Hy threre, we're evaluating ILB as a DNS Loadbalancer and so far, it looks quite promising. Thank you all for your awesome work.
However, during the tests, some questions arose, we couldn't answer. (If this is not the right place to, ask i'm thankful, if you point me to right direction.) Right now, we're using the ILB plain with round robin. OK, easy. There are 3 other Algorithms mentioned in the man page (hip, hipp, hipv) and i assume they are generating hashes from client-ip, client-ip & sourceport and client-ip & vip. But what happens then? How is the hash processed to decide which backend should be used? Should one consider to assign Backend Server IPs in a special pattern because this would affect the algorithms or is this completely seperate? Thanks for your help Lutz