chm::/cm_8_des_02_92k 3.htm

    mk:@MSITStore:C :\Program%20Fil es\Microsoft%20 SQL%20Server\80 \Tools\Books\cr e
    atedb.chm::/cm_8_des_02_33j s.htm

    mk:@MSITStore:C :\Program%20Fil es\Microsoft%20 SQL%20Server\80 \Tools\Books\cr e
    atedb.chm::/cm_8_des_02_224 8.htm


    John

    "Ragu" <ragudba@sbcglo bal.net> wrote in message
    news:vgToc.1089 0$TC5.2751@news svr23.news.prod igy.com...[color=blue]
    > Hi
    >
    > I joined a project where 100,000 rows were added everyday. Now due to
    > additional customers the expectation is 2 million reocrds/day ie 10 GB[/color]
    worth[color=blue]
    > of textfiles. We have to estimate the hard disk, memory, # of CPUs etc.We
    > will have one yearworth of data in the db. Rest will be in tapes etc.
    >
    > We will be using WIN2000, SQL Server2000.- Any comparable server sizing[/color]
    will[color=blue]
    > be appreciated.
    >
    > 1. Tohandle every day load, I thought that we will have a table for each[/color]
    day[color=blue]
    > (pre created in the database )and have a view with union all selecting
    > fromall these 365 tables. (This is the only way to partition in MSSQL[/color]
    Server[color=blue]
    > right?).
    >
    > 2. The requirement is to populate datawarehouse tables with all the data.
    > However there will be only inserts mostly but there can be updates too[/color]
    which[color=blue]
    > happenned in the past 12 days.Hence we have to use the data from the last
    > 12 days and massage it etc and populate into datawarehouse tables.
    >
    > How can I do this so that I will have the datawarehouse tables with n-12
    > days of data and I will alwys add the last 12 days data to it.
    >
    > Do you have any suggestions?
    >
    > Ragu
    >
    >
    >[/color]


    Comment

    • John Bell

      #3
      Re: sql server sizing planning - architecture help

      Hi

      In addition....

      On one of the CDs accompanying "Inside SQL Server 2000" by Kalen Delany ISBN
      0-7356-0998-5, there is a stored procedure which will give you the size.

      John

      "John Bell" <jbellnewsposts @hotmail.com> wrote in message
      news:7empc.92$M w2.5784667@news-text.cableinet. net...[color=blue]
      > Hi
      >
      > Check out "Estimating the Size of a Database" in Books online:
      >[/color]
      mk:@MSITStore:C :\Program%20Fil es\Microsoft%20 SQL%20Server\80 \Tools\Books\cr e[color=blue]
      > atedb.chm::/cm_8_des_02_92k 3.htm
      >
      >[/color]
      mk:@MSITStore:C :\Program%20Fil es\Microsoft%20 SQL%20Server\80 \Tools\Books\cr e[color=blue]
      > atedb.chm::/cm_8_des_02_33j s.htm
      >
      >[/color]
      mk:@MSITStore:C :\Program%20Fil es\Microsoft%20 SQL%20Server\80 \Tools\Books\cr e[color=blue]
      > atedb.chm::/cm_8_des_02_224 8.htm
      >
      >
      > John
      >
      > "Ragu" <ragudba@sbcglo bal.net> wrote in message
      > news:vgToc.1089 0$TC5.2751@news svr23.news.prod igy.com...[color=green]
      > > Hi
      > >
      > > I joined a project where 100,000 rows were added everyday. Now due to
      > > additional customers the expectation is 2 million reocrds/day ie 10 GB[/color]
      > worth[color=green]
      > > of textfiles. We have to estimate the hard disk, memory, # of CPUs[/color][/color]
      etc.We[color=blue][color=green]
      > > will have one yearworth of data in the db. Rest will be in tapes etc.
      > >
      > > We will be using WIN2000, SQL Server2000.- Any comparable server sizing[/color]
      > will[color=green]
      > > be appreciated.
      > >
      > > 1. Tohandle every day load, I thought that we will have a table for each[/color]
      > day[color=green]
      > > (pre created in the database )and have a view with union all selecting
      > > fromall these 365 tables. (This is the only way to partition in MSSQL[/color]
      > Server[color=green]
      > > right?).
      > >
      > > 2. The requirement is to populate datawarehouse tables with all the[/color][/color]
      data.[color=blue][color=green]
      > > However there will be only inserts mostly but there can be updates too[/color]
      > which[color=green]
      > > happenned in the past 12 days.Hence we have to use the data from the[/color][/color]
      last[color=blue][color=green]
      > > 12 days and massage it etc and populate into datawarehouse tables.
      > >
      > > How can I do this so that I will have the datawarehouse tables with n-12
      > > days of data and I will alwys add the last 12 days data to it.
      > >
      > > Do you have any suggestions?
      > >
      > > Ragu
      > >
      > >
      > >[/color]
      >
      >[/color]


      Comment

      • Erland Sommarskog

        #4
        Re: sql server sizing planning - architecture help

        Ragu (ragudba@sbcglo bal.net) writes:[color=blue]
        > I joined a project where 100,000 rows were added everyday. Now due to
        > additional customers the expectation is 2 million reocrds/day ie 10 GB
        > worth of textfiles. We have to estimate the hard disk, memory, # of CPUs
        > etc.We will have one yearworth of data in the db. Rest will be in tapes
        > etc.
        > ...
        > 1. Tohandle every day load, I thought that we will have a table for each
        > day (pre created in the database )and have a view with union all
        > selecting fromall these 365 tables. (This is the only way to partition
        > in MSSQL Server right?).[/color]

        Yes, you set up a partitioned view over all this tables. Be careful
        to apply your check constraints accordingly.

        However, I'm in doubt that having one table per day is really a good
        thing. Yes, loads will be fast, if the table is empty and clean
        each day. But with a clustered index on date, you would append the
        new rows at the end. One table per month seems more manageable....
        [color=blue]
        > 2. The requirement is to populate datawarehouse tables with all the
        > data. However there will be only inserts mostly but there can be updates
        > too which happenned in the past 12 days.Hence we have to use the data
        > from the last 12 days and massage it etc and populate into datawarehouse
        > tables.
        >
        > How can I do this so that I will have the datawarehouse tables with n-12
        > days of data and I will alwys add the last 12 days data to it.[/color]

        Too little information to say anything substantial, but a common technique
        is to first load into a staging table, and then take it from there. Note
        that for the previous days, you would need to do both and UPDATE and an
        INSERT, presuming that new rows may appear too.

        --
        Erland Sommarskog, SQL Server MVP, sommar@algonet. se

        Books Online for SQL Server SP3 at
        SQL Server 2025 redefines what's possible for enterprise data. With developer-first features and integration with analytics and AI models, SQL Server 2025 accelerates AI innovation using the data you already have.

        Comment

        Working...