On 23/05/2016 08:07, "Fischetti, Antonio" <antonio.fische...@intel.com> wrote:
>Hi Daniele,
>I added one comment below.
>
>Regards,
>Antonio
>
>> -----Original Message-----
>> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Daniele
>> Di Proietto
>> Sent: Tuesday, May 17, 2016 1:57 AM
>> To: dev@openvswitch.org
>> Subject: [ovs-dev] [PATCH v3 05/16] conntrack: Periodically delete
>> expired connections.
>>
>> This commit adds a thread that periodically removes expired
>> connections.
>>
>> The expiration time of a connection can be expressed by:
>>
>> expiration = now + timeout
>>
>> For each possible 'timeout' value (there aren't many) we keep a list.
>> When the expiration is updated, we move the connection to the back of
>> the
>> corresponding 'timeout' list. This ways, the list is always ordered
>> by
>> 'expiration'.
>>
>> When the cleanup thread iterates through the lists for expired
>> connections, it can stop at the first non expired connection.
>
>[Antonio F] This ordering mechanism is interesting, but could we loose its
>benefit in case of many TCP retransmissions?
>Or for ex. in a Telco environment, when we're tracking UDP connections
>(GTP tunnels) between S-GW and P-GW? In this case all the apps running
>on one mobile phone can share the same S-GW/P-GW UDP connection.
>So the overhead of pushing back on the conn list could be big.
Hi Antonio,
I'm afraid I'm not familiar with the use case you're describing. I can tell
that from an OVN perspective, for example, there isn't much value in sending
encapsulated (VXLAN, Geneve or STT) traffic to the connection tracker.
In my benchmarks I found that pushing a connection to the back of a list is
very cheap (I couldn't notice the difference) compared to the other
per-packet operations that we have to perform, like the lookup in the hash
table. After all we're using a doubly linked list, and the complexity of
pushing back an element is O(1) in the number of connections. Is this the
overhead you're worried about?
Have I completely misunderstood your question? :-)
Thanks,
Daniele
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
