Single vs Multi-Tier Backup

Most backup software available in the market today follow the single-tier design architecture. This page discusses the differences as well as pros and cons of the two design approach.

Single TierMulti-Tier
On machine is responsible for read the source and writing to destination Two machines are involved. One machine reads from the source, while another machine writes to the destination
Typically, rewrites the entire file from source to destination, even if only a small part of the file is modified Has the ability to transfer the modified part within a file, which is known as delta. This is done by dividing a file into multiple blocks and computing checksum signatures on both ends.
Geared towards a single-user environment Geared towards a multi-user environment
Well suited for backups where the source and destination are close to each other, such as a USB drive Well suited for situations where source and destination are far apart. For example, across the Internet
Backed up files are normally not accessible from the Internet without help from a third-party Files can be accessed from anywhere, simply by using a web browser
It is static. The source and destination must always be connected to each other. For example, your work laptop could only be backed up when you're in the office, or close to where the destination is It is dynamic. The source machine can be moved relative to the destination. For example, users can be on the road and will still be able to run backups to the central server from anywhere
Encrypting data on the target is typically not necessary Encryption is normally required on the target since data resides on a machine that can be accessible to other individuals
Typically, no provision to synchronize multiple machines. Folders on multiple machines can be synchronized through the central repository
Files on the destination serves just one purpose: backup Backup repository can serve for purposes other than just backup. For example, it could be used to share files among colleagues

Remote Backup with Single-Tier

Many companies offer remote backups even with a single-tier backup. For example, where destination is:
  • Storage on public cloud
  • An FTP server
It is important to understand what happens in the background. In both cases, the backup is sent in its entirety. For example, when backing up a 1GB file to this remote destination, the entire 1GB will be transferred every time even if a small portion of the file is modified. This is because the receiving end does not get involved in determining the blocks that are modified.

Sending the entire file whenever a small portion changes does not work with larger files as it uses tremendous amount of bandwidth.

Why Single-Tier systems cannot do block-level backup

A block-level backup requires following steps:
  • Every file is broken down into many blocks (typically over 100,000). These blocks are created on both ends.
  • Checksums for these blocks are compared to create a delta.
  • Delta is sent to the destination
  • The destination server merges incoming delta with the actual file
There is no way to calculate checksums of the blocks on a remote machine without reading the entire file, which means the remote data must come to the local machine. Moreover, there is no way to merge the incoming delta on the destination without a local process on the other end. Therefore, it is not possible to do a block-level backup using a single-tier backup system to a remote destination.


Social Media

Powered by