| Sorry for being late with this, but it might still help:
1. I don't have my manual at home to check this, but Sybase claims that
you can. I don't have an answer at this point.
2. Sybase still depends on a user-written coordinator. They have been
promising this for a long time, but still no results.
3. They are talking about the use of Triggers and the local security
provided by the Sybase Server. The Triggers are used to do data
integrity checking by rejecting transactions that don't match the
business rules. They can also be used to enforce other business rules
(such as auto-delete of detail records when a master records is
deleted). Rdb's use of Constraints and Triggers make this much easier
to do.
4. I thought they did table file mirroring, which involves opening the
same table file on two different disks and duplicating all writes to the
second file. Obviously, this puts a somewhat higher strain on the CPU
than the use of VMS based disk shadowing, because of the need to
redundantly issue two QIOs to perform the writes.
Transaction logging requires a second 'hot' server which applies all of
the transactions from the first system on the second system. Double
your CPU and disk to get less performance.
5. Sybase can add and delete indexes, but locks up the table to do so.
To add or delete a column they need to copy the table to a temporary
table (which has had the change made to it) and then delete the
original table and copying the updated table back to it. This happens
more-or-less online, providing no-one wants to use the table while you
are making the changes. Updating of the index statistics table
requires exclusive access to the table.
Tables can be create/deleted while the database is operating. Sybase
also claims to be able to do online backup, while the database is
operating. Any changes made to pages that have been backed up are
automatically appended to the backup file (so they claim). I am not
quite sure how you would get one of their online backups to stop
backing things up.
6. Sybase does have multi-file tables. A table can start in one table
file and overflow into succeeding files. They can also place an index
for a table into a separate file (called a Device in Sybase terms).
They can also cluster a table around an index but they must be on the
same device.
Note that a lot of this became available in Sybase V4.0
Hope this helps.
Neil
They do not have the same level of control over placement that Rdb has.
|