Re: Retrieving records in order they were placed.

Hello,

Not sure why your sequences would need to rollover in your lifetime.

For instance, imagine you want to insert one million items per second into a database. This gives about 3.1 E13 events per year. A 20 digit sequence number will use 11 bytes of space and rollover in about two million years.
Ok, thats silly, how about a thousand events a second and a sequence number which occupies the same space as a date? Rollover in 292 years.

Note: Oracle will work with integers with more than 40 digits of precision. It is silly accuracy.

I have tried a 20 digit sequence - works fine.

graham

On Fri, 09 Jul 1999 12:57:03 GMT, b_rich2_at_my-deja.com wrote:

>I have to put records into a table, and later retrieve them in the order
>in which they were placed. I have a very very large number of entries to
>be placed in a live system. The records are logs of important events in
>the system.
>
>My first thought was to use the unix time stamp to place as the field to
>order on, but there will probably be many entries entered in rapid
>succession. Thus many entries could be entered at the same time stamp
>interval.
>
>I then though of having a sequence number associated with each record,
>increasing one by one with each record stored. This will work well until
>the sequence numbers begin to wrap. If they are retrieved in order of
>sequence number, and then they wrap, the records entered after the wrap
>(sequences 1,2,3,4 etc) will all of a sudden be pulled out before the
>earlier records with high sequence numbers.
>
>Another idea is to just use the sequence numbers, and re-boot the box
>every few weeks when the sequence numbers get high, but this is really
>unacceptable in the environment where this will be run.
>
>Does anyone know how to get around this problem? Any help would be
>GREATLY appreciated.
>
>
>Sent via Deja.com http://www.deja.com/
>Share what you know. Learn what you don't.
Received on Fri Jul 09 1999 - 19:18:09 CEST