I am relatively new to Netbackup as I've only used BEWS in the past. I'm being taught by one of the other techs where I work. He indicated that about once a month I need to eject the partially written to tapes because NetBackup will not use them for new backups and jobs will fail if there are not enough tapes in the robot that don't have no time assigned to them.
This is not only causing us more time to administer the changing out of the tapes, but it is costing us money because we don't have as many tapes in rotation and we have to keep buying new ones because the partials are being stored offsite with those tapes that are written to capacity.
I had conflicting information from my boss who said that is not a design flaw in Netbackup but a configuration inefficiency and to ask Symantec about it -- which is what I'm trying to do here.
We are using Netbackup for Windows version 7.1 with LTO4 and 5 drives and tapes.
Thanks for any input on this question.
My experience has been exactly the opposite - NBU will first use already written tapes that are assigned to the requesting media server, is in the required pool and the same retention period, not Full, Frozen, Suspended, etc. Only then will it take from scratch.
Please see http://www.symantec.com/docs/HOWTO32854
The problem with partially full tapes is that you don't know exactly how much space is left on them. If you don't have enough slots in the robot for the partially full media as well as spare scratch tapes, you could end up with status 96.
In our enviroment we are using LTO3 tapes where the full media size varies from 390GB to 890GB. Am not able to find the reason. Is it because of compression technology.
Not the compression technology but the inherent compression possible with different types of data. Database files for instance will compress well, files like jpg's wont. Others might disagree but I also believe the transfer speed you get impacts the amount you can place on a tape.
Agree with Marianne, if a tape is not full NetBackup will use it. Only issue that you might face then is the expiration times are different i.e. if you have weekly backups that you retain for 4 weeks, the first weekends weekly goes on to the tape, now the expiration is 4 weeks from now, if you leave the media in the robot and the weekly runs next week, then your expiration is another 4 weeks from now. By ejecting the media after the first weekly backup you know you'll have media expiring in 4 weeks. You can imagine how this could cascade if you keep it in there for a while. NetBackup will only use a tape once the last image has expired, even if that is just 1Gb written and the very end and the rest is "open".
You didnt mention if you're using all those drives on 1 server. If you are then you'll need quite a few fibre cards (2 or 3). If you dont have that many you don't really have enough bandwidth to run the drives simultaneously at maximum speed. The same also applies for the Network bandwidth, the theoretical limit of a 1gbps card is around 100MBps which means its less than what the tape drives can do and its therefore a bottle neck. In either of these situations you're using more drives than you can effectively stream to, which means you're using more tapes than you should (that get filled up partially)
I have found that if NB is in the middle of a backup "run" & the current media becomes FULL it will not use partially full media to complete this backup but rather take a "new" tape from the appropriate volume pool or scratch if none are available. It will start a batch of scheduled jobs using partially full media, but from that point until the "run" completes it always utilises unused tapes to complete the backups.
Previously, in our situation, for certain pools there should've been a maximum of 2 partially full media in our library at any one time (due to number of jobs that were active at any one time). For some reason in the past this could creep up to 4 or 5 say. As NB utilises those partially full media in order of those that were used less recently, we found that to get the number of partials back down to the "expected" 2 then we had to eject enough media to get it back down to that level - this was media that should've remained in the library, there's not a lot you can do about those that have to go off-site on a regular basis, you can't really leave them until they are full! It wasn't a big deal, just me being tidy (or OCD).
Just my 2p.
The reason it looks like transfer speed affects how much data you can fit on a tape, is because the better the compression you get also means you get higher transfer speeds. If the drive is writting fewer 1's and 0's to the tape, because of high compression, it can empty it's buffer and ingest more new data faster.
You can set a limit on the number of partially full media on a per Volume Pool basis. Just make sure you leave it high enough that you don't limit your overall throughput/performance.
What I've seen happen sometimes, when a job needs a tape, instead of waiting for a partially full tape, even if it's being rewound and ejected in the same drive it'll be using, it will put a reserve on a new tape and put it in the next free drive as soon as it's available. Setting this limit will cause NetBackup to instead wait for the "in-use" tape before pulling an un-used tape.
I like to set it to the same number of drives as are in my library, and enable media sharing between media servers to make the best use of my tapes. Check your NetBackup Documentation for more details, and be sure to consider your overall environment when picking the limit to use.
I don't however recommend allowing multiple retention on the same tape. It may seem helpful up front, but you end up with tapes that has a few longer retention backup image left on it long after the short retention images have all expired, that otherwise would have been available for re-use much sooner; causing you to need more tapes in the long run. Of course if your retention periods are relatively close together, this might be a good trade off for improved performance, but then you might also be better off slightly increasing the retention of some of your backups.
Keep in mind that if you do send multiple retentions to the same Volume Pool, setting the limit applies on each Volume Pool and Retention combination. If you have multiple media servers without media sharing, that too will cause you to have additional partially used tapes. In other words if you have a 2 week and 4 week retention in the same Volume Pool, with a limit of 2 partial full tapes, you'll end up with 4 partially full tapes, 2 at 2 weeks retention and 2 at 4 weeks retention (if you have two media servers you can end up with 8 partially full media).