Quoting Mike Jackson <mj at sci.fi>: > Hi Pierangelo, > The fact that LDAP directory servers are not intended to support a > high frequency of write operations means that the term "write load > balancer" is not the correct term to use when describing the benefits > of multi-master versus single-master replication - unless you are > arguing how to support systems architects who intentionally (or > perhaps out of ignorance) use LDAP technology in an incorrect manner > in their designs. The correct term to use in this context, IMO, is > "highly available write operations". I thought that more accurately describes it as well. And my approach would be to have a "primary" master and a "secondary" master. Throw all your writes at your primary master and if they all go through there, no need to worry about consistency. In the event of a disaster, the secondary master is quite capable of taking the writes as well and keeping things running. This is almost akin to your master/promotable slave concept, except the promotable slave is not really a slave but a standby master. If you're concerned about load, throw in a bunch of slaves that are read-only and point your read-only lookups at the slaves, reserving your masters for applications that need write access. End result - highly available write operations, that minimize consistency issues. Highly available read operations. It requires intelligent design of a directory infrastructure, as you note - sometimes I think we expect software to design that too :) Kevin -- Kevin M. Myer Senior Systems Administrator Lancaster-Lebanon Intermediate Unit 13 http://www.iu13.org
