Hello everyone!
I'm hoping that there are other shops out there using Replicate 5.5. with DB2z V11 as the source who can help us solve our throughput issues. Here's some background:
When we first started using Replicate (older version 5.3.3.45 MVS 32-bit) it was remarkably slow with horrible throughput ranging from a couple hundred records per second to maybe 3000 records per second. On million and billion row tables the bulk load process took forever to say the least. This older version of Replicate connected to DB2z natively using a daemon on the mainframe. To speed up the process, our Architect figured out that we could bind certain packages that replicate was using with DEGREE ANY (turning on parallelism) and our throughput jumped not only into the tens of thousands but over 100 thousand records per second at times.
Fast forward to earlier this year when we installed Replicate 5.5 and connected via ODBC. It's like we've started over again in the throughput department. On a "good" load we maybe get 2100 records per second. Sometimes we don't even get 1000 records per second! We did some research to see if we could turn parallelism on and discovered that the following packages are being used with this new version:
SYSLH200
SYSSH100
SYSSH200
SYSSTAT
We bound those packages with DEGREE ANY but this time our throughput has not increased. Here are additional details of our environment:
Replicate: V5.5.0.345, 363_r4db2
ODBC: 11.1.0.1527
Source: DB2z V11 non data sharing, Z/os V2.2
Target: Teradata 15.10
Does anyone know how and what we need to configure so that we can get back to throughput of at least tens of thousands of records per second? Any ideas are greatly appreciated.
Thanks!
Candace
I'm hoping that there are other shops out there using Replicate 5.5. with DB2z V11 as the source who can help us solve our throughput issues. Here's some background:
When we first started using Replicate (older version 5.3.3.45 MVS 32-bit) it was remarkably slow with horrible throughput ranging from a couple hundred records per second to maybe 3000 records per second. On million and billion row tables the bulk load process took forever to say the least. This older version of Replicate connected to DB2z natively using a daemon on the mainframe. To speed up the process, our Architect figured out that we could bind certain packages that replicate was using with DEGREE ANY (turning on parallelism) and our throughput jumped not only into the tens of thousands but over 100 thousand records per second at times.
Fast forward to earlier this year when we installed Replicate 5.5 and connected via ODBC. It's like we've started over again in the throughput department. On a "good" load we maybe get 2100 records per second. Sometimes we don't even get 1000 records per second! We did some research to see if we could turn parallelism on and discovered that the following packages are being used with this new version:
SYSLH200
SYSSH100
SYSSH200
SYSSTAT
We bound those packages with DEGREE ANY but this time our throughput has not increased. Here are additional details of our environment:
Replicate: V5.5.0.345, 363_r4db2
ODBC: 11.1.0.1527
Source: DB2z V11 non data sharing, Z/os V2.2
Target: Teradata 15.10
Does anyone know how and what we need to configure so that we can get back to throughput of at least tens of thousands of records per second? Any ideas are greatly appreciated.
Thanks!
Candace