New Set of High Performance Networking Patches Available

Chris Rapier rapier at psc.edu
Fri Aug 5 02:55:28 EST 2005



Darren Tucker wrote:
> Chris Rapier wrote:

> I can see 1+ MB sitting in the server's TCP send queue.  I suspect it's 
> some local problem limiting TCP throughput in the high-BDP configuration 
> (they're not super beefy hosts but a direct connect gives me ~6MB/s so 
> they're capable of more than the ~ 500-600 KB/s I'm seeing with the 
> latency).

Yeah, thats definitely low. Can you try a loopback to your localhost and 
dump the data into /dev/null to see what sort of limit the CPU is 
imposing? Also, what cipher? 3des is justa nightmare. Most of our tests 
are using arcfour and blowfish.


> You can specify it in the config file on a per-host basis and/or as a 
> default so I think it's OK.  In fact, it's probably more convenient for 
> end users (as opposed to people testing ssh mods :-) since they can just 
> enable it for the appropriate hosts then forget about it.

That works. Honestly, this was mostly an afterthought - one of the users 
that connects to us needed something like this for various reasons. They 
did good stuff with it though (near term storm forecasting 
http://www.psc.edu/publicinfo/news/2005/2005-07-05-caps.html - all data 
xfered with hpn-ssh).


>>> This would allow it to be set either per-connection or globally, and 
>>> may be passed through from the scp command line with the "-o" option.
>>
>>
>> Okay, so thats basically the same thing. I'd suggest using a shorter 
>> name though but thats not a mahjor point to get hung up on.
> 
> 
> There's an existing config TCP option (TCPKeepAlive), I picked it to be 
> consistent with that.

I just hate having really long command lines but consistant nomenclature 
is important. I have no objections to a longer more descriptive name.

>>> The latter would also mean that scp would need less modification (and 
>>> scp's code is mostly shared with rcp, so that's also a plus).
>>
>>
>> I honestly think changing scp's code isn't a bad thing. :)
> 
> 
> Yeah but you don't have to maintain it :-)

Yeah, but if we change it enough we'll be forced to maintain it. So this 
would be an excellent opportunity to pass the torch :D




More information about the openssh-unix-dev mailing list