Open Stack

Hello everyone.

As far as I'm concerned, a neutron network is actually a pure virtual
concept layer which holds couples of subnets. Subnets are the ones connect
and provide virtualized network access, internal ip arrangement, and basic
layer-2 isolation. When using a GRE tunnel mode, the isolation between
tenants' networks will be done by "Tenant Network ID", which provide a
layer-3 isolation.
However, it just makes me curious that, what does a shared network do? By
design it should isolate network flow on layer-3 level, which means only
instances within the same network will have the chance to communicate with
each other. As I can see, a shared network may allow different tenants to
access the identical network resources created by others. But what about
the connectivity? The network flow may be isolated by both the network id
and the tenant id for instances owned by different tenants in a shared
network. So what does the network actually shares? Only the "fixed IP"
arranged? If instances cannot communicates to each other then why bother to
share a network?

If I'm wrong at any point please guide me. Thanks in advance.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140918/87620ac6/attachment.html>