New Media Set Date in the middle of the Backup Week

[WinstonWolf]

Hello ,

I find an strange behavior .

In my Media Pool i started only every Saturday at 00:05 an new Media Set .

An now in my Media Pool i see the following :



In the middle of the Week one Tape in my Media Set becomes an other Date . Yust for Info . This Job uses two Tapes with the Parallelprocessing Option activated .

[Dima P.]

Hello,

This Job uses two Tapes with the Parallelprocessing Option activated

When you enable parallel processing for the media pool it keep more than one media set opened to be processed simultaneously. Check the sequence number you have three tapes with the #1.

Also I strongly recommend you to keep the %ID% postfix in the media set name – it allows you to easily identify the “parts” of the media sets with parallel processing being enabled.

Below is the list of three tapes being processed simultaneously (as a result three media sets were opened).

Tape XXX11 - Media set XXXX ... ID1 - Sequence number 1
Tape XXX21 - Media set XXXX ... ID2 - Sequence number 1
Tape XXX31 - Media set XXXX ... ID3 - Sequence number 1

[WinstonWolf]

Sorry this is totally nonsense !!!!!

First : The only Point where i can define when a new Mediaset is started sholud the Option in the Media Pool under "Media Set"

Second : How should i distinguish in an MediaPool with 60 Tapes and 6 Mediasets only with the ID and not with the Date of the Week which Tapes i Need in case of Recover .

Thirt : What has the usage of one or Two Tape Drives for an Tape Job to do with the creation of an new Media Set ???????

Fourth : The Reason why i stamp an New Mediaset with the Date is to have unique Media Sets for an Week in an Media Pool . With your Solution i have many ID1s and many ID2s etc. .... what Nonsens ?!?!!?

Sixth : The Mediaset one Week before ( 31.01.2016 ) uses on an Special day two Tape drives for the Backup Job too , but no Mediaset Date is created ????

Ist a Bug for me and not a Feature , Sorry .

Veeam finally take you Homework with the Tape Option .

[WinstonWolf]

No answer is also an answer !!!!

[Dima P.]

Hi WinstonWolf,
I am sorry for keeping silence. First, I just realized that there is no prompt explanation of the parallel processing logic in the Help Center – thanks for pointing this out. I’ll let Technical Writers team know.

Next, let me explain the logic of the parallel processing to make sure we are on the same page.

For better understanding let’s start with the media pool without parallel processing being enabled. Media pool (basically, a scope of tapes with set retention) should have one sequence of tape media for successive tape job processing. This sequence is called media set where tapes are added one by one (you can manage this stream creation logic and set the custom name). If you scheme the media pool and media set it should look like this:

Tape Job 1 > Media Pool A > Media set 1 > Tape A
Tape Job 2 (will wait for the Tape Job 1 to finish) > Media Pool A > Media set 1 > Tape A (if not enough space next Tape B is used)

Keeping in mind that this jobs are running successively and parallel processing is disabled.

Now we enable the parallel processing for two drives (for example, if you what to run two or more tape jobs at the same time). Concurrent tape jobs pointed to the same media pool should start their own data stream at the same time. Since they can’t use one tape in two drives at the same time, every job will take it’s own tape media and open a media set for each tape. It should look like this:

Tape Job 1 > Media Pool A > Media set 1 > Tape A
Tape Job 2 (will start at the same time) > Media Pool A > Media set 2 > Tape B

That is the reason you may get two media sets opened at the same time and two tapes with sequence number 1. The ID variable I propose above is just a way to easily track such media sets.

Hope it makes sense.

[WinstonWolf]

Hello and thanks for the Answer,

First : Is it possible to say the Tape Job in parallel processing that he dont start the both Backup Jobs at the same Time to the both drives ? For example 10min difference . ( perhaps per Script ? )
Then parallel processing should use only one Media Set , if i understand u correctly .

Second : I have Problems with the %ID% Postfix . Now it Looks so :



We have no Numbers .


Thanks

[WinstonWolf]

Can it be tha the # is missing in the %ID% Postfix ? #%ID%

[PTide]

Hi,

Can it be tha the # is missing in the %ID% Postfix ? #%ID%

Please try to type in a lowercase - %id% instead of %ID%

Thank you.

[WinstonWolf]

Perfekt , it works .
Have you a solution for the other Problem ?
I can explain you why i want to use parallel Processing for the "big" DAILY_EXCH-SQL Tape Job .
On Saturday Morning he creates an new Media Set for this Job and therefore an new virtual Full . Because we want on every Week on every Media Set one virtual Full and the vibs for the Weekdays .
But this virtual Full processing needs very long Time . I think it is because we have an forward incremental forever with 30 retention points . We dont have free space for syntetic fulls on our Storage .


First Question , why the "Processing Full Backup" needs so long Time ?
Second question is the solution with parallel processing the right solution for this behavior ? I Think the Problem is on the Storage Side ?
Goes the Job slowlyer when i use normal processing ?

Pleas i need Help for this Problem .

Thanks

[WinstonWolf]

Need a fast solution for this behavior , Sorry .

[Dima P.]

Hi WinstonWolf,

Is it possible to say the Tape Job in parallel processing that he dont start the both Backup Jobs at the same Time to the both drives ? For example 10min difference .

Then tape jobs are going to be started one by one, new tape media is taken by the dalayed job, thus new media set is going to be started anyway.

Processing for the "big" DAILY_EXCH-SQL Tape Job

If I do remember your previous posts correctly, this source job contains one backup chain, right?

First Question , why the "Processing Full Backup" needs so long Time ?

Since the bottleneck is 'Target' please clarify how your library is connected to the tape proxy

Goes the Job slowlyer when i use normal processing ?

It should not be faster. Generally, parallel processing allows writing two (or more) tapes simultaneously but performance remains the same in both cases (although, you have an option to process in parallel multiple backup chains within one source backup job but that does not seem to be the case).

[WinstonWolf]

No , the Source Backup to Disk Job contains 2 Backup Chains ( Two VM Backup to Disk Jobs ) as you can see in the Screenshot .

The last Question that i ask you is about that u tell me when i use parallel processing it is normal that an new Media Set is created . when i dont use parallelprocessing than i have only one Media Set , right ?

What needs so long time , 30 Hours for this Backup to Tape Job ?

The Tape Drives are connected to the Backup Server via Fibre Channel and SAS ( 2 Tape Drives Fibre Channel , one Drive vie SAS )

[WinstonWolf]

Can i send you the Log Files of this Job and you can see what happens there ?

[WinstonWolf]

On which way i can sent you this file ?

[Dima P.]

when i use parallel processing it is normal that an new Media Set is created . when i dont use parallelprocessing than i have only one Media Set , right ?

Yes - multiple media sets required for parallel processing. Basically, you will have X number of media sets in the media pool, where X is the number of drives you specified at parallel processing step of the Media Pool wizard.

What needs so long time , 30 Hours for this Backup to Tape Job ?

Now let’s investigate this problem. As I understood you are on LTO 5 drives which average transfer rate is ~140 MB (at least if I remember the LTO specifications correctly), is that correct?

You have one backup to tape job with two source backup jobs (two VMs) parallel processing – can you check and share the completion time in the job session for every task (in the left of the backup stratus you have two task which are generated by source job VMs included into tape job)

Can i send you the Log Files of this Job and you can see what happens there ?

You can upload the logs by submitting the case at customer care portal and posting ID to this thread.

[WinstonWolf]

Two of the LTO5 Drives are connected via Fibre Channel , and one drive is connected via SAS ( Drive 3 )

[Dima P.]

Thanks. I did some calculations and, even though I do not expect your tape drive was running all the time on the average 145 MB/s, according to the rough estimate the processing of this huge W2008R2-FILE… should be finished in ~15 hours . So it’s either the statistics or performance is not displayed correctly.

I might be wrong and that’s why I am asking you to open a support case, so we could check the actual logs. Don’t forget to share the case ID please so I could forward it directly to QA team. Cheers!

[WinstonWolf]

I opend already a Case for this Problem , Case ID is : 1699248

[Dima P.]

Thanks - passing to QA team.