Kbase P38754: Binary dump is much slower than a probkup.
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
8/25/2003 |
|
Status: Unverified
SYMPTOM(s):
Binary dump is much slower than a probkup.
CAUSE:
The reason why a probkup is quicker than a binary dump is because a probkup is backing up entire blocks rather than individual records. e.g. 8Kb blocks at a time. The binary dump would have to read the index block, read the data block, extract the record, then append it to a file. I/O calls are expensive.
Also, because records for the same table aren't likely to be stored sequentially and would be scattered across the entire database, the same blocks are likely to be read from disk multiple times during the dump process. e.g. Record 1 is likely to be stored in a different block than Record 2, but Record 100,000 might be stored in the same block as Record 1. However by the time it needs Record 100,000 the block probably isn't in memory so it must read it from disk again.
FIX:
A comparison between probkup and a binary dump isn't very appropriate.
A binary dump will dump records from within blocks, doesn't process database blocks sequentially, and has the potential of reading the same blocks multiple times.
A probkup will process the entire block, will work sequentially through the database, and will only read each block once.