We need to store a 250K voice blob but the client will be sending the blob
in separate transactions in 50K segments. We'd like to
use locators to append to the blob as follows:
1. Trns A sends 1st segment
- Insert buffer data into db table
2. Trns B sends 2nd segment
- Retrieve lob locator for blob in db table (1st
segment) - LOCA
- Use Values clause to concat LOCA with buffer of 2nd
segment
to create lob locator - LOCB
- Update db table with LOCB
3. Trns C sends 3rd segment
- Same as #2 but LOCA now represents the 1st and 2nd
segment and
LOCA will be concatenated with the buffer of the 3rd
segment.
First, would this work? If yes, what is the storage/performance
implication?
Do blobs work like varchar where every time you increase the size there is
a potential of creating an overflow record? We are NOT using
COMPACT. The manual says that when a LOB is inserted some freespace is
inserted to assist in future appends. How much freespace, 200k worth?
I'm concerned about my voice blob being fragmented. The other possibility
is to insert the segments into a work table and once the last segment is
received
use locators to contanate the work table blobs (5 rows 50K each) into one
row (250K blob) in the master table and cleanup the work table.
Any advice would be greatly appreciated.
Mary Macatee
IBM 60 West Street Suite 400
Annapolis, MD 21401
Telephone: 410-267-9256 x128 Fax: 410-267-8254
email: [EMAIL PROTECTED]
=====
To unsubscribe, send 'unsubscribe' to [EMAIL PROTECTED]
For other info (and scripts), see http://people.mn.mediaone.net/scottrmcleod
