Effect of indexes upon performance

[PatriciaW]

When I worked with other DBMSs, I did not create indexes on fields where the cardinality (# of distinct values) was low. So for example, if there was a field with two possible values (such as yes/no) I would not have built an index because it would not improve performance and would in fact damage performance because of the time required to build, maintain this index. This was related to the "page size" used. is there an equivalent concept in FM?

I have two questions:

1) is this a valid concern for FM7?

2) are the indexes built when the file is loaded (opened) or are they stored as part of the file?

[PatriciaW]

When I worked with other DBMSs, I did not create indexes on fields where the cardinality (# of distinct values) was low. So for example, if there was a field with two possible values (such as yes/no) I would not have built an index because it would not improve performance and would in fact damage performance because of the time required to build, maintain this index. This was related to the "page size" used. is there an equivalent concept in FM?

I have two questions:

1) is this a valid concern for FM7?

2) are the indexes built when the file is loaded (opened) or are they stored as part of the file?

[PatriciaW]

When I worked with other DBMSs, I did not create indexes on fields where the cardinality (# of distinct values) was low. So for example, if there was a field with two possible values (such as yes/no) I would not have built an index because it would not improve performance and would in fact damage performance because of the time required to build, maintain this index. This was related to the "page size" used. is there an equivalent concept in FM?

I have two questions:

1) is this a valid concern for FM7?

2) are the indexes built when the file is loaded (opened) or are they stored as part of the file?

[Brian C]

Indexes are stored as part of the file and do take up additional space.

So far as I know v6 and v7 both benefit greatly from indexing regardless of the cardinality of the data. The only drawback being the additional diskspace used.

I am not sure if there is any real need to rebuild indexes, but they can be rebuilt manually by turning off indexing for a specific a field, closing the file (to force a removal of unused blocks) and then reopening the file and turning indexing back on. The index will then be rebuilt as soon as you exit the define fields window.