Block Delta Backup

Backup for Workgroups uses a built in Block Delta Engine to compute the differences between the current version of a file and the previous version of that same file.  This process allows Backup for Workgroups to determine the portion of a file that changes.  As a result the Backup Client can send just this change information to the Backup Server to reduce the time it takes to transmit the file being backed up to the Backup Server.

When you perform a backup over the Internet, the Delta backup approach can reduce the backup time significantly.

What is a Block Delta Engine

A Delta Engine is a process that computes what has changed in a file.  The Delta Process take two versions of a file, usually the current version of the file and the most recent past version.  It compares the two versions of the files and determines what has changed.  The goal is that the delta is significantly smaller than the size of the current original file.  The more correlated the revisions are, the smaller the delta will be.  From a backup perspective the goal is to speed up the backup by working with smaller data than the original file.

Delta engines can operate at different granularities on the two versions of the file.  They can compare the files either on a per byte basis or in blocks.  A Block Delta Engine will examine the file versions in chunks of the file called Blocks.  The difference is then computed at the resolution of a block size.  The goal of using Blocks is that the delta computation can be slow, the smaller the block size the more pieces of the file must be compared.  Byte based deltas can be very slow.  By chunking the files into Blocks, the speed of the delta process is great increased. 

Explain Block Delta Backup process

The best way to explain the Block Delta Backup process is to walk through the steps that are performed when a file is backed up, changed and then the changes are backed up.

Setup

Backup for Workgroups defaults to including file that are larger than 100MB as candidates for the Block Delta Process.  Files under this size tend to have little benefit from the Delta Process.  Also since the Delta process operates one the past revision of the files to backup, this past data needs to be stored and will cost local hard drive storage, as a result we recommend limiting the use of the Delta Engine to large files instead of all files.   

You can change these settings at the Backup for Workgroups Backup Client -> Tools -> Preferences -? Block Delta Backup panel.

Step 1 - The first revision

When the backup process sees a new file that has not been backed up, it will backup the file and send a full compressed copy of the file to the Backup Server for storage.  If the file is larger than the threshold specified for consideration of the Block Delta Engine, then the backup process will not only perform the complete backup of the file but it will also construct a file that contains information about the file being backed up.  Since the Block Delta Engine needs to compare 2 versions of the files, this current version is being prepared to be compared to a future version.

The file being backed up is processed in Blocks and a signature is computed for each Block.   The signatures relating to all Blocks in the file being backed up, are stored in a file on the computer being backed up.  The size of this signature file is about 1.5% of the size of the original file being backed up. 

Step 2 - The second revision

Over time as the files that have been backed up are changed, the Backup Process will detect which files have changed and backup those changed files.  Normally this backup process will backup the complete file, compress it and send the compressed data to the Backup Server for storage.  The normal compression is about 50%.  With the Block Delta Engine, instead of sending the entire file to the Backup Server, the Client can compute the changes to the file and send just the changes.  Typically that can achieve a reduction of 99% of the file size.

When a file that was previously backed up by the Block Delta Engine has changed, the Block Delta Engine will read through the current version of the file and cross compare the file with the Block signature file that was computed during the last backup of this file.  The Block Delta Engine will compare the past Block Signatures with the Blocks of the current file to compute what has changed between the two versions of the file. 

The Block Delta Engine computes the changes to the current file and this delta is then compressed and sent from the Backup Client to the Backup Server.  The typical size of the the compressed delta is commonly 1% of the size of the original file accomplishing a 99% file size reductions.

The Backup Server uses the Block Delta to compute a copy of the current file being backed up.  The constructed copy of the backup file is compressed and stored at the primary storage drive.

The Backup Client completes the backup by computing signature information on the blocks contained in the original file being backed up.  These signature are stored locally to replace the previous signature file.  This new signature file will be used to to determine what changes when the next backup occurs.

What are the benefits of using Block Delta

The primary benefit of the Block Delta process is to reduce the amount of data that is sent from the Backup Client to the Backup Server.  When you are backing up over a slow link such as the Internet this reduction can be very significant.

Let's say you have a 50GB Exchange Database mailbox store.  Normally each time you perform a backup, a compressed copy of this 50GB needs to be sent from the Backup Client to the Backup Server.  Assuming the compression ratio is 50% then each backup needs to send 25GB of data.  If the link between the Backup Client and the Backup Server runs at a typical uplink speed of 50KB/Second then it would take 138 hours. 

The Block Delta Engine computes what changes in a file, and with highly correlated files such as Exchange mailbox stores, it is common to see the Block Delta size of 1% of the original file size or in this example 500MB.  At this same transfer rate the Block Delta would take just 2 hours.

Are their any problems when using a Block Delta Engine

In order for the Block Delta to compute the changes to a file it needs to keep information about the previous version of the file.  The Block Delta Engine reduces the amount of storage needed from the entire original file to just signature information on the blocks of the original files.  These signatures are about 1.5% of the original file size.  This information is stored on the computer and as a result you need to have enough local disk space to hold the Block Delta Signature information.  By default the Backup Client will limit the amount of local storage consumed to 1GB.  You can control this storage size, in the preferences at the Backup Client.

One other consideration when using the Block Delta is that the Delta takes more CPU time to compute.  During the backup process you will notice a higher CPU load than the normal backup process.