Dear RIPE local-IR WG member,
you chairperson has received (:-) the following request. Since this
proposal is very much relevant to the local-ir WG I would appreciate
your comments however short.
Thank you
Daniel
------- Forwarded Message
Date: Wed, 26 May 1993 13:28:44 +0200
From: Daniel Karrenberg <Daniel.Karrenberg(a)ripe.net>
To: RIPE WG Chairpeople <wg-chairs(a)ripe.net>
cc: RIPE Chairpeople <chair(a)ripe.net>
Subject: PRIDE Project
As you know the current route server project will end in July this year.
In order to contnue work on the routing registry I have put together the
following project proposal called PRIDE which I will also put in the
RIPE document store shortly.
Already some national members of RARE have indicated their willingness
to fund part of the project provided that the RARE technical committee
agrees that the project is useful, which I am confident they will do.
At the same time it is neccessary that the RIPE community discusses this
and comes to a position on it. Since time is pressing I would
appreciate comments from you personally and your working group no later
than Monday June 7th. It would be ideal if the chairpersons of the WGs
and the RIPE chair could agree by then that RIPE supports this proposal,
which would give it added weight.
Note that I regard the proposal text to be finished and I am reluctant
to make drastic changes to this version. Clarifications and additions
to make it clearer as well as small additions are of course welcome.
Even more welcome are offers of contributions to the funding, however
small.
Daniel
P R I D E
Policy based Routing
Implementation, Deployment in Europe
A Project Proposal
Daniel Karrenberg
Manager RIPE NCC
- 2 -
The Need
In todays Internet environment, policy based routing
technology providing routing of traffic between different
network operators is a key technology. While tools are
available to apply routing policy they are not used in a
coordinated way if used at all. In a general mesh topology
applying policy without coordination and prior simulation of
the consequences will eventually lead to an unmanageable
situation. This potential problem has been noted in [1] and
several of the references cited therein.
As an immediate measure the GIX (Global Internet Exchange)
has been proposed in [2] and a proposal for a GIX routing
implementation has been made in [3]. It is noteworthy that
the GIX only solves part of the problem: routing consistency
and connectivity at the global level. In order for the
Internet to cope with its current growth, the routing
problem will also need to be solved at the regional and
local levels. The "RS" project (1) has recognised this by
spending significant resources on establishing consensus
within RIPE on how routing policies are stored in the
European routing registry[4]. Once registered the
information can be used to ensure proper operation of the
European part of the Internet. The project is making
significant progress in this area. The routing registry and
tools being developed are much more general than those used
and proposed so far for use in the current Internet. It is
expected that the architecture will eventually allow
multiple major interconnect points as envisaged in [1].
_________________________
1) "Implementation of a Route Server for policy based
routing across the GIX", a joint RARE/RIPE project
currently funded by SURFnet.
- 3 -
In order to promote the European routing registry and the
associated technology two key ingredients are needed:
Implementation
A set of tools for use by local network operators needs
to be developed. The "RS" project deals only with the
tools needed by the route server itself. While some of
these can be adapted there are not sufficient resources
to properly produce tools for local network operators.
These tools will enable them to use the routing policy
stored in the routing registry to perform such tasks as
check actual routing against policies defined, ensure
consistency of policies set by different operators, and
simulate the effects of policy changes.
DEPloyment
In order to be useful the routing registry and
associated tools need to be deployed rapidly by all
significant network operators in the European Internet.
This means there is a big need for information and
training of the network operator staff, coordination of
deployment and support activities. If enough
information and education pressure can be applied there
is a good chance that the technology will be deployed
outside Europe as well. First signs of this are
already visible as the CIX (Commercial Internet
eXchange) association has announced their intention to
deploy a route server using the RIPE routing registry
technology.
The urgent need for these two ingredients motivates this
project proposal and has suggested the name PRIDE: Policy
based Routing Implementation and Development in Europe.
- 4 -
The Results
The tangible results of the project will be the following.
Implementation
In the implementation area of the project, tools will be
developed, documented and made publicly available for use by
network operators. A complete list of those tools cannot be
specified in advance since specific needs are likely to
evolve during deployment. Flexibility to meet those needs
and agree about priorities with the network operators is a
key element to ensure acceptance and thus the success of the
project.
The tools are expected to include:
prcheck A tool to check the consistency of routing
policies stored in the routing registry.
This tool will flag if two neighbouring
network operators specify conflicting or
inconsistent routing information exchanges
with each other and also detect global
inconsistencies where possible.
prpath Extract all (AS-)paths between two networks
which are allowed by routing policy from the
routing registry.
prconn Display the connectivity a given network has
according to current policies. This will of
course also be able to find the set of
networks a given network can not reach.
prtraceroute A version of the existing traceroute tool
which will be able to display whether a route
in use is allowed by policy and where
deviations from policy occur.
The range of implementable routing policies is currently
limited by the destination based routing and forwarding
technology. There are efforts underway to enable forwarding
decisions based on the source of packets as well as the
destination. The routing registry and tools will need to
both follow and influence developments of destination based
routing and forwarding in IPv4 as well as next generation
IP.
- 5 -
Deployment
This is the key part of the project. Without widespread
deployment of the routing registry at least in Europe the
results of the "RS" project and the Implementation part of
the PRIDE project will be academically interesting but not
much more. The result of this part of the project is thus
widespread deployment of the routing registry and associated
tools as possible.
The tangible results will be:
- instruction and training material about the routing
registry as such and the way routing policies need to
be expressed for registration
- delivery of the information and training to key
communities in Europe
- coordination of the actual deployment of the tools and
especially the registration of routing policies in the
routing registry
- general presentation material about the routing
registry
- delivery of the presentation material to key
communities worldwide
The delivery of training and coordination and deployment
will consume the bulk of the project resources. It should
be noted however that the proposed resources are not
sufficient for a general support or help desk function.
This would need significantly more resources. Because we
believe network operators will invest here in their own
interest we propose to focus on targeting information and
training well and to provide coordination only. The RIPE
community will then form the network for mutual support as
it has done successfully in the past.
- 6 -
The Partners
In order to realise the goals of this project, close
cooperation with as many service providers as possible will
be necessary. European service providers are already
following and influencing the developments closely through
RIPE. Both the amount of input received during the design
of the routing registry and the rate at which information is
being registered show this works well.
Worldwide contacts with groups involved in similar
developments are also already established via RIPE, IEPG and
IETF. The close coordination of all parties deploying route
servers on the GIX and the recent announcement by the CIX
association of their intention to use the RIPE routing
registry technology are good examples of this.
The project will exploit these already existing channels and
be open to new ways of reaching the service providers in
particular. All service providers will have equal access to
the tools, the routing registry itself and the information
and training materials.
- 7 -
The Resources
The resources needed for the PRIDE project are estimated as
follows:
- 2 senior engineering staff for 12 months representing
24 FTE months. These will need renumeration equivalent
to senior network engineer levels including overheads.
The estimated cost of this is 84kECU.
- The deployment part will necessitate significant travel
for delivery of the information and training. The
estimated travel cost is 20kECU.
- Professional document design for the materials is
desirable. It is difficult to estimate the cost for
this at this point but 10kECU should be sufficient.
- Computing resources. Minimally a personal WS and some
storage capacity. 10kECU will be sufficient.
- 1/3 of senior technical management for 12 months.
Technical management of the current RS project is
provided by the RIPE NCC. This withdraws resources
from NCC core activities which need to be replaced.
Estimated cost of this is 20kECU.
The total project cost for 12 months is thus estimated at
84+20+10+10+20 = 144kECU.
The above resource levels are purely for the work specified
and do not include formal project management and formal
(non-technical) reporting.
After completion of the project and successful deployment of
the routing registry, a level of maintenance effort will be
needed for the tools and the routing registry. This should
be a structural activity much like the current RIPE NCC core
activities.
In order to start up quickly the implementation part can be
started first as a sub-project with one engineer. It should
be noted however that without the deployment effort
following, this will have only a very limited effect.
- 8 -
References
[1] T. Kalin: "Global Network Interconnect", Amsterdam, 8
Jan. 93, EC(92)093v3
[2] G.Almes, P.Ford, P.Lothberg: "Proposal for Global
Internet Connectivity", IEPG Working Document, June
1992
[3] Tony Bates, Daniel Karrenberg, Peter Lothberg, Bernhard
Stockman, Marten Terpstra: "Internet Routing In a Multi
Provider, Multi Path Open Environment", Document RIPE-
82, March 1993
[4] Tony Bates, Jean-Michel Jouanigot, Daniel Karrenberg,
Peter Lothberg, Marten Terpstra: "Representation of IP
Routing Policies in the RIPE Database", Document ripe-
81, March 1993
------- End of Forwarded Message
