This is a discussion on Re: Outstanding queries vs. source port randomization entropy - DNS ; > On 2008-09-01 00:46, Mark Andrews wrote: > > Additionally with sendto() you will tend to get locked onto > > a open set of ports as you need to wait for the next response > > whenever you re-use ...
> On 2008-09-01 00:46, Mark Andrews wrote:
> > Additionally with sendto() you will tend to get locked onto
> > a open set of ports as you need to wait for the next response
> > whenever you re-use a port. With connect() each transaction
> > becomes almost completely independent of other transaction.
> I don't follow; when you have ~2^16 distinct source ports bound, how can
> you get locked into a sub-optimal set of ports?
Ususally you don't have 2^16 ports in use.
> Your point about combining connected UDP sockets with SO_REUSEADDR in
> order to take advantage of kernel filtering and avoid impacting on other
> uses of UDP is well taken, but a drawback I see is that your file
> descriptor set has to continue to grow past 2^16 if you don't actually
> reuse existing sockets.
Yes. There are tradeoffs everywhere in this space.
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742 INTERNET: Mark_Andrews@isc.org
to unsubscribe send a message to email@example.com with
the word 'unsubscribe' in a single line as the message text body.