TurboDB Engine Documentation

Optimization

Previous  Top  Next


When you have the feeling that your TurboDB application is slower than it should be, there are many possible reasons. Check the following questions and follow the appropriate instructions.

One or more select statements on a large table or on a set of tables takes too long to execute

There are basically two ways to speed a query: Create the necessary indexes and/or optimize the statement.

Setting a filter on a table component (VCL library) takes too long

Adding an index to your table may also help in this case. Another way is to use the range feature instead of the filter.

In file access mode local operation is quite fast, but as soon as a second application just connects to the database, everything slows down

The first thing to check in this case, is whether your network has problems. Because file access mode makes use of network functionality very heavily, it often happens that poor network performance was not noticed before the TurboDB application was installed.

The network is ok, but when a lot of people edit in the database, completing an operation takes very long

With a lot of people working concurrently on the database, the locking overhead grows and a lot of waiting for access to the database occurs. The first step in this case is to use explicit locking to form bigger database operations and thus less locking overhead. If this won't help enough, you can still use TurboDB Server.

In addition to those specific hints, there some general hints that can help increase database performance:

Set the flush mode to fast

The flush mode determines the buffering degree within the database. Setting it to fast will maximize the internal buffering and therefore increase performance. However, if the application crashes, data loss can occur in this mode.

Use exclusive access if possible

In case your application is conceived for a single-user only. You should put the database in exclusive mode to eliminate multi-user access overhead.