cancel
Showing results for 
Search instead for 
Did you mean: 

What would be the size of full snapshot backup? Is it same of size of data? what data set/ selection which is going to backup?

Chethan
Level 3
Certified

What exactly is the difference between VMware snapshot and normal file system snapshot

1 ACCEPTED SOLUTION

Accepted Solutions

Dan_42
Level 4
Partner Accredited

VMware snapshot backups (this is true of any provider, VEEAM etc., not just NBU, as this is via VMware's APIs) ask VMware to take a snapshot of the VM. What this does is it causes the VM to run off delta VMDKs rather than their normal VMDKs. 

I'll give you an example

So normally if you look in the datastore, you'll have a folder MyVirtualMachine. Inside, you'll have a number of files

myvirtualmachine.vmx (the config file for your virtual machine)

myvirtualmachine.vmdk (the C: drive) 

myvirtualmachine_1.vmdk (the D: drive) 

etc.

When you create a snapshot you'll see 

myvirtualmachine.vmx (the config file for your virtual machine)

myvirtualmachine.vmdk (the C: drive) 

myvirtualmachine-0001.vmdk (delta file for the C: drive)

myvirtualmachine_1.vmdk (the D: drive) 

myvirtualmachine_1-0001.vmdk (delta file for the D: drive)

etc.

The -0001 delta files start out as 0 bytes, and are just change logs. They point to their parent vmdks, and all writes now go to the -0001 files rather than the .vmdk files. Reads are based on the content of the base vmdks and "filtered" through the delta files to determine whether anything on the base disks should be superseded when doing reads (not entirely precise, but you can think of it like that). Before the delta files are created, you'll typically quiesce/prepare the VM (get filesystem to a consistent state and perhaps perform application processing). Now you have a set of base vmdks that are not going to change until the snapshot is removed. The backup software then grabs these VMDKs off the datastore via one of several methods (san, nbd, hot add, etc.) and processes them. This affords the opportunity to grab the whole server in one big consistent state, catalog/index files as necessary (NetBackup has technology to read/interpret the content of the vmdks), perform application processing via VIX (for example, truncate Exchange logs once we're done) etc. Once everything is backed up, the snapshot gets deleted, which involved consolidating whatever writes took place to the delta (-0001) files back onto the base vmdks.

This explains full. Once you've done a full, subsequent backups can be run as "incrementals" by utilizing VMware change block tracking technology, which in effect allows NetBackup to ask VMware what blocks have changed on the VMDK since the last time it was backed up, and retrieve only those (still following the previously mentioned snapshot create/process apps/delete snapshot workflow).

 

On the Windows side, "snapshots" via VSS provide a consistent, quiesced version of a drive(s) that can be read as a singular point-in-time instance. When a snapshot is created in this fashion, writes to files on the drive get redirected elsewhere until the snapshot session is over. Once it's cleared, things get cleaned up. 

 

Does that answer your question? 

View solution in original post

4 REPLIES 4

SymTerry
Level 6
Employee Accredited

A VMWare snap is going to backup the whole machine (VM). A FlashBackup, with proper config of policy, will get all local drives and a snap of the current system state (Volume Shadow Copy components and System Service writers). 

NetBackup for VMware (HOWTO70878) and FlashBackup-Windows (HOWTO50362) support Windows NTFS file compression, for backup and restore. However, they both do not support NetBackup's compression options.

You will have to test, but the VM might take more space.

A major difference between the 2 is restoring this data. With a VM its just a restore to a ESX box and boot the VM. Flashbackup requires a already installed OS with netbackup client on it to restore on to, and is not a dissimilar restore (TECH150156)

 

Chethan
Level 3
Certified

Hi

Thanks for a swift respose !

What backup does only backup the index or pointers to the data table. Since we say snapshot is point in time copy how it can backup whole data with single click of time. How we can restore data using pointers?

For example ,If we configure local drives to backup in snapshot method. what data it is goint to backup, whole data or index list to data

 

 

 

Dan_42
Level 4
Partner Accredited

VMware snapshot backups (this is true of any provider, VEEAM etc., not just NBU, as this is via VMware's APIs) ask VMware to take a snapshot of the VM. What this does is it causes the VM to run off delta VMDKs rather than their normal VMDKs. 

I'll give you an example

So normally if you look in the datastore, you'll have a folder MyVirtualMachine. Inside, you'll have a number of files

myvirtualmachine.vmx (the config file for your virtual machine)

myvirtualmachine.vmdk (the C: drive) 

myvirtualmachine_1.vmdk (the D: drive) 

etc.

When you create a snapshot you'll see 

myvirtualmachine.vmx (the config file for your virtual machine)

myvirtualmachine.vmdk (the C: drive) 

myvirtualmachine-0001.vmdk (delta file for the C: drive)

myvirtualmachine_1.vmdk (the D: drive) 

myvirtualmachine_1-0001.vmdk (delta file for the D: drive)

etc.

The -0001 delta files start out as 0 bytes, and are just change logs. They point to their parent vmdks, and all writes now go to the -0001 files rather than the .vmdk files. Reads are based on the content of the base vmdks and "filtered" through the delta files to determine whether anything on the base disks should be superseded when doing reads (not entirely precise, but you can think of it like that). Before the delta files are created, you'll typically quiesce/prepare the VM (get filesystem to a consistent state and perhaps perform application processing). Now you have a set of base vmdks that are not going to change until the snapshot is removed. The backup software then grabs these VMDKs off the datastore via one of several methods (san, nbd, hot add, etc.) and processes them. This affords the opportunity to grab the whole server in one big consistent state, catalog/index files as necessary (NetBackup has technology to read/interpret the content of the vmdks), perform application processing via VIX (for example, truncate Exchange logs once we're done) etc. Once everything is backed up, the snapshot gets deleted, which involved consolidating whatever writes took place to the delta (-0001) files back onto the base vmdks.

This explains full. Once you've done a full, subsequent backups can be run as "incrementals" by utilizing VMware change block tracking technology, which in effect allows NetBackup to ask VMware what blocks have changed on the VMDK since the last time it was backed up, and retrieve only those (still following the previously mentioned snapshot create/process apps/delete snapshot workflow).

 

On the Windows side, "snapshots" via VSS provide a consistent, quiesced version of a drive(s) that can be read as a singular point-in-time instance. When a snapshot is created in this fashion, writes to files on the drive get redirected elsewhere until the snapshot session is over. Once it's cleared, things get cleaned up. 

 

Does that answer your question? 

Chethan
Level 3
Certified

Hi Dan,

Nice explanation and thanks for it.