My Two Best Lectures at TechEd So Far...

First I have to say that San Diego is GORGEOUS! Perfect weather and amazing landscapes all in a big city that doesn't feel TOO big. Now on the geeks stuff...

SAP Solution Manager: New Features

Drop the Notes Transport Database and walk away slowly. SM 3.2 gives you all of that and more. Very slick interrogation of the whole transport management/development process and the actually propagation of the transports. After my other hands on class experience with SM and upgrade/implementation project management, I'm looking forward to getting SM going.

Higher Performance with Archiving

A good discussion of how table growth and size can affect performance and why. Examples of chronological data (sales orders and invoices) vs non chronological data (material master) were given. In chronological data, storage size really won't impact performance since newer items are always searched first. The material master doesn't have the a dating luxury so the larger it is the worse performance is when the transaction relies on tables like MSEG.

The point is that focusing your archiving efforts on large tables will not automatically give you a performance boost. You are definitely getting disk space with those tables but to gain performance you'll have to dig into your tables without dating info and find a way to determine what should be swept away. More difficult since there is really no retention period time frame to go by.

Another great point of the class focused on indexes. As your database grows so do your indexes. In many cases, the indexes tied to a large table can surpass the actually data size of the table. Without archiving you have a huge problem of indexes multiplying your actually data growth.

One last final point was access and process time of retrieving archived data. Your intuition probably makes you think that archived data retrieval is always slower than online data lookup. This is definitely not the case. If your archive still resides on your file system then getting to it is much faster for the end user than if it was still in your database. Since it's archived, you'll have a reference of exactly where that record is in the archive. The data online still has to be retrieved by the normal multiple table scans. So even if you leave the archive data on your file system and don't reap the rewards of reduced disk usage you still win with increased performance and a smaller R/3 database. Performance gains are seen from less dialog steps being spent gathering those requests for archived data. A smaller R/3 database wins from reduced daily backup times.

And the winner is----

The cool archiving class. I'm sure you could have guessed by the write size disparity, hmmm. =)