Re: Logical log size

Hi Carlson: I have done this kind of updates some times, and in general the first threshold that I reach is the number of LOCKs, not the LOG size. Usually, specially if you are planning to update an indexed column, you will encounter that unless you have an instance with a large number of locks and little log space, locks will be the limiting factor. Tino > Carlson@WHSmith wrote: > > > > I looked all around and couldn't find any info, so, here goes . . . > > > > How much logical log space is used on an update? I will be updating two > > columns (smallint) of each record on various tables in my system, along > > the lines of . . . > > > > update table_name > > set upd_a = new_value_a, > > upd_b = new_value_b > > where upd_a = old_value_a > > and upd_b = old_value_b > > > > The two ideas I have now would be to : > > 1. select subset of table > > begin work > > update subset of table > > commit work > > repeat until no more subsets. > > > > or > > 2. select all rows from table > > begin work > > update > > commit work > > select rows from next table.> > > > > > The problem I have is one of granularity. I'd like to be able to update > > an entire table on a transaction, but I'd like to plan ahead to see if > > it is even possible. Is there any way to determine how much logical log > > space a two-column update would take?? > > > > TIA > > > > John Carlson > > Informix DBA > > WH Smith, Inc. > > > > Last seen: Trying to get all projects caught up after the Conference.