WAN Optimization/ TCP accelerator over satellite link

[aut8133024]

Hello

I have a question with respect to test of TurboIP-G2 (kind of TCP accelerator) and would really appreciate if you let me know its answer and guide me in this regard.

I am dealing with a predicament regarding testing TurboIP-G2 on 4.5 Mbps satellite link.

My test scenario is as follows:

I want to measure Bandwidth throughput of my link with and without mentioned-device.

I would have to say that I configured TurboIP-G2 according to its manual available on its website, step by step.

For generating TCP packets I use “Psping” S/W (An estimation bandwidth tool in order to measure available BW) and when I use TurboIP-G2 on both side of satellite link the software shows the available BW is approximately 16 Mbps.( nearly fourfold ) I tested this scenario using other tools (iperf, ttcp, spruce and some Linux-based software) and surprisingly I got the same results.

Is this due to compression over satellite link or something else?

I am looking forward to hearing from you.
Best Regards.
H

 

[Samir]

This has to be compression as there is no other explanation if the true bandwidth is 4.5. Unless the 4.5 is not absolute, then it's something else.

 

[Dreamslacker]

There is definitely compression going on.

One way to check for compression is to transfer a heavily compressed file to see if you see any or as much benefit.

Additionally, if this device tunnels your traffic (like with a VPN), it is possible that proxies (MitM agents) help with handling certain protocols to reduce the handshake overheads. e.g. Some WAN accelerators recognize SMB/ CIFS traffic and can 'hijack' the data to re-arrange them in a manner that removes/ reduces the overheads involved in regular transactions.

 

[aut8133024]

thank you for your answer

 

[aut8133024]

I know that TurboIP-G2 (WAN accelerator) uses SCPS-TP protocol over satellite links and compression is enable as well. but my question is that is it possible that compression rate would be over 400% ?
I get confused when i saw the same result with different scenario and tools

 

[sinshiva]

somebody mentioned tesing with a highly compressed file because that would give your link as much original data that cant be further compressed to test with. if you are familar with linux, another method that may work well is to pipe /dev/urandom or whatever via nc (netcat) across the link, to generate data that cant be compressed so easily. the compression you are seeing from the tools you are using is probably not abnormal.

 

[stevech]

Top over geosynchronous orbit satellites suffers the long delay. Satellite providers offer bridges with special protocols to transport encapsulated IP. But in the final result, the 600mSec round trip delay is never good for web apps.

It's not the data volume... It's the delay for ACKs... And that's why TCP gets encapsulated.

 

[Dreamslacker]

I know that TurboIP-G2 (WAN accelerator) uses SCPS-TP protocol over satellite links and compression is enable as well. but my question is that is it possible that compression rate would be over 400% ?
I get confused when i saw the same result with different scenario and tools

Yes, it's possible. If you create a 50MB text file consisting of nothing but the same repeating text and zip it (or TAR, GZ etc), you can get the compressed file down to a matter of KBs. Link compression works in the same way.

You can think of it as that your traffic is compressed at the local accelerator to a smaller sized package, then sent to the remote accelerator and decompressed. Your actual bandwidth does not change but the amount of data you're sending in the same time period effectively increases.