Traffic Squeezer - Documentation Coalescing Introduction

Traffic Coalescing Techniques

Packet/Frame Coalescion(or Combination):
When majority of packets of the corporate traffic are small, then packet headers consume substantial bandwidth in comparison to the amount of end-user data transferred. Packet coalescing combines multiple user packets traveling between the same two sites into a single coalesced packet.

Used in conjunction with header compression, this amortizes a single header over multiple packets thus decreasing overhead, and therefore bandwidth requirements. Packet coalescing is particularly beneficial for web applications, VoIP, and interactive applications, like Citrix.

Note: In general "Packet Coalescing" refers to the technique in which multiple packets travelling to the same site is coalesced into one packet. But Traffic Squeezer employs this technique in a bit different way. Since the Traffic Squeezer is deployed for a point-to-point corporate WAN link, all the traffic which is flowing accross this link is destined to the same destination. So, packet coalescing is done on all the packets which arrive the Traffic Squeezer box and the header+payload information is as such retained. Or in other words there would not be any single header derived for multiple packets. The header+payload is just combined with multible packets and made into a single packet (~ near size of MTU) and sent out.

The figure below explains the overall picture of the packet flow in a packet coalescing enabled network:

Traffic Squeezer - Packet Coalescing, Implementation overview:
So in order to support the Frame Coalescing Traffic Squeezer implements a frame container called as "Bucket". The in coming frames are collected into this bucket and once in a while once the bucket is almost full the same is taken out and sent out for further processing. Here is the way the bucket is managed:

Rule 1: The bucket should accumulate only till a specific accumulation time.

Rule 2: The overall accumulation time is pre-set. If this duration exceeds then immediately the contents in the bucket should be sent out.

Rule 3: If after coalescing if the overall packet size exceeds MTU, then previous bucket's coalesced packets/frames should be sent for compression. The new packet should be kept in a fresh bucket.

Rule 4: If the packet size is near MTU then don't add this packet in the bucket. Flush this packet out and also the bucket's coalesced packets.

Rule 5: If there are no packets in the bucket then don't initiate accumulation time time. (I.e) the accumulation timer start ticking only if there is one or more packets in the bucket.

Improved efficient WAN Packet delivery with Packet Coalescing:
Yet another illustration of efficiency improved with TS Packet coalescing feature with TS Packet Coalescing feature. In this illustration, one can well see what happens in case there are say some 500 multi-sized packets need to be pushed across a large distance network such as WAN with many intermediate ISP networking devices.

When TS Packet Coalescing is enabled these 500 multi-sized packets are converted into just 250-400 packets, and hence reduction of the overall processing latencies in those intermediate ISP networking devices.