newsNode Manager - Community IP/Name Management - News: New Data Model

Latest News
Italian support! posted by JamesSchopp, Wed 14 Aug 2002 01:32:25 AM UTC - 0 replies
New Data Model posted by JamesSchopp, Wed 14 Aug 2002 01:15:41 AM UTC - 0 replies
[Submit News]
[2 news in archive]

New Data Model

Item posted by James Schopp <JamesSchopp> on Wed 14 Aug 2002 01:15:41 AM UTC.

I am investigating a completely new data model that will make NodeManager MUCH more flexible. NodeManager was initially designed to handle just a single community (like MadridWireless). However, the need is arising to support several groups at the same time, which means more IP pools, and more administration/management.

The new datamodel has been set up at a test site here:

Here is a description of the new datamodel with an explanation:

Imagine 2 tables, initially in the following state:

Available Subnets             Allocated Subnets
-----------------             ----------------- <empty>

Then, some jerk asks for a subnet that supports just 13 hosts. The tables would then be in the following state:

Available Subnets             Allocated Subnets
-----------------             -----------------       

We allocate the subnet with the number of hosts required, but then regroup the remaining addresses in the best way possible.

If someone asks for a subnet that supports, say 100 hosts, then we would find the available subnet that has the smallest mask that supports at least that many hosts (25bits, 128 hosts, which would be in this case). If a subnet is available with EXACTLY that subnet mask, then we just move it into the "allocated" table. If none was found, then we must find the next largest subnet, and break it down as we did above.

If an allocated address has been freed, then in must be moced back into the available table. However, this means that we might have the ability to re-consolidate smaller subnets into larger ones. For example: imagine that the same jerk then realizes that for administrative reasons it was bad to allocate so few hosts from such a large pool. We then need to aggregate it all back into

This would allow for some "international group" to allocate an address space to "RedLibre" (for example). They could turn around and allocate addresses to Madrid, who in turn allocatse address for a specific node owner. All of this could be managed through a single site.

No messages in New Data Model


Back to the top

Powered by Savane 3.5