Understanding WAN acceleration techniques

The sheer amount of data explosion over a WAN (Wide Area Network) in today’s digital world has propelled WAN acceleration techniques to greatly evolve with time and a WAN today is much smarter than it was decades earlier. Here is an understanding of important WAN acceleration techniques that help in optimizing todays WAN:

Data reduction

This technique eliminates duplicate data within data that is sent over the WAN by examining duplication of data in real time and sending references to data instead of repeating the data again – the data is stored in local data stores. Data reduction can reduce 90% of WAN bandwidth by preventing repetitive data traversal as well as reducing disk costs due to lower storage space requirements. There are mainly two types of reduction techniques – one at file level and other at block level (i.e. data in a file) with block level giving even better performance at the cost of requiring more processing power.

Compression

Compression techniques are applied to data patterns along the WAN. Compression and decompression is done by using WAN acceleration appliances on sending and receiving ends.  Compression algorithms range from simple ones detecting repeated bytes in a packet to smarter ones, which can find duplicates across packets and flows. Header compression reduces packet header information while payload compression algorithms detect repetitive short byte sequences. Compression can typically reduce WAN utilization by 50% thereby doubling bandwidth.

Latency mitigation

Latency, which is the packet trip time between host and destination, has a big impact on the performance of applications across the WAN. Latency can severely limit throughput as TCP congestion control limits the amount of unacknowledged data in transit – new data transmission is postponed until older data is acknowledged, on reaching the TCP congestion window size. Acceleration techniques used to overcome the latency issues are TCP refinements such as window-size scaling, selective acknowledgements, layer 3 congestion control algorithms and even collocation strategies placing applications near the endpoint. Others include CIFS acceleration and application specific acceleration – such as tuning of application protocols including HTTP, HTTPS, SQL, MAPI, NFS and CIFS etc.

Quality of service (QoS)

When demand exceeds capacity of a WAN link and traffic is contending for the same limited resources, less important traffic can hog bandwidth of business critical applications. To counter this, some WAN acceleration techniques implement QoS to prioritize and classify traffic based on different criteria. These include packet classification based on source, destination addresses and/or applications as well as queuing and service mechanisms that are used to apply service policies based on these classifications.

Loss mitigation

Forward error correction (FEC) is a technique applied to correct errors at the physical layer – in TCP; unacknowledged packets need to be retransmitted in congestion prone networks. An additional error recovery packet is added after every ‘N’ packet across the WAN. If a packet is lost during data transfer, the FEC packet can be used to reconstruct the lost packet. This eliminates the need to retransmit lost packets across the WAN, which dramatically reduces the latency of the link.

Packet coalescing

For small packets, the packet header information consumes more space than the actual user data. Packet coalescing combines multiple user packets travelling between the same two sites into a single packet. This often involves header compression and a single header over multiple packets reduces bandwidth requirements especially for VoIP and interactive web applications.

Wide area file services (WAFS)

WAFS is a caching technology to access remote files as local files. It enables users to access file services faster and reduces costs.

There are additional techniques such as protocol spoofing and traffic shaping which can be configured by network operators/administrators to provide data acceleration through WAN.

In summary, today’s WAN performance can be accelerated through a number of effective techniques and the best solution may be to implement a combination of these.