Tape usage extremely high

[OpsAnalyst]

We just upgraded our entire Netbackup system to 7.5.0.4 and we feel we are using far too many tapes.  We have four dedupe appliances and they are all duplicating to one LTO5 tape library with twelve drives.  We have each of the appliances set to use three tape drives.  We sometimes kick out as many as twenty tapes a night and most of them don't even have 1 TB on them...they can hold up to 3.  The setting for partially full media is set at 0, but our vendor suggested setting it to 12 to match the number of tape drives.  Well that made us kick out more than thirty tapes, so we set it back to 0. This entire setup was recommended to us by our vendor but I'm not very trusting of them now. Any ideas?

[Mark_Solutions]

The most likely reason here - no accounting for encryption being used at the client side - is that your schedules are using different retention levels which cause each retention level to use a different tape

By this i dont mean different retention period as a retention period of 12 months using retention level 3 is actually different to a retention period of 12 months using retention level 4

Likewise infinite retentions by default go from level 9 to 20 but all are different and will write to different tapes

If this happened after an upgrade then perhaps previosuly you allowed multiple retentions per media (Master Host Properties) and the upgrade has reset that value

Give a little more detail about you policies / schedules / settings to see if we can dig deeper

[Marianne]

Another reason for high media usage is lack of Media sharing. 

If you enable global media sharing, all media servers/appliances can append to the same tapes.

Without media sharing each media server requires own set of dedicated tapes.

Try to limit amount of volume pools as well.

[Nicolai]

What is the transfer rate from dedupe appliance to tape ?

If we go below the tape drive variable writing speed the tape drive will start to do "shoe-shining" (e.g. do start/stops). Each start/stop operation require a interblock gab to separate where writing ended to where writing start. That tiny space with no data may add up significant data capacity  if all the tape drive do is start/stops. 

https://en.wikipedia.org/wiki/Tape_drive (see techinical problems)

[OpsAnalyst]

Thanks, Marianne.  I actually just noticed this morning that this possibility would really help us.  Where do I change that setting. I know I've seen it before and can find it again, but I'm sure you can point me in the right direction fatser. :)

[Mark_Solutions · Accepted solution]

The setting is under Devices - Server Group

Add a Media Sharing Group

But from the number you stated it still sounds like something else may be wrong

[OpsAnalyst]

Okay I actually found the setting.  I checked the "Enable Unrestricted Media Sharing for all media servers" box on the four media servers using the new library.  I did not check it on the master server because we have other media servers and libararies that I don't want to use that setting for.  Does this sound okay to you?

[OpsAnalyst]

Actually we only use two retention periods and they are set in the SLP.  For daily diffs and weekly fulls we use 1 month and for monthly fulls we use 1 year.  Even on the nights when we have no monthlies running we are still kicking out at least ten tapes that have under 1 TB written to them.  I also wonder about changing the parameters for the SLP duplication session. I believe we chose the default which is 5 minutes. Should that timeframe be increased?

[Mark_Solutions]

I thought that setting was only available on the Master server and it allows all media servers and NDMP hosts to share media

For this to work you should not have any Media Sharing groups settup

If you want just four Media Servers to share media then you should use a media server group for media sharing and add just those four servers to the group

[OpsAnalyst]

Okay I undid that setting and set up the media sharing group for those four media servers.  Now do I need to change any of the storage unit settings?

[Mark_Solutions]

That should be all you need to do - although a fresh load of expired tapes will help as any already used may already be owned by individual servers

You can see their status in the robot section under media if you add the Media Owner and Server Group columns you will see if the tapes can be used by the group or not

Hope this helps

[Mark_Solutions]

Just noticed your other answer - so if you just have 2 SLPs (Daily and Weekly) then that is not the problem as you can only have 2 levels

What about compression and encryption?

[Marianne]

Media sharing is either configured per server group or unrestricted.

Host Properties > Master Servers -> Master -> Media

Select 'Enable Unrestricted Media Sharing for All Media Servers'.

[OpsAnalyst]

I have set up a server group. I'm wondering if it's still best to have each media server set to only use 2 or 3 maximum concurrent write drives.  And what about the maximum streams per drive setting?  And does anything need to be done at the policy level or in the SLP to direct it to use the server group?