Security Mechanisms
charter-ietf-secmech-01

There exists a disconnect between the IETF's security frameworks.
Although these frameworks have very similar goals, the set of
mechanisms available depends upon the choice of framework. There are a
number of issues that make a compelling case for converging the way we
develop mechanisms for these frameworks.

• The actual mechanisms in each of these frameworks have very similar
  goals of authentication and establishing a cryptographic context. In
  many cases frameworks are developing new functionality that bring them
  closer together. An example of this is the addition of a PRF API to
  access key material from GSS-API.

• There is a desire to standardized EAP mechanisms and there currently
  is no working group with this on its charter. It would be desirable to
  work on this in conjunction with other efforts in the security area
  including work on GSS-API enhancements in KITTEN working group and SASL
  enhancements in the SASL working group.

• There is pressure to adopt a particular framework because of the set
  of mechanisms available not because of the capabilities and upper-layer
  interface of the framework. This recently was an issue with ISMS, but
  there has also been a desire to use EAP to authenticate other
  applications. We should be in a situation where the choice on mechanism
  was dictated by the deployment requirements and the choice of framework
  dictated by protocol design and implementation simplicity.

• There is a duplication of effort in the development of security
  mechanism that support similar credential types and infrastructures.
  This is problematic because the development of security mechanisms is
  both difficult and time consuming. It would be good to leverage the
  work and expertise required for developing a mechanism across all the
  frameworks.

• Often the cost of deploying a security mechanism is in the
  infrastructure and not the implementation of the mechanism itself.
  There limited set of mechanisms available to particular frameworks
  makes the coordination and administration of security between
  applications that use different frameworks more difficult.

The first tasks of a SECMECH working group would be to document a set
of evaluation criteria/guidelines explaining what standards-track
security mechanisms need to do and how we will evaluate them and to
document how we're going to go about specifying a security mechanism
for use in the frameworks that are in scope. The working group would
also be chartered to define a set of standards track mechanisms. The
mechanism work would complete after the first two tasks have completed.