Database restructuring

Fellows Informixers, We are currently in the process of restructuring our database. The things that we are trying to accomplish are as follows: 1) move "hot" tables in there own dbspaces -> better disk space management. 2) move all Blob chunks up top with respect to chunk numbers -> minimize the time that users cannot access the Blobs. 3) place "hot" tables on their own disks -> minimize disk contention. Sounds routine right? Well, lets throw in a 30 Gig database and time constraints and see what happens. In this database, there are 27 tables. The biggest table is 22 Gigs in size, with about 2 million rows. Each row has an associated Blob located in its own Blobspace averaging 16k. On a small table, what we would usually do is a dbexport/dbimport. On a medium table, we would do a tbunload/tbload. In our case, we cannot do a tbload on tables greater than 256,000 rows for the following reasons: 1) tbload does row-level locking when it creates the tables. 2) tbload locks each Blob pages when it loads. Because of the above "feature", and with OnLine lock limit of 256,000 rows, we have the following options: 1) perform an isql unload/load -> way too slow: ~60Mb/hr 2) dbexport/dbimport -> basically same as an isql unload/load 3) write an esql/c program that would select from system1:database1.table1 and insert into system2:database2.table2 4) perform an isql select * from system1:database1:table1 insert into system2:database2:table2 -> 500Mb/hr 5) get Informix to fix tbload We have not yet established any benchmarks for step 3. We would like to accomplish the restructure process under 48 hours. QUESTION: Has anyone done anything similar? If so, what was the best approach? Does anyone have any utilities? What is the best way to do this? Thanks in advance for any advice ############################################################## David Nguyen Internet: davidn@geis.geis.com GE Information Services UUCP: uunet!ge!davidn 401 North Washington Street Voice: 1-301-340-5461 Rockville, MD 20850 USA ##############################################################