Smarter Compression Technology

Peribit's SR device increases bandwidth by recognizing traffic.

June 9, 2003

2 Min Read
NetworkComputing logo in a gray background | NetworkComputing

Compression Tests

Perebit SR-20click to enlarge

I set up my test environment to simulate a central office with two remote offices connected via T1 (1.5 Mbps) lines. First I installed the SR-55 and designated it as the registration server, which keeps track of the other SRs on the network.

For my baseline test I transferred a 1-MB text file via FTP. It transferred in 5.61 seconds at 187 KB per second (approximately 1,496 Kbps). I then transferred the same file five times consecutively with the SRs on. The transmission time progressively decreased from 2.94 seconds to 0.18 seconds at 5,799.64 KB per second. I repeated the test with a 1.1-MB PDF file; its transfer speed went from 206.23 KB per second to 387 KB per second. Finally, I transferred a 50-MB Windows Cabinet File. The initial test ran at 182.14 KB per second and subsequent runs took just as long. That's because there is very little repetition in a Cab File. Overall, I saw an average 25 percent to 50 percent reduction in file size.

Customizing

Peribit's device lets you create application classifications so you can instruct the SR to ignore encrypted or previously compressed data.

The SR also has some limited reporting capabilities. It can generate graphs for throughput, data-reduction rates and total data transferred, and display them by application.

Peribit's CMS is primarily for large deployments. I installed a copy of CMS on my LAN segment, and after specifying the registration server's IP, I was given a list of the SRs on the LAN. I could extract from or load software images and configuration files to all my SRs. To test this capability, I extracted the configuration from SR-2 and specified it be loaded on SR-3. The unit rebooted, and a minute later the config was there.

The CMS lets you view traffic reduction stats, though it offers only a graphical representation of how much data is reduced for each tunnel; it doesn't provide hard numbers for the entire network.

Michael J. DeMaria is an associate technology editor at Network Computing's Syracuse University's Real-World Labs®. Write to him at [email protected].

Post a comment or question on this story.

SUBSCRIBE TO OUR NEWSLETTER
Stay informed! Sign up to get expert advice and insight delivered direct to your inbox

You May Also Like


More Insights