Re: [Karlnet] RE: Ping Jitter

[Chris Conn <cconn@abacom.com>]

Hello,

I find it curious that people use ICMP rtt to measure anything.  This is even 
flawed in tradition land-based networking for various reasons, however compound 
the fact that the Karlnet polling algorithm is designed to maximize throughput 
to satellites who need airtime can easily explain this "jitter" as you call it. 
 If you even moderately load the satellites with traffic, the ICMP rtt times 
stabilize greatly.  I have not corrolated this phenomenon to be associated to 
lower SNR values, however I have easily seen this on my own networks.

It is not too difficult to explain that appearances can be deceiving.  ICMP rtt 
is not a bandwith measurement tool, it is a REACHABILITY tool.  I can 
understand that if your ICMP rtt is 1000+ms constantly there may be an issue 
either of bandwith saturation or other (retransmits on your lower SNR 
satellites???), but if it jumps from 30 to 250 momentarily, this does not 
represent lack of bandwith, especially when you take into consideration again 
the polling mechanism.

So, would you rather have a network that _seems_ to have low latency, or do you 
truly want to maximize what little spectrum can be used to carry data from and 
to satellites that need it?  It may be difficult to explain to a clueless 
customer that uses whatever el-cheapo network analysis tool ala-Net.Medic, 
however I think it should be understood by the network operator to at least be 
able to attempt to explain this to the customer.

My 0.02$,

Chris