Fear and Loathing at the IETF Fear and Loathing at the IETF (Ten - - PowerPoint PPT Presentation

Sepucha_Date_01

Fear and Loathing at the IETF Fear and Loathing at the IETF (Ten Years with No Pay) (Ten Years with No Pay)

Note: Note:

NEVER joke about your presentation title with the Panel Chairper NEVER joke about your presentation title with the Panel Chairperson!! son!! My apologies to HST My apologies to HST

Sepucha_Date_02

IETF Security Standards IETF Security Standards Challenges and Opportunities Challenges and Opportunities

Hugh Harney

SPARTA, Inc. 7075 Samuel Morse Dr. Columbia, MD, 21046 410-872-1515 x203

Sepucha_Date_03

Agenda Agenda

• Are the ground rules changing?

– Are we forgetting end to end? – Are we passing adequate policy information? – Are the mechanism flexible enough?

Sepucha_Date_04

Are we forgetting end to end system design? Are we forgetting end to end system design?

• Store and forward systems need

system level mechanisms.

– Service desired is secure end to end. – Communications push us to store and forward solutions. – Pairwise connections to servers are not end to end secure. » Source authentication » Data integrity » Does the data source really know the access control rules?

Router Router Router Router Router Router

Peer Peer Peer Peer Peer SECURE Server

? ?

END-TO-END ARGUMENTS IN SYSTEM DESIGN J.H. Saltzer, D.P. Reed and D.D. Clark*

Sepucha_Date_05

Adequate policy information? Adequate policy information?

• Mobility makes it difficult to have

static policy for a single identity – Policy changes based on location – Policy changes based on environment

• Large efforts are adding policy

negotiation capabilities to IKE for network layer SA s

Sepucha_Date_06

Adequate policy information? Adequate policy information?

• Policy is commonly assumed to flow from a trusted authority
• In truth, policy is composed of many local policy authorities

with conditional dominance – This is true for security access control / authorization policy, and for information flow policy. » E.g., an ISP in two different countries have different supported mechanisms. – The Data policy may need to transcend the local policies to find adequate protection services. » E.g., Country X may require an algorithm set that does not meet the protection requirements, the security policy would have to interact with the routing policy to ensure protection.

• Standards are needed to expand the definition of policy

negotiation. – Coalitions between countries and industry are prime examples.

Sepucha_Date_07

Adequate policy information? Adequate policy information?

• Dominance based policy is being expanded.

– Policy decisions based on a composed policy. » Multiple policy authorities. » Conditional rules based on location and environment. » Turning complete languages? – Lattice based policies are more adaptable to event based information policies.

• These efforts need additional policy payloads to exchange system

level information. – There is working code and working systems – Need this type of work to filter into the IETF

• We really can’t keep the one Certificate Authority model for policy

anymore. – Security policies need to be written with a mind toward multiple authorities and mechanisms.

Sepucha_Date_08

Are the mechanism flexible enough? Are the mechanism flexible enough?

• Flexible mechanisms allow us to create standards that stand the

test of time. – Can the standard provide algorithm flexibility? – Does the standard provide hooks for additional services? – What happens when algorithms age? – Intelligent flexibility can enhance performance. » Simple doesn’t always equate to fast » Service modes allow for optimized service modes – TCP MD5

• No key id field: static keying
• No service options: one mode
• Not authenticated TCP: MD5

– If you need to use something other than MD5 write a standard.

• Use of RSA signatures in IPSec ESP and AH may have the

same issues.

Sepucha_Date_09

What will help? What will help?

• The IETF core principles are sound,

– Working code and rough consensus.

• The group of participants have changed.

– Focus on products, not the services that the products provide – Users are absent, mostly.

• We need more user based challenges pushing the

standards – Work is going on, we need the users to provide input into the standards process.