Talk by Liuba Shrira (A modular "alway there" and "not in the way" past state system)

TITLE: A modular "alway there" and "not in the way" past state system

SPEAKER: Liuba Shrira, Brandeis University

DATE and TIME: Friday, October 14, at 11 am

LOCATION: ICCS X836 (CS Boardroom)

HOSTS: Bill Aiello and Andy Warfield

ABSTRACT:

Demanding historical data analysis for forecasting and forensics, formerly dependent on specialized data warehouses and temporal databases, can become available in everyday data stores.

The challenge is how to add efficient support for ``always there''  and "not in the way" historical states without extensive prohibitively costly modifications to the data store internals.

Our approach integrates a low-level consistent snapshot system into a data store, allowing to run data analysis programs against long-lived past states from normal applications running against the current state.

The approach is attractive for several reasons.Frequent snapshots do not disrupt performance, unneeded snapshots can be garbage-collected, and snapshots can be accessed efficiently, independent of history length.Importantly, the snapshot system can be implemented in a modular way by a modest extension of standard data store protocols.

The talk will describe some of the underlying techniques and present some promising preliminary performance results from a prototype we built in Berkeley DB.

(This joint work with Ross Shaull.)

BIO:

Liuba Shrira is a Professor in the Computer Science Department at Brandeis University, and is affiliated with the Computer Science and Artificial Intelligence Laboratory at MIT. She has been affiliated with Microsoft Research, Cambridge, UK, Microsoft Research Asia, Beijing, and Computer Science Department, in the Technion, Haifa.Her research interests span aspects of design and implementation of distributed systems and especially storage systems. This includes fault-tolerance, availability and performance issues. Her recent focus is on long-lived transactional storage, time travel (in storage), software upgrades, and support for collaborative access to long-lived objects.