Re: Comments w/r/t data mastered in many places...

["Ed Reed" <ED_REED@xxxxxxxxxx>]

John Merriells wrote...
>Alan Lloyd wrote:
>
>
>>         But then how will I get the master entry if I need it. I may be
>> connected to a replica server for my address book details, but would
>> really like the military target information from the master in case its
>> been updated in the last ten minutes and has not been propagated yet.
>>         Users must have the mechansims to get to master info - in case
>> of inconsistencies caused by latent replica updates -
>>         See X.500 - Dont Use Copy, Copy Shall Do, its defined, it works
>> and its useful.
>
>So far, no one's suggested this as a requirement.

<eer> Actually, I think it is, for the reason you raise...
>
>I'd wonder though... that if your application has real-time constraints then
>you probably don't want to store your data in some loosly consistent store.
>
That's one of the reasons I've not been comfortable calling any writeable replica a master of the data...it's not,
it's just an updateable replica.  The master replica needs to be a designated replica to which special operations
can be directed, for instance when an application wants high confidence that tight consistency is being enforced...
this is most easily done by directing all such tight-consistency operations against the designated master replica.

Another use of the master replica is to treat it as the transaction coordinator for directory information (distributed
knowledge) operations which require tight consistency, such as adding or removing a replica from the known
set of replicas for a particular partition of the namespace, or to merge two partitions into a single partition,
or to create a new partition of the namespace.

High confidence has value in several scenarios and probably needs to be re-introduced into LDAP, perhaps
as a new control (which is really what it is...a control on the resolve name/get referral operation embedded
in the high level LDAP operation being performed (read, modify, create, whatever).  It would, of course, be
dependent on the available knowledge in the directory itself with regard to replicas and their types of the particular
partition holding the objects being so operated upon.
>Also... if an entry is mastered by many servers... then you won't know
>which server holds the 'correct' value anyway.  So it doesn't sound like
>this buys you much.
>


-------------------
Ed Reed, Technologist
Group Technology Office
Novell, Inc.
+1 801 861 3320