Emergency Contacts ECON drafthardieecritiris02 Andrew Newton, VeriSign Ted Hardie, Qualcomm Hannes T

1
Emergency Contacts (ECON)draft-hardie-ecrit-iris
-02Andrew Newton, VeriSignTed Hardie,
QualcommHannes Tschofenig, Siemens

• Andrew Newton
• IETF ECRIT Working Group
• 2 August 2005
• Paris, France

2
Service Type

• Current approaches to service type are
  hierarchical lists.
• Certain functions are duplicated in the list.
• Hierarchy implies preference order.
• Do forest rangers have a policing function or
  just a fire prevention function?
• We have broken it down into
• Function
• Area
• Target
• We are not wed to this solution.
• Many of the approaches are questionable as to
  their true applicability to users and user
  interfaces.
• ltlistServicesgt query added to support
  locally-scoped queries.

3
Caching

• We added caching of answers by end clients.
• In the case of civic addresses
• If your civic address does not change within X
  number of minutes, do not requery.
• In the case of geo
• If your coordinates stay within polygon Y for X
  number of minutes, do not requery.

4
Database Replication in ECON

• We take no single position on database
  replication with ECON.
• It most likely will differ greatly throughout the
  world.
• Isnt it out of scope?
• But we have identified 3 methods of conducting
  database replication with ECON.
• Serialized database entries to a file as
  specified in IRIS.
• And the file transfer protocol of your choice.
  Many people like SFTP.
• ECONREP (ECON Replication)
• Interactive IRIS profile.
• Replication of entries before they become active.
• Incremental replication.
• Anything you find that works better for your
  situation.
• RDBMS replication
• Shared Network Memory
• Osmosis, crystal balls, and strong hope

5
Caching / Database Replication

• Problematic
• Are the requirements for a cache or for database
  replication?
• What are the aggregation properties?
• What about aggregation of data from different
  schemas?
• How are relationships managed?
• Do we really know all the requirements for
  specifying this?
• Intermixing server-to-server semantics with
  server-to-client semantics is bad.
• If a client detects a one millimeter change in
  location, does it requery?
• Charter Issue
• Isnt this out of scope for the working group?

6
Object Signing Considered Harmful

• My house is on fire. Who do I call?
• Please update your client with the proper trust
  anchors.
• My house is still on fire.
• Please cryptographically verify these URIs.
• Its getting hotter.
• Please check this CRL.
• Did I mention that my house is on fire?
• Object signing is useful for diagnosing problems.
• But that happens after the incident, not during.
• All the mechanisms to get object signing to work
  seem to be a pretty heavy price to pay for a
  diagnostic tool.

7
Dont Believe the FUD

• IRIS was created in the CRISP working group by
  TLD operators.
• We know a thing or two about high resolution
  loads, operations of highly available services,
  and moving data around the globe.
• If simple framing of XML is a concern, then we
  have bigger problems to worry about in ECRIT.