Subject | Re: [Firebird-Architect] Cloud database |
---|---|
Author | Jim Starkey |
Post date | 2008-07-23T21:15:45Z |
paulruizendaal wrote:
which should be intuitive to the well trained practitioner of
multi-generational concurrency control. Certain things have to appear
to be serialized -- sequence values, unique indexes, updates to a
particular record -- but most things can happen in parallel perfectly
happily as long as each transaction sees a consistent view of the
database. That fact that transactions on different nodes will see
different database states is no different than a pair of concurrent
transactions on a single system. So we don't need serialization to
maintain database consistency (once a heretical thought). This opens up
all sorts of opportunities for batched, asynchronous replication
messages and a great opportunity for careful bookkeeping.
It has long been my opinion that there are two types of people, those
two break all questions into a small number of categories, and those who
think there are an infinite number of ideas, most of which haven't yet
been thought.
--
Jim Starkey
President, NimbusDB, Inc.
978 526-1376
> Hi all,There are alternatives in between his "lazy" and "eager" models, some of
>
> Sometime ago Jim discussed cloud databases. I came across a paper
> that folks interested in that topic might find a good read: "Don't be
> lazy, be consistent":
> http://www.cs.mcgill.ca/~kemme/papers/vldb00.pdf
>
>
which should be intuitive to the well trained practitioner of
multi-generational concurrency control. Certain things have to appear
to be serialized -- sequence values, unique indexes, updates to a
particular record -- but most things can happen in parallel perfectly
happily as long as each transaction sees a consistent view of the
database. That fact that transactions on different nodes will see
different database states is no different than a pair of concurrent
transactions on a single system. So we don't need serialization to
maintain database consistency (once a heretical thought). This opens up
all sorts of opportunities for batched, asynchronous replication
messages and a great opportunity for careful bookkeeping.
It has long been my opinion that there are two types of people, those
two break all questions into a small number of categories, and those who
think there are an infinite number of ideas, most of which haven't yet
been thought.
--
Jim Starkey
President, NimbusDB, Inc.
978 526-1376