IDD Topology Reconfiguation Plan
Mitch Baltuch
May 23, 1995
Overview
The original topology for the IDD was put together before Unidata had much
practical experience with large volume dataflows over the internet. Also,
the very nature of the internet was changing due to the switch from a
federally funded and run network, to a commercially run network. This change
coincided with the implementation of the IDD. As a result of both increased
experience and knowledge on the part of Unidata, and the changes brought
about by the reorganization of the internet, It has now become necessary to
reconfigure the IDD topology.
This will not be a one-time only reconfiguration. The internet is dynamic,
rather than static, and this will always necessitate that we monitor the
overall network state and modify the topology as needed. However, we can
implement a new topology that is well thought out and valid for current
conditions, such as they are. Once the new scheme is in place, future
modifications should be less sweeping and easier to implement.
Implementaton Steps
The following steps need to be taken to take to accomplish the topology
reconfiguration. Each of these steps will be done on a tier-by-tier basis.
In other words, the topology will be implemented for tier, starting from the
top. Interim lower tier changes will be done only to accomodate the upper
level changes.
For Each Tier
- Design preliminary target topology
- This entails designing a topology based on some initial assumptions
that may, or may not be valid. Testing in the next phase will confirm
the assumptions, or force changes. In addition, negotiations with
possible new relay sites will be done.
- Run connectivity tests at all sites
- Using monitoring scripts already in place at IDD sites, we will ask
all of the sites to check connectivity with selected downstream
sites to test the assumptions made in the preceding step. This will
be an iterative process depending on what the results from the test
show.
- Design final topology configuration
- Using the results of the testing, we will design a topology that
gives up the most efficiency in data transfer.
- Implement topology configuration
- Implementation will be coordinated beforehand with all sites involved.
During the implementation, each site will also be assigned formal
failover sites and a list of all sites, their primary and failover
feeds and the site administrator information will be listed and made
available on the Unidata Web server.
Milestones
- First Tier Sites July 14, 1995
- peliminary target topology May 26, 1995
- connectivity tests June 23, 1995
- final target topology June 30, 1995
- implementation July 14, 1995
- Second Tier Sites October 13, 1995
- peliminary target topology August 18, 1995
- connectivity tests September 15, 1995
- final target topology September 29, 1995
- implementation October 13, 1995
- Third Tier Sites December 8, 1995
- peliminary target topology October 20, 1995
- connectivity tests November 10, 1995
- final target topology November 17, 1995
- implementation December 8, 1995
- Fourth Tier Sites February 9, 1995
- peliminary target topology December 15, 1995
- connectivity tests January 12, 1996
- final target topology January 19, 1996
- implementation February 9, 1996
Mitch Baltuch <mitch@unidata.ucar.edu>
Last modified: Tue May 23 10:08:33 1995