This post is more than 5 years old
207 Posts
0
400
March 18th, 2010 11:00
SRDF IBM Mainframe DB2
Hello, we will start replicating a DB2 database with SRDF/A. If the DB is 1TB what is the best way to seed it or establish it on the R2 side for the first time. Once it's established only changes will be replicated. Do we use SRDF and just wait until all the tracks get from CA to TX. Or is it possible to do something like send a backup tape to seed the database in TX and then start sending the changed tracks to TX.
This does not seem too logical to me but I am just looking for some alternative to getting the new DB to TX the first time.
No Events found!
Allen Ward
6 Operator
•
2.1K Posts
0
March 18th, 2010 12:00
So, this really is as simple as setting it up and letting the initial sync go. The only real difference is that you will not start out in async mode. You would be in Adaptive Copy with Disk mode initially until most of the tracks are over. This mode doesn't give you consistency on the R2s, but you won't have that until the first sync completes anyway, and acp_disk mode will get the data there faster.
There is no good way to estimate how long this initial sync will take, but it is dependant on your rate of change on the R1 devs. Also on the link you have running between sites. Your EMC CE might be able to get someone to look at all the numbers and give you a (very) rough estimate.
Just as a side note, even if you did try to "preload" some of the data on the R2 devices, the first time you established the link it would ignore everything there and refresh from the R1s. So you wouldn't gain anything. You just establish in acp_disk mode, and when your invalid tracks get low enough you change mode to async to finish up.
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
1
March 18th, 2010 12:00
"silvering" with tape will not work as DMX will not know which tracks are owned to the R2 side. When we started replicating our 5TB Oracle DB, we put it in acp_disk mode and let it run at its pace. When it gets close we change back to SRDF/S (that's what we use).