[Interop-dev] Decentralization

[Federico Capoano]

How could we conceive a decentralized database model?

Here some of our networks (we call them "islands" cos on they map they
look like isolated networks) will need to access the device via SSH.
Many people already talk about developing something similar to
aircontrol in the near future to manage the network.

Of course it wouldn't be smart to put all the passwords of the devices
of the entire network of an entire country on a single database. We need
to be able to split it. Decentralization is also needed so the checks on
the devices (SNMP for example, but in Slovenia they made a custom way to
retrieve stuff via HTTP) can be performed in autonomy by every "island".

We mostly all need the nodes to be grouped, nodewatcher call it project,
some other nodeDBs call it zone or area.
So that you can have project "Rome", project "Pisa" and project "Milan",
until here is easy.
Suppose Pisa wants to be autonomous in managing their nodes, they might
want to host their database on one of their machines in their network
and performs monitoring checks in complete autonomy.
Suppose instead that Milan doesn't want to do so because they don't yet
have enough expertize or volounteers to do it, so they want to stick
with the main database until one day they might switch.

Since we are conceiving something new, we are doing it for our future,
it would be a good idea to start with decentralization in mind. What do
you think?


*Federico Capoano*
Web Designer & Web Developer
Portfolio/Blog: nemesisdesign.net <http://nemesisdesign.net>
Twitter: @nemesisdesign <http://twitter.com/nemesisdesign/>
PGP Key ID: 308BD46E
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.funkfeuer.at/pipermail/interop-dev/attachments/20120509/7135d010/attachment.html>