openstack team mailing list archive

[Ed Leafe <ed.leafe@xxxxxxxxxxxxx>]

On Feb 19, 2012, at 12:53 PM, Mark Washenberger wrote:

> For this reason, whatever name we choose I would hope we prefix it with "compute-" (i.e. compute-zone or compute-cell) so that we aren't letting language trick us out of some of our better implementation options, such as allowing deployers to scale compute, volume, network, and api resources separately.

	I certainly don't want to preclude implementation options, and like the clarification that this is an issue of scaling compute, but I have two problems with the use of a prefix. First, it indirectly implies that scaling other entities would be done in the same manner: e.g., that an api-cell would have its resources independently deployed and with the inter-cell communication design as compute-cell. If the desired outcome is that we encourage different entities to implement their scaling independently and in the best manner for that entity, having a common name would seem to encourage the opposite.

	Second, it just seems cumbersome. We should ensure that documentation about cells is clear that this is a way of scaling compute, but for referring to them in code and in discussion, a simple name like 'zone' or 'cell' is simpler and cleaner.


-- Ed Leafe