Ay, I think you might have to reconsider your whole situation!
Backupping with multiple streams to a single disk volume, although supported is a bad practice in reality. This causes a great deal of fragmentation and will slow down the fastest array to a crawl because of the random read nature (and write) after a while. Special pre-cautions can be taken using settings for the stripe size, format cluster size, buffer number and sizes, etc. But in the end, it will eat your performance away because of how NTFS works and how disks work.
But a bigger problem even, is that the situation you have built is *not* supported and not really useful. You do not backup to disk and then backup that folder again to tape. You will get into trouble if you actually have to restore something because Netbackup will have no idea what is going on.
Sorry to be so hard on you. It's a common mistake, if you’re not really into it, not your fault.
Before testing clients and bpbkar, etc. (which is good practice) you'll need to fix your server side first.
From what I understand, you are trying to use the disks to accelerate your backups. Contrary to what you may have been told, this is not always possible and disk can actually slow down your backups considerably, just like is happening with you now. Only a highly tuned situation can get you a faster backup and completion overall.
Best is to figure out what your fast clients are and which are your slow clients. You can do this by backing them up to disk or tape using a single stream in total. Then you get good KB/sec values in your admin console and judging from that, we can go on.
Slow clients will need to go through disk not to wreck the tape drive, fast clients can go directly to tape with some multiplexing on top of it to get everything going at a good pace. In your information you do not mention if you need duplicate tapes or disk restore ability, so I will assume you do not. If for instance disk restore ability is required, you can add those clients to the “going through disk” selection. But a period to keep on disk cannot be set. This requires a different setup.
Set up your tape drive correctly (please read performance guide very carefully, un-tuned or tuned can give you double if not TRIPLE the performance). And don’t set your multiplex astronomically high, start with 8 and check your BPTM logs to see if you need more or not. Also, setup monitoring of your network cards (NetMeter is a good program to visualize traffic) and performance monitor looking at your tape drives and disk I/O, queue length, etc. to find your bottleneck. Right now, during the backup you will probably see a very high queue length, and during the backup to take probably a queue length of 1. ;)
The disk part takes some work too. Talk with your storage guy on how disks, your array, bandwith, stripe size and clusters work. In the end you’ll probably end up with a few DSSU’s (Disk Staging Storage units) which you can use for your smaller clients. Then, an automatic schedule will take care of duplicating these to tape and freeing space when needed on the disk again.
Ok, I’m leaving it at this for now. If you have any more questions, feel free to ask. I will not go into any specific values or such, because well, then you’d have to hire our firm. But I am certainly willing to help and get you on your way!
