PRDB Minutes from Regional-Techs Meeting in Ann Arbor
Dale S. Johnson
dsj
Tue Oct 5 21:48:22 UTC 1993
These are notes from Monday's presentation at the Regional-Techs
meeting in Ann Arbor. Merit will be compiling some kind of results
from the meeting, but I'm about to be out on vacation for a couple of
weeks and I'd like to get these to Regional-Techs before I leave.
Apologies for the separate mailing. --Dale
Status of Merit PRDB support of CIDR (and other assorted enhancements):
--------- PRDB Support for CIDR:
"CIDR-1" (The ability to accept and propogate aggregates): Finished; ready
for testing & deployment when GateD is ready. (Includes NACR and auto-NACR
updates, gated generation, reports, whois support).
Input: Identical to old NACR, except that "x.x.x/len" is now supported
where "x.x.x" used to be supported. The old format will still
work: if the length is left off, this will be assumed to be
a Class A, B, or C as appropriate.
Output: In all published output, "x.x.x" is replaced by "x.x.x/len" .
(Exception: nets.unl.now has the fields separately. The
Right Thing is done within the GateD configuration files).
"CIDR-2" (The ability for NSFNET to create aggregates on behalf of its
peers): Targeted for mid-November, to be ready for BGP-4 in GateD.
Input: Identical to the old NACR, with the addition of zero or more
lines labelled "trigger"* following "netname":
netnum: 193/8 [exact]
netname: All_of_Europe
trigger: 193.0.0/24 exact
trigger: 193.1/16 refines
If trigger lines are present, then the "netnum" line may
contain an optional keywork "refines" or "exact" to modify
the way this aggregate is exported from AS-690.
The contents of the "trigger" line may be any classless IP
address ("x.x.x/len") which is fully contained within netnum,
followed by at most one of the words "refines" or "exact".
Output: Aggregates created this way will be included with all other
nets in all current reports of nets (whois and published files).
All aggregates created this way will be automatically entered
into the Aggregate Registry. This database will be searchable
by whois, reports will be published (ftp, gopher), and an
"exchange format" of the entire contents of the registry will
be published. (See Aggregate Registry discussion later).
A separate report of AS 690 Created Nets with their triggers
and full definitions (flags) will be produced.
--------- Other PRDB enhancements:
Whois Server: All the functionality of the "dbshow" clients is now
available via "whois -h prdb.merit.edu". "Whois help" for help.
Nets reports: "nets.unl.now" ("unload" format) and "nets.tag.now"
("tagged format") are now published containing all information
about nets.
Home AS: Very soon accepted in NACRs, and published in nets reports.
This inspired a lively discussion about how the community can
encourage each other to fill in this information, since it is
not required for NSFNET routing, and since we have about 13000
current networks registered without this information. But the
PRDB software is ready and willing to be used.
US State: Accepted in NACRs, and published in nets reports and in a new
nets.by.state report.
policy.as_690: Modified/expanded RIPE format of AS 690 AS policy, published
for automatic updating of RIPE database.
Shared Whois project: negotiations currently underway to modify our
Organization Address fields to be compatible with the NIC,
RIPE, and others. Then, to compare and update each others'
data. (See SWIP presentation).
CLNP Support: We have a preliminary schema and some initial forms and
report formats for supporting CLNP in the PRDB. Nodes and
Applications (with their addresses), RDIs, and contacts are
included. Gated will be generated; information will be
available via whois, ftp, and gopher.
Motif Support: We are looking for ways to speed development, and are
experimenting with a Motif GUI builder. Is there an X-based
application that would be most useful to you that we should
experiment on first? (Anyone want to trade notes on GUI
builders?)
--------- Aggregate Registry Development (Non-NSFNET Specific):
We spent quite a bit of time on this; details to follow. The "Aggregate
Registry" concept of RFC 1482 will be split into an AS-based "Routing
Registry" and a separate "Routed Net Registry", which will include
not only Aggregates but also any other "Classless Net" that needs
registration. (Possibly every net on the connected Internet?). All
such nets will be entered only by their Home AS or their creators
(in the case of aggregates); not by everyone who announces them. If
other organizations build similar registries, this one will exchange
information with them on a peer basis or otherwise simulate a full
distributed database solution.
-------- Upcoming non-backward-compatible changes:
It is likely that the format of the Organization Address field in the
new nets.unl.now and nets.tag.now reports may change as a result of the
SWIP ("Shared Whois") project reconciling Organization Address schemas
between Merit, the InterNic, RIPE, and others. Merit's single field
will lilely be broken into multiple independent subfields within the
current single field.
The format of all AS 690 announcement lists in all PRDB input and
output everywhere *might* have to change as a result of supporting
multiple gateways to single ASs. (E.g. a syntax something like:
1:200(NSS1) 2:200(NSS2) ). Then again, maybe not. Your opinions on
ways to minimize disruption in the reports are welcome.
Many Net reports are scheduled to go away Nov 1. See the file
nic.merit.edu:nsfnet/announced.networks/reports.summary for details.
More information about the NANOG
mailing list