Hi Folks, I've often heard about the infinite-scalability that will be created by something like ZEO and I've always wondered about one thing : how is the load-balancing being done ? Just say we have 3 front-line servers (ZSERVER_1, ZSERVER_2 and ZSERVER_3 in the diagram below), each running Zope with identical application logic replicated by ZEO (or simply copied across manually); the front-line servers all use the same database (DB) for storing sessions. Our load-balancer distributes incoming requests between the 3 servers : +---[ZSERVER_1]----+ | | REQUEST-->[LOADBALANCER]----+---[ZSERVER_2]----+------[DB] | | +---[ZSERVER_3]----+ What are people using for the load-balancer (and I'm looking for real load-balancing, not DNS round-robbin) so that the multiple front-line servers appear as one big server... ie. so that they are transparent to the outside world. In particular, so that cookies set by any one of the front line servers will be shared by the others. (ie. they can not have different fully-qualified domain names or else cookies set by one will not be shared by another). Does anybody with experience have any recommendations for the load-balancer ? Thank you very much, chas ps. In the past I've got around this by ensuring that if the very first request from a given browser is dealt with by, say, server silo8.domain.com, then all subsequent requests from that browser are also dealt with by by that server. But that (a) introduces head-aches with application logic and (b) is not good if silo8.domain.com suddenly goes down.