summary: individual cahced table data files --> individual cached table data files
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2003-05-16
Logged In: YES
user_id=199381
I agree. Would allow much more efficient operation
under cases like:
some really big read-only CACHED tables, some
collections of CACHED tables that are updated only
occasionally, some collections of CACHED tables that
are updated frequently.
A one-one mapping is a bit limiting, however. Better to
allow DDL to assign tables to or transfer them between
table spaces (cache files or collections of them)
We'd be better off redesigning the binary file format to
get close to what innoBD and Oracle do first, however,
as that is, first and formost, where many of the
performance limitations lie for CACHED table
persistence.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2003-05-17
Logged In: YES
user_id=199381
BTW:
You can get much the same effect with TEXT tables,
excpet that TEXT table indexes are built and maintained
100% in-memory, so they do impose a limitation that
CACHED tables do not (CACHED table index structures
are persisted on disk as well as their row data).
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
It would be very good to configure separate files for CACHED
tables. But data and indexes could go into same file.
Now, it would be possible to transfer table from one machine
to another installation by replacing the file only (+copy
rowindicator from top of .script file). Currently, cached tables
cannot be moved with copypaste.
Regular tabledata is possible to copypaste from .script file.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Logged In: YES
user_id=199381
I agree. Would allow much more efficient operation
under cases like:
some really big read-only CACHED tables, some
collections of CACHED tables that are updated only
occasionally, some collections of CACHED tables that
are updated frequently.
A one-one mapping is a bit limiting, however. Better to
allow DDL to assign tables to or transfer them between
table spaces (cache files or collections of them)
We'd be better off redesigning the binary file format to
get close to what innoBD and Oracle do first, however,
as that is, first and formost, where many of the
performance limitations lie for CACHED table
persistence.
Logged In: YES
user_id=199381
BTW:
You can get much the same effect with TEXT tables,
excpet that TEXT table indexes are built and maintained
100% in-memory, so they do impose a limitation that
CACHED tables do not (CACHED table index structures
are persisted on disk as well as their row data).
Logged In: YES
user_id=367429
It would be very good to configure separate files for CACHED
tables. But data and indexes could go into same file.
Now, it would be possible to transfer table from one machine
to another installation by replacing the file only (+copy
rowindicator from top of .script file). Currently, cached tables
cannot be moved with copypaste.
Regular tabledata is possible to copypaste from .script file.
Logged In: YES
user_id=82101
see, this is an example of how ideas just pop up which this
would help.