cancel
Showing results for 
Search instead for 
Did you mean: 

Multiplexing

OpsAnalyst
Level 4

At some of my sites we are still on Netbackup 5 and I was wondering how best to handle multiplexing.  If a site has two drives (LTO2) and we backup about 100 jobs a night what should the media multiplexing be set at?

1 ACCEPTED SOLUTION

Accepted Solutions

Mark_Solutions
Level 6
Partner Accredited Certified

If you were already on 10 and have reduced it to 6 this may have had an impact

Every system has a bottle neck some where and i appeciate you restrictions but if you have jobs running for that long then the system is seriously lacking

A good first step would to be to introduce disk storage where far more jobs can be run at the same time but it also sounds like you may need an additional Media Server and more tape drives to reallytackle your situation.

You policies need to be examined in details to optimise the multi-streaming in the best way possible

If you are stuck with the hardware that you have then your only option in this situation is to increase multiplexing but i would not take it to the limit and you have to be aware of the effect that this will have on restores

Have you double checked that all schedules have multiplexing set?

I hate to say it but in you position, and ahead of your upgrade which hopefull involves more resources, then go for a level of 12 for now as i dont see any other options

If you need advice / planning etc. there are a lot of us here that are Symantec Partners that offer those services so some consultancy may well help you out:

http://partnerlocator.symantec.com/public/search;locale=/

View solution in original post

9 REPLIES 9

Mark_Solutions
Level 6
Partner Accredited Certified

Multiplexing can go as high as 32 but it has an impact on restores

A good level to use is 6 but rememebr to also set this in your schedules to make it work and if any clients run mutiple streams then make sure the maximum jobs per client in the Master Servers Host Properties is high enough

Other than that you really need to upgrade as you are running unsupported software

teiva-boy
Level 6

There is no set number to use for multiplexing.  4, 6, 10, I've seen a bunch.  It all depends on your environment and what it can support.  

That said, LTO2's speed is 15MB/s minimum, upwards of 35MB (More if you are getting compression from the drive upwards of 50MB/s) at its top speed.  The main thing is to make sure you hit that minimum speed I listed, and tune accordingly to hit that max speed value.

This gets you the most from your drives, prevents drive wear and tape wear, and ultimately the fastest backup speeds.  

Of course multiplexing has its downsides.  As you up the value to 2, 4, or 32, you are interleaving backups up servers and volumes on to the tape.  When you need to restore, the tape has to read and skip over backup sets to get to the data needed, thus increasing your restore times.

Remember we backup to restore, not backup to save time!  So balance multiplexing with your RTO.

 

OpsAnalyst
Level 4

Thanks for the input so far. I was also wondering what the relationship is between multiplexing and Max Fragmentation?  Is there a best practice concerning Max Frag?

teiva-boy
Level 6

Fragment size technote:

http://www.symantec.com/business/support/index?page=content&id=TECH15692

Tuning involves, change ONE setting, test, change ONE setting, test.  If there are no positive gains in the last test, go back one setting and test again.  No point being on the ragged edge...  Start with multiplexing first before playing with other values.

And as always, you need to test restores too after you make changes.  In some tape techologies when you change the tape block size it could make restores not work.  So you would test not only backup performance, but restore performance too.

Mark_Solutions
Level 6
Partner Accredited Certified

As a rule we always use 5000MB as the fragment size

It gives good performance and fast file location during restores

If using multiplexing it will produce its own fragments anyway unless only one job is running

Has your initial question now been answered? If so dont forget to close the thread by marking the answer that helped you most to assist others in the future when searching the threads

OpsAnalyst
Level 4

I guess I should explain my prob a little more in depth.  We have certain backups that run for more than 24 hours sometimes because they are so large and our current Netbackup resources are so limited.  We don't like them to run that long obviously so we had tried to up the multiplexing, but then it started using a lot more tapes.  We don't like that either. So now we are trying to find the best balance for the next few months until we get our system upgraded.  Last night I reduced the multiplexing on all schedules in all policies to 6 from 10.  I upped the Max Jobs Per Client setting from 2 to 6. And I changed the Max Fragment Size from 1TB to 2GB.  The jobs are still running, which is already about an hour and half longer than the previous few days.  And it really didn't reduce the tape amount either.  So basically the changes I made had a negative impact on performance (but a livable one for now) and no impact on tape usage.  Are there any suggestions based on this new info?

Mark_Solutions
Level 6
Partner Accredited Certified

If you were already on 10 and have reduced it to 6 this may have had an impact

Every system has a bottle neck some where and i appeciate you restrictions but if you have jobs running for that long then the system is seriously lacking

A good first step would to be to introduce disk storage where far more jobs can be run at the same time but it also sounds like you may need an additional Media Server and more tape drives to reallytackle your situation.

You policies need to be examined in details to optimise the multi-streaming in the best way possible

If you are stuck with the hardware that you have then your only option in this situation is to increase multiplexing but i would not take it to the limit and you have to be aware of the effect that this will have on restores

Have you double checked that all schedules have multiplexing set?

I hate to say it but in you position, and ahead of your upgrade which hopefull involves more resources, then go for a level of 12 for now as i dont see any other options

If you need advice / planning etc. there are a lot of us here that are Symantec Partners that offer those services so some consultancy may well help you out:

http://partnerlocator.symantec.com/public/search;locale=/

OpsAnalyst
Level 4

Our weekly fulls will run tonight so i am going to make some changes based on your suggestions and hope for the best.  I'm going to increase multiplexing to 12 on each schedule in each policy.  I am also increasing the Maximum Fragment Size to 5GB and the Max Jobs Per Client to 12 and the Max Multiplexing Per Drive to 12.  Based on what I've heard here I'm sure this will help on speed...I just hope it doesn't double the tape usage. :)

Just so you all know we are running Netbackup 6.5 at our main data center (about to upgrade to 7)  These older systems are at our remote sites.  I appreciate all of your help on streamlining these ancient systems.  My predecessors really dragged their butts on getting these systems brought out of the stone age.  I am working on it, but it's a big organization and nothing here happens overnight, so bare with me if you all see more questions or thoughts on streamlining my backups on these.  Thanks.

Mark_Solutions
Level 6
Partner Accredited Certified

OK - glad this has all helped and dont forget if you need someone to come and help / re-design / implement everything to get in touch or find a partner near to you (cant see from your profile where you are) and do your best to get some disk in the system - you will be amazed the difference it will make

It is not as expensive as it seems - you can get 180TB in a 4U array for less than 50,000 GBP these days

It can work out cheaper than buying tape drives, new libraries and all the licenses that go with them.