> In Jim's definition, a cluster is static and a cloud is elastic. Not
> sure if those definitions are commonplace, but I find them practical.
>
> I think that partioning data across an elastic cloud is a major bit of
> code. Probably it could be layered into a "distributed disk server"
> component and the other layers of code would not be affected much.
> Would that lead to optimal perfomance? One would for instance give up
> the ability to let the optimiser make different plans for local & non-
> local data.

But if we talk about horizontal partitioning in such a way, that all
related data are always (ok, 90% of time) are stored on the same node,
for most queries we get situation when only one node will be affected in
read and in write operations. So we don't put one table on one node,
another one - on another node, but instead each node contains all
tables, only the data sets are different. Also you can do another
strategy: keep static data replicated to all nodes and partition the
dynamic data only.

Do I miss anything here?

Roman