On 2011-01-11 12:40, Yong Lucy wrote:
>> If you want to
>>> tell me "oh, I take the modulus of the flow label with the
>>> number of servers I have to select which server", or any
>>> similar stateless algorithm, I can do it with that hash just
>>> as easily. And yes, that gives me a session predictably going
>>> to the same server. But it doesn't allow me to balance loads.
>>> To balance loads, I need to estimate the load on individual
>>> paths, and assign new data streams to paths as they change in
>>> their actual loading.
>> Why is that better than a statistical approach? I assume we're
>> talking about thousands of flows. In any case, in massive
>> server farms, the server load may be a more important parameter
>> than the network path load.
>
>
> [LY] Statistical approach works well under the consumption that there are
> thousands of flows and they have the similar rates. Today's Internet may
> have the flows that only have few packets and the flows that have thousands
> packets per second and last long. Hash does not work well under this
> condition.
In fact no solution works well for short flows and the problem isn't important
for long flows with few packets. So I think it's OK to discuss a solution
that works for long flows with many packets, since that covers most
of the load.
>
> [LY] the server load balance and the network path load balance have
> different criteria.
Very true. It isn't clear to me that balancing the network load
without knowledge of the server load is a good idea.
Brian
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
