NonStop SQL DDL Replicator User's Guide (Update 7)
Table Of Contents
- Legal Notices
- 1 Introducing SDR
- 2 Installing SDR
- 3 Configuring SDR
- 4 SDR Operations
- 5 SDR Monitoring and Control
- 6 SDR Commands
- A SQL DDL Statements
- B SDR EMS Messages
- C Testing SDR
- Index

SDR Operations
HP NonStop SQL DDL Replicator User’s Guide—545799-007
4-6
Special cases
Because there are additional files involved, there are potential issues in replicating 
CREATE COLLATION relating to the use of these files.
The collation compiler input file
Since edit files are not replicated by RDF, you must ensure that the same character 
processing rules are present on both systems before executing the CREATE 
COLLATION on the primary system.
If the input edit file does not exist on the backup system, SDR will continue to retry 
the CREATE COLLATION until the input file is created or until you cancel the 
operation. 
Not only must the input edit file exist on the backup system, but it must also be in 
sync with the input file on the primary. Problems will occur if the primary input file is 
updated and an out-of-date file still exists on the backup.
The audited collation output file
The audited updates that the collation compiler makes to its output file cause RDF to 
stall, because the corresponding audited compiler output file does not exist on the 
backup system.
Without SDR, the operator has to create a collation manually to allow RDF to continue 
processing.
With SDR, the operator no longer needs to intervene. SDR detects the absence of the 
audited file and inserts a CREATE COLLATION into the depot file, before the collation 
compiler has a chance to generate its audited updates. Thus, CREATE COLLATION is 
replicated before RDF encounters the compiler generated audited updates.
User Transactions
Normally, SQL creates and commits a transaction for each DDL operation. No user 
transaction management is required. But, you can execute DDL in a transaction that 
you create and commit. This leaves SDR with a dilemma, as it does not know the final 
status of the user transaction when it is time to replicate the DDL.
Note. If you have automated the replication of non database files between your primary and 
backup systems, be sure to include the collation compiler input edit file into your automatically 
replicated file sets.
Caution. The downside of this strategy is that the CREATE COLLATION operation could fail 
on the primary system, for example due to syntax error in the compiler input, but succeed on 
the backup system. This can be avoided by making sure that both systems have the same 
input edit file.










