[nsd-users] NSD TCP performance

nsd at dclg.ca nsd at dclg.ca
Fri Nov 9 17:06:20 UTC 2007

>>>>> "W" == W C A Wijngaards <W.C.A.> writes:


W> Colm MacCarthaigh wrote:
>> On Thu, Nov 08, 2007 at 06:30:56PM -0800, Aaron Hopkins wrote:
>>> On Thu, 8 Nov 2007, nsd at dclg.ca wrote:

>>>> NSD always sends the two byte response size as a separate TCP
>>>> packet (causing the requestor to send a separate ACK) to the main
>>>> body of the request.

>>> I reported this in May 2006.  Found in nsd-3.0.6/doc/TODO:

>>> - From Aaron Hopkins: write tcp length and tcp data in one write
>>> operation, instead of multiple calls to write. Avoids Nagle algo
>>> delay in this case.  preallocate 2 bytes in front of buffer to put
>>> them into.

W> Yes, this is what NSD does. Simple code.

W> TCP performance is not considered important, as it should rarely be
W> used with DNS. UDP performance is the goal. Therefore the
W> (otherwise excellent) patches from Aaron did not get applied, they
W> introduce extra code, portability attention, where it is not
W> needed.

I'd like to have a look at this patch.  Maybe the patch can be worked
ina more acceptable manner.  My client is very concerned about TCP
performance because of DNSSEC being on the horizon.


|David Gilbert, Independent Contractor.       | Two things can be          |
|Mail:       dave at daveg.ca                    |  equal if and only if they |
|http://daveg.ca                              |   are precisely opposite.  |

More information about the nsd-users mailing list