"We dug up old memos and reverse engineered the code"

Dr C Mohan is best known for his pioneering work on database transactional
management. He is the primary inventor of the ARIES family of recovery and
locking methods, and the industry-standard Presumed Abort Commit Protocol. He
was recently at the IBM Research Lab in New Delhi, where he spoke to Bhaswati
Chakravorty of Dataquest.

How did you get started on the Algorithms for Recovery and Isolation
Exploiting Semantics (ARIES) project?

In the mid-80s, basic Concurrency Control and Recovery (CC&R) was
believed to be a dead research topic by general DBMS community. After the R*
project at IBM Almaden, Starburst was initiated to design a new extensible DBMS.
A few of us (non-Sys-R) decided to revisit Sys R legacy and found major unsolved
problems in the area of CC&R. Important algorithms were left undocumented.
Some significant original design flaws still remained in product version of Sys
R, for example, space reservation. We dug up some old unpublished Sys R memos
and reverse engineered the code of Sys R, SQL/DS, DB2/MVS, Sys R, SQL/DS, DB2/MVS,
and IMS. We consulted with the developers of mainframe DB2 to learn why its
recovery differed significantly from Sys R.

What are the key features of the ARIES?

Every page has a Log Sequence Number (PageLSN), which helps keep track of
pages. The Buffer manager tracks dirty pages using RecLSNs, allowing for use of
short-duration latches, rather than longer duration locks, for physical
consistency. ARIES facilitates client server systems with client catching of
data beyond transaction end and server managing data and log disks: simple
technique exploiting recovery info for locking and its optimizations; and
managing messages with high concurrency and linear hashing locking and recovery.

What is the role of a technologist in today's economy?

The economy as well as the IT environment is very dynamic today.
Technologists are required to work in a strategic manner for adequate alignment
between research and products. Therefore, it is important to strike the right
balance between research and the commercial implications of your research. At
IBM, we follow a funding model where a part of the fund comes from the product
organization.