I expanded the first section a bit and added it to the "Architectural Overview" page of the AUG (http://directory.apache.org/apacheds/1.0/architectural-overview.html). Further comments or contributions are welcome :-)
Christine 2007/4/5, Trustin Lee <[EMAIL PROTECTED]>:
Sounds very good! Trustin On 4/5/07, Alex Karasulu <[EMAIL PROTECTED]> wrote: > To clarify from previous threads ... > > Definition > ------------- > > A partition is a physically distinct store for a subset of the entries > contained within a DSA (Directory Server/Service Agent A.K.A the LDAP > server). The entries of a partition all share the same suffix which is the > distinguished name of the namingContext from which the stored entries in the > partition are hung from the DIT. A partition can be implemented using any > storage mechanism or can even be backed in memory. A partition simply has > to implement the Partition ( 1.5) interface and by doing so can be mounted > in the server at it's suffix/namingContext. > > The server can have any number of partitions (with any implementation) > attached to various namingContexts which are published by the RootDSE (empty > string dn "") using the namingContexts operational attribute. So if you > want to see the partitions served by the server you can query the RootDSE > for this information. > > Motivation For Terminology > ------------------------------------ > > I decided to use the term Partition to denote this structure in the server > rather than use the term backend. Why? Well to me the backend is > everything after the protocol which is basically all of the core. The > frontend is the MINA based LDAP protocol service provider. Another reason > for this choice was the fact that a partition is a separate physical store > that separates a subset of the entries. It partitions the DIT, the > namespace hence the term "partition" made sense to me. > > Another crazy reason for this comes from my days as a Linux system developer > while writing device drivers (ahhh those were the days). Anyways I really > like to parallel the concepts of the Linux file system terminology since it > is almost exactly the same concept but in another domain. Basically file > system partitions which can be backed by any kind of random access store can > be mounted at mount points on the file system. To me the mount point is > analogous to the namingContext used to hang the entries stored therein off > the DIT (here I'm relating entries to files). Note this analogy does > breakdown in some places but over all it's a descent analogy which warranted > using the name Partition rather than Backend. > > Alex > > -- what we call human nature is actually human habit -- http://gleamynode.net/ -- PGP Key ID: 0x0255ECA6
