|
On 17 May 2002, Rached Ben Mustapha wrote:
> On 17-May-2002 Wensong Zhang wrote:
>
> [snip]
>
> > I don't know how many clients are used in your test. You know that the SH
> > scheduler is statically mapping algorithm (based on the source IP address
> > of clients). It is quite possible that two or more client IP addresses are
> > mapped to the same server.
>
> I've done another test and you're right. With 5 client IP addresses, there is
> load balancing. I've done the test with only 3 different machines which were
> on
> the same network in my previous mail. I would have thought the SH algorithm
> could have a "perfect" balancing scheme.
> Could the kernel parameter IP_VS_TAB_SIZE influence in that way ? I've put 16
> bits instead of the default 12.
>
No, the IP_VS_TAB_SIZE doesn't influence the SH scheduling.
In the SH scheduling, there is a static hash table in which destination
servers are added in round-robin manner, and a function which calculate
the hash key based on client source IP address. We hope that this function
can help schedule many client requests to the servers evenly. But with few
client requests, it is quite possible that they are sent to the same
server, because the same hash key are calculated based on their source IP
addresses.
Regards,
Wensong
|
