Our NW server is a Sun X4440,32g, 4 cores with 2xJ4400 (48-1T SATA "SATAVOL") and 2xJ4400 (48-300G SAS "SASVOL") with a StorageTek SL500 with 8 LTO-4 drives, 10Gb fiber connected. SATAVOL and SASVOL are zpools.
2 AFTDs on the "SATAVOL" - SATAVOL/ebs* - all fulls go to here
2 AFTDs on the "SASVOL" - SASVOL/ebs* - all non-fulls go here
We are beginning to migrate from using a Linux clustered NFS solution (backed up using NW Linux client) to an Oracle 7000. Our plan was to use 3-way NDMP. There is approximately 15.35T on disk compressed down to 4.57T on the Oracle NAS. I added 4 more AFTD's (2ea on the zpools) and created 2 new pools for non-full and full. Running an NDMP schedule with just one dataset is working fine. But when I tried to add the other 15 this weekend it has slowed down considerably. I'm looking for thoughts and solutions. I can read and write from the zpools at 250MB/s simulaneously. So I don't really think that is my issue. I can also stream from the pools to tape at 250MB/s. Where can I tweak things to be able to back this thing up in 3 days or less?