1. Policies

We discussed the question of gathering more information from network providers 
and users about the kind of policies that they (might) apply and expect. The 
conclusion was that I would spend some time talking to people (if I could find 
appropriate people!), but that it would not be a major deliverable.

2. Traffic management

2.1. Management within the network

We agreed that I would spend 1 month investigating what the fore switch in the 
department can do - looking in detail at the MIB and then writing a report. We 
will collaborate with Cambridge on providing SNMP manager access. Writing 
SNMP manager s/w is something I can do, but it is not very appropriate use of 
resources. We will decide whether we invest programming effort later - if 
Cambridge want to write this, that is fine :)-. I will also look into whether we can 
buy some SNMP manager s/w as this is probably the cheapest solution - I will 
talk to jon about budgets.

In terms of traffic management, we are both more interested in end system 
management (stack and applications). We discussed what monitoring information 
would be useful - this will come out of the work in (1).

2.2. Protocol stack management

Ian's project is approaching this by a providing an adapter layer above the stack 
rather than modifying the stack itself. I asked how one would control existing 
applications transparently (ie without changing their code). We discussed our DTI 
project and Ian said that BT would be interested in joing the project on a limited 
basis - not asking the DTI for funds, contributing a small amount of time coming 
to project meetings (eg 10 days), possibly also providing access to a BT network 
testbed. Ian took the S2 form away - I will send an email to all parties in the 
project shortly.

We agreed that managing the stack would be a major part of the work in the first 
part of the project (no specific timescales as yet). Exact technical approach will 
depend on whether DTI funding comes through and examination of 2 approaches 
to stack management.

2.3. Applications management

Ian described the various applications being developed at BT. We discussed the 
possibility of writing a proposal for funding where we involve end user partners 
with distributed applications - Greg Brougham of Tadlon is working at BZW, 
Communic8 know Barclays, I can talk to one of the more enthusiastic people at 
Texaco etc.


3. Policy language and network management models and policies.

Along with protocol stack management, this would be a major part of the work in 
the first part of the project (again no specific timescales as yet).

We spent some time discussing our work to date. I described how the OSIMIS 
GDMO compiler works and gave Ian a copy of the technical information. I 
described the ASN.1 SNMP MIB interpreter and showed how MIBs could be 
automatically compiled and interpreted cdoe executed within an agent. I have 
been working on a report which will be ready shortly.

The main themes discussed (which I will focus on in the report) is that a policy 
language has a number of functions:- it is an interpreter for a specification 
language that enables interpreted code (behaviour) to be added to a MIB; the 
same policy language should be capable of working across different management 
models and the syntactic conventions of each model (eg IDL or ASN.1 encoding) 
should be entirely transparent to users of the policy language; the interpreter 
should understand the constraints of different models (eg different naming 
conventions) and check for correctness (the constraint implemented in the SNMP 
MIB interpreter is that variables have to be  valid MIB identifiers); the interpreter 
needs a flexible way of interfacing with the features of the particular DPE etc.

We discussed possible platforms - using the ASN.1 SNMP MIB interpreter in 
simpler solutions (small machines, network rather than applications management 
etc) and using IDL (Orbix) and BT's environment. We can use the report on the 
ASN.1 MIB interpreter as an initial specification of what is required of the 
platform (eg can we add symbolic information to objects so that they can be 
interpreted). Jon and I will discuss budgets to see how much we can afford.

When the report is finished and Ian and his team has looked at it, I will go to 
Ipswich for a meeting - I suggest the last week in November.

After that, we will arrange a meeting with Imperial to discuss policy languages.