Mike Bantick
Thursday, 18 June 2009 06:00
Business IT -
Technology
Page 2 of 3
The Juniper WX devices form their own network; a tunnel is formed between each device in the Juniper network in either a Hub-Spoke or Mesh form. Each Juniper network requires at least one ‘registration’ device that coordinates the compression network. Here licensing is important, as certain devices will only allow up to a particular number of tunnels within the compression network.
Traffic identified for compression enters a WX device, if another device on the compression network advertises the intended traffic destination, then the WX device will compress the traffic, replacing strings with tokens and then re-wrapping the IP packet with headers forcing the traffic to its sister device.
On the far side, the sister WX device unwraps the packet and reconstitutes the data using the passed token.
This all happens using a propriety and DNA sequence based algorithm. Depending on the type of traffic involved, compression rates in the high ninety percentile are easily achieved.
The image below shows typical rates of compression using a pair of WX devices. As is shown, typical compression rate of 90 percent is normal, with the graph at the bottom of the screen showing data in as the grey line, data out as orange.
The Juniper WXC devices and their inbuilt hard drive manage, in most circumstances, to achieve even greater percentage of throughput. Traffic such as large FTP’s that occur on a regular basis where the data has only comparatively small changes is ideal for this kind of processing. Compression rates improve over time as the WXC understands the data transfer better with each pass.
The compression, remarkably, is unnoticeable in adding transmission times, and with the WAN acceleration option turned on (where the far side Juniper box will provide TCP/IP ACK acknowledgements in proxy of the near side devices) the result is a sped up network using only a fraction of the capacity originally required.
But there are some pitfalls to be aware of.
CONTINUED on page 3