For those who have not given much thought to WAN optimization for their Cloud backup, this may be an eye opener. There are several little known secrets of WAN optimization that can make your Cloud backup experience a pleasant one. Cloud backup service providers are realizing the importance of this facet of Cloud computing and have quickly transformed WAN optimization from a point-product to a strategic cornerstone application that is well integrated with their online backup product.
At the outset, let us define what we mean by WAN optimization. This will bring some clarity and focus to our discussion.
WAN optimization is a collection of techniques that are used for increasing data transfer efficiencies over Wide Area Networks. The key words are “Collection of techniques” and “data transfer efficiencies”.
What are the techniques that make up this collection and improve efficiencies?
De-duplication: This technique is used to make a binary byte-level comparison between two or more files and folders to ascertain whether they contain are unique data bytes or duplicate data bytes. The duplicates files are then eliminated and only one copy of the file is saved to the storage repository with a reference for the duplicate at the point of occurrence in the data set. When rebuilding the file for recovery, the duplicate file is recreated from the original file that was uploaded to the storage during backup. The process reduces the volume of data that is transferred over the Internet and this saves on bandwidth and online storage space.
Latency optimization: This is not a single technique. It is again a collection of techniques known as TCP tuning, including—Bandwidth delay, Buffer, Windows size scaling, selective acknowledgements, Layer 3 congestion control algorithms, collocation strategies, etc.
- Bandwidth delay-product is a concept that is used in conjunction with TCP to understand the number of bits that can fill up the TCP path or the number of bits that can be simultaneously be transmitted over the network. If the network has a high performance bandwidth, the implication is that data transfer will be faster.
- Buffer is a TCP configuration that is also known as a TCP receive window size. A 64 KiB-1 will support up to 65,535 bytes with small round trip times. This is normally inadequate for large data transmissions, which require high performance networks. The buffer size must be scaled appropriately to prevent large delays in data transmission times.
This article is Part I in a four part series on WAN Optimization.