Our current mission is to develop a feedback/feedforward infrastructure for run-time monitoring and repair/reconfiguration of component-based distributed systems.

The DARPA Dynamic Assembly of Systems for Adaptability, Dependability, and Assurance (DASADA) Program involves research into software probes and corresponding measurement gauges.  The program is developing a standard for the structure of probe events that will be processed by gauges.  Columbia University’s Programming Systems Lab, OBJS, BBN, and other DASADA participants have developed an initial version of the proposed schemas.

An example message using the schema is located here.  The schemas are intended to function as SOAP blocks.  A SOAP message has a header section, a body section, and an optional faults section.  Each section can contain one or more blocks.  Further information on SOAP can be found at http://www.w3.org/2000/xp/.  Current usage is to put the context block in the header and one or more content blocks in the body.  There are currently two content blocks defined, one very low level, one very high level.  It is assumed that most users will want to define their own content blocks as well.

The low-level probe content block contains information about a particular function/method call, including name, parameters, value of “this,” return value, and exception information.  The high-level probe content block identifies an architectural mutation, involving components, connectors, and so forth.  A possible usage of this probe format is for the probe to generate low-level events, which are augmented by successive processing with higher-level information.

• Autonomizing Legacy Systems (talk at Almaden Institute Symposium, 4/02)
• An Active Events Model for Systems Monitoring (paper at CDSA, 12/01)
• Available Technologies (ppt) (pdf)
• Year 1 Coordination Plan
• Kickoff Briefing (ppt) (pdf)
• Revised Technical Proposal
• Overview Slide (ppt) (pdf)

MEET is Columbia’s Multiply Extensible Event Transport. It aims to provide an easy-to-use publish-subscribe infrastructure with the following features:

• Scalability: 10⁶ – 10⁹ publishers and subscribers.
• Survivability: arbitrary connection topology, with rapid switchover to alternate routes. Redundant distributed storage of routing information.
• Extensibility:
  • new datatypes and operations can be defined for event filters. An example is the XML processing filter which allows XPath expressions to be used as predicates over XML fields. A subset of operations will be realtime compatible.
  • routing policies are extensible. Event publishing / multicast is an open problem, and specific algorithms may be optimal for particular situations.

Worklets micro-workflow mobile code adapts computation to component context

Workflakes integrates Cougaar (BBN) macro-workflow for community coordination

Process-aware system repair: KX “continual coordination”

• Worklets Flyer (6/01)
• Mobile Agent Approach to Process-based Dynamic Adaptation of Complex Software Systems
• Process-Orchestrated Software: Towards a Workflow Approach to the Coordination of Distributed Systems (PhD Thesis Proposal, 5/00 Talk)
• see Cougaar

A joint project with Profs. Gail Kaiser,  John Kender and Jason Nieh.

Flyer (6/01)

Screenshots:

Flyer (6/01)
Old flyer (6/99)

CHIME: A Metadata-Based Distributed Software Development Environment (ESEC/FSE, 9/99)
– Paper
– Talk

Distributed Software Development with CHIME (SE over Internet wksp, 5/99)
– Paper
– Talk

CHIME: A Metadata-Based Distributed Software Development Environment (Telcordia, 6/00)
– Talk