Interval Tables

[joo_mariano2]

Hi SDN's,

My program has access a table of 4 million register of sequential form. To improve the performance I want to mount one another table with the intervals of the primary key. Which the best way to make this?

Thank's

[Former Member]

Hi,

While selecting from the table:

1. Select only required information

2. Keep the result set small

I dont understand what you mean by "mount one another table". It would be better if you could create an index on that table.

Regards

Wenceslaus

[joo_mariano2]

I dont to create a new Index for my selection criteria(Field type DATE).

I want to create a table like this:

pk_low and pk_high

Instead of having access the table an only time, access some times by means of this interval.

In Standard exists some thing to become this?

[Former Member]

I would just select min(pk) and max(pk) from your table, then split this interval evenly and select your data for each interval... but this approach with intervals will help ONLY if you are not going to check whole table - otherwise you are going to scan whole table anyway if your DATE field in not in index.

If you know from either previous run or from any other source an interval of PK you are interested in - then splitting and selecting by intervals will help. You can even select different intervals in parallel - I believe you can find some examples on SDN on how to do it.

[joo_mariano2]

i resolved my problem.

Thank's.

In standard exists a program where you pass the object name and the number of intervals and he create intervals in another table. For this, i needed to create a function module with logic of interval (table and field names) and associate function to event in transaction FQEVENTS.