Smart Blob Bottleneck?

IDS 11.50.UC5 Linux I currently have two smartblob spaces each consisting of 5 100-GB chunks to which I am taking data currently in a blob space and pushing it into the smart blob space -- we want to establish HDR Replication on the instance in the future. I have established 21 cpus, affinitied and set to no age. I have 1.25 million buffers available. RTO_SERVER_RESTART and AUTO_CKPTS are managing the resources, with RTO set to 60 seconds. I created a new table without indices, figured it was best to add the indices later. I started with 5 insert_into_select_from processes and was only able to achieve around 50,000 inserts into the new table over a one-hour period. I added 5 more processes; over the next hour, only 54,000 additional rows were inserted. The instance has plenty of buffers and is not showing any foreground writes; the buffers are getting quite dirty but the LRUs (128) are doing their job. I am beginning to think that perhaps the bottleneck in the process is just the two smart blob spaces (round-robin inserts via PUT statement). Before dropping the 8 extra chunks and recreating them as their own smart blob spaces, I thought I would ask if there was a way to determine if the smart blob spaces are indeed a possible bottleneck ... and if anyone thought the additional smartblob spaces (also inserted via round robin via PUT statements) may show an increase. Here is the onstat -p: Profile dskreads pagreads bufreads %cached dskwrits pagwrits bufwrits %cached 2924885 2325491 19153481 84.73 3456398 3644417 14279549 75.79 isamtot open start read write rewrite delete commit rollbk 14806599 1535464 10281 68223 128067 46 525 101 0 gp_read gp_write gp_rewrt gp_del gp_alloc gp_free gp_curs 1020452 255070 4897834 1024 1 0 33 ovlock ovuserthread ovbuff usercpu syscpu numckpts flushes 0 0 0 243.46 396.65 4 4 bufwaits lokwaits lockreqs deadlks dltouts ckpwaits compress seqscans 222835 0 4285658 0 0 11 21 344 ixda-RA idx-RA da-RA RA-pgsused lchwaits 146 0 357062 357094 61243 Here is the top of the onstat -P: Buffer pool page size: 2048 partnum total btree data other dirty 0 1214227 0 1214226 1 222941 I wish appreciate any insight into this ... especially from those that have dealt with this type of "conversion". Thanks in advance. Clifton _________________________________________________________________ Hotmail: Powerful Free email with security by Microsoft. http://clk.atdmt.com/GBL/go/196390710/direct/01/