Requirements for the Dynamic Partitioning of Switching Elements
draft-ietf-gsmp-dyn-part-reqs-03

2003-03-13 - note from randy
To: Avri Doria
Cc: Scott Bradner , Bert Wijnen ,
  IESG Secretary
Subject: Re: [Fwd: Re: draft-ietf-gsmp-dyn-part-reqs-03.txt]
References: <3E70A586.1050206@acm.org>

>>    as requirement 3 allows starvation of resources such as cpu,
>>    perhaps this needs to be mentioned in sec cons
>>
>
> As i mentioned below it don't see how it is a security consideration
> since only a trusted PM has access.  I can see it happening through
> error, but not malice.

you're right.  i withdraw my objection

and i do know what real nits are

randy

2003-03-06 - randy points out that one of his comments
was not addressed in revision - sent note to WG chairs

randy points out that one of his comments on draft-ietf-gsmp-dyn-part-reqs-03.tx
t
was not addressed

  as requirement 3 allows starvation of resources such as cpu,
  perhaps this needs to be mentioned in sec cons

this needs to be fixed before he will let it go

while you are at it please fix the Intellectual Property Considerations
to say what RFC 2026 sec 10.4(D) says it should say

Scott

2003-03-04 - comments from WG in response to IESG comments

Specific replies to the IESG comments below:

--

Q. IESG Evaluation  Comment:2002-10-03 - iesg discussion - comment to WG
are SEs only layer 2, or can a virtual router be an SE?

A. Since support for ForCES is not covered in the current
requirements, no.  If ForCES support were to be added to the
GSMP charter then the answer would be yes.

---

Comment: microsoftisms in the text

A: oops.  Fixed.

---

Comment: in intro para 2, the enumeration omits the case where a single
logical SE or controller might be implemented by multiple devices

A: Further information was adding of more possibilities for SEs.  This is
now in Para 3.

---

COMMENT: in the discussion

  Dynamic Partitioning

  Static repartitioning of a SE can be a costly and inefficient
  process.  First, before static repartitioning can take place, all
  existing connections with controllers must be severed.  When this
  happens, the SE will typically release all the state configured by
  the controller.

you might make clear that one or more static partitions of the SE
may not be affected by the change(s) and hence would not be
disturbed.  e.g. one could have an SE with O(10^3) partitions and
only be mucking with a few.


A. The text has been added to cover this point.

---

COMMENT: as requirement three allows starvation, this needs to be mentioned
in sec cons

A: Requirement 8 was added to disallow for the possibility of starvation.

---

COMMENT: sec cons says

  Only authorized PMs MUST be allowed to dynamically repartition a
  SE

etc.  but there is no hint of security relationships.  are SEs
statically bound to PMs and vice verse?

A: Text has been added to indicate how the relationships between
PM CE/SE can be established.

---

COMMENT: what are the implications of a requirements document having ipr?

A. We couldn't figure out how and so removed the statement.

2002-10-03 - iesg discussion - comment to WG
are SEs only layer 2, or can a virtual router be an SE?

---

microsoftisms in the text

---

in intro para 2, the enumeration omits the case where a single
logical SE or controller might be implemented by multiple devices

---

in the discussion

  Dynamic Partitioning
   
  Static repartitioning of a SE can be a costly and inefficient
  process.  First, before static repartitioning can take place, all
  existing connections with controllers must be severed.  When this
  happens, the SE will typically release all the state configured by
  the controller.

you might make clear that one or more static partitions of the SE
may not be affected by the change(s) and hence would not be
disturbed.  e.g. one could have an SE with O(10^3) partitions and
only be mucking with a few.

---

as requirement three allows starvation, this needs to be mentioned
in sec cons

---

sec cons says

  Only authorized PMs MUST be allowed to dynamically repartition a
  SE

etc.  but there is no hint of security relationships.  are SEs
statically bound to PMs and vice verse?

---

what are the implications of a requirements document having ipr?