Call for stats : Backup jobs combine vm's or just seperate them in seperate jobs and Ardchive (+ validating times)
i see the following scenarios / type o backups jobs when you want to backup vm's
Backup
1) Backup a single vm into a dedicated/single Archive file
- many backup jobs and Seperate archive files
2) Backup multiple vm's into a dedicated/single Archive file
- a big archive and one big archive dedicated to a backup job
3) Backup multiple vm's into a shared/single Archive file
- a combined backup job a big archive shared by other backup jobs..
4) Backup a single vm into a shared/single Archive file
- a single backupjob for each item but storing it in a big shared archive
i wonder about the validation proces, does it then do deduplication?
and combine the seperate tib files into one big TIB file?
i do see longer validation times when the TIB files has many servers into one tib file
some take 15 minutes to complete and some take 6 hours to complete (comparable backup size)
my question to you all is what type of backup do you use (1,2,3 or 4) and do you validate and how long does it take! and while you are at it what is your average backup speed?
Also does anyone have experiences with validating over ftp?
i do wonder how this would work.. while a validate over a lan takes a long time.. over ftp remote will be even worse
purhaps also a call to support to explain how validate actually works.
woudl you say the validation is not required? it has no other benifit then checksumming? it does notthing else like deduplicationg after the backup?
so then then other idea comes.
on a weekly backup schedule
- make a backup every day without validation except 1 day to Archive1.tib
- then on the 1 day that is excluded you will enable validation on Archive1.tib
also keeping in mind the change rate of vm's to make one big TIB or just 1 TIB for a server
do also many vm's of the same class benifit from deduplication? and to avoid mixing linux and windows vm's purhaps but keep windows X of the same version together
Hello Jan,
Thank you for your reply.
It might related to the performance of your deduplication. For this, we have this article in our Knowledge Base.
I suggest you go through the manual and the settings accordingly. The storage node itself should run on an x64 machine that has enough capacity available.
Instead of validating the whole archive, you should validate only the last backup. This approach saves time and performance on the machine with the Acronis Storage Node.
Let me know if you need additional help please.
Thank you.
At first I was using method #2 to backup three guests, but I found that the backup tasks were taking a very long time because the entire archive was being validated after each guest was backed up (adding about 3-4 hours per guest!). Also, the validations would start failing after several backup tasks had been performed.
Now I am using method #1 and it has been working flawlessly for months and the tasks completely quickly. However I do not like this method as I have to create a separate task for each guest, which somewhat negates the convenience objective of using virtualization-aware backup software to to backup an entire host.
What I would like to see is the validation to be performed AFTER all of the guests have been backed up - I don't care if the validation runs all day long, but I do care if the actual guest backups take half of the workday to complete because more time is spent validating than is spent backing up.