ARIES: A Transaction Recovery Method Supporting
Fine-Granularity Locking and Partial Rollbacks
Using Write-Ahead Logging

C. Mohan
Data Base Technology Institute
IBM Almaden Research Center
San Jose, CA 95120

Tuesday, April 21, 1998, Volen 101, 9am-10am. (Refreshments at 9am)

In this talk, I present a simple and efficient method, called ARIES (Algorithm for Recovery and Isolation Exploiting Semantics), which supports partial rollbacks of transactions, fine-granularity (e.g., record) locking and recovery using write-ahead logging (WAL). I introduce the paradigm of repeating history to redo all missing updates before performing the rollbacks of the loser transactions during restart after a system failure. ARIES uses a log sequence number in each page to correlate the state of a page with respect to logged updates of that page. All updates of a transaction are logged, including those performed during rollbacks. By appropriate chaining of the log records written during rollbacks to those written during forward progress, a bounded amount of logging is ensured during rollbacks even in the face of repeated failures during restart or of nested rollbacks.

I deal with a variety of features that are very important in building and operating an "industrial-strength" transaction processing system. ARIES supports fuzzy checkpoints, selective and deferred restart, fuzzy image copies, media recovery, and high concurrency lock modes (e.g., increment/decrement) which exploit the semantics of the operations and which require the ability to perform operation logging. ARIES is flexible with respect to the kinds of buffer management policies that can be implemented. It supports varying length objects efficiently. By permitting parallelism during restart, page-oriented redo and logical undo, it enhances concurrency and performance.

ARIES is applicable not only to data base management systems but also to persistent object-oriented languages, recoverable file systems and transaction-based operating systems. I show why some of the System R paradigms for logging and recovery, which were based on the shadow page technique, need to be changed in the context of WAL. I compare ARIES to the WAL-based recovery methods of DB2/MVS V1, IMS and Tandem systems.

ARIES has been implemented, to varying degrees, in IBM's DB2 Family of Products, MQSeries, ADSM, Lotus Domino/Notes, Starburst and QuickSilver, in the University of Wisconsin's Gamma, EXODUS and Shore, and in Transarc's Encina product suite. More details on ARIES can be found in a paper in the March 1992 issue of the ACM Transactions on Database Systems and also in the IBM Research Report RJ6649.

This is joint work with D. Haderle, B. Lindsay, H. Pirahesh and P. Schwarz.

Host: Liuba Shrira


Dr. C. Mohan joined the IBM Almaden Research Center as a Research Staff Member in 1981. In June 1997, he was named an IBM Fellow for being recognized worldwide as a leading innovator in database transaction management. He received the ACM SIGMOD Innovations Award in 1996 in recognition of his innovative contributions to the development and use of database systems. In 1992, he was elected to the IBM Academy of Technology. Mohan is currently leading the Dominotes project whose goal is to enhance Lotus Domino/Notes's data integrity and scalability by introducing log-based transactional recovery in R5. Prior to this, Mohan led the Exotica project which was focussed on the IBM workflow product FlowMark, the IBM messaging product MQSeries and the Lotus groupware product Notes. Mohan was a designer and/or implementor of the R* distributed DBMS, the Starburst extensible DBMS and DB2. He is the primary inventor of the ARIES family of recovery and locking methods, and the Presumed Abort commit protocol. He has had major impact on IBM and non-IBM prototypes and products. Mohan's research results and designs have been incorporated in the IBM products DB2/MVS, DB2 UDB (DB2/6000, DB2/NT, ....), MQSeries, S/390 Parallel Sysplex Coupling Facility, AdStar Distributed Storage Manager (ADSM), SQL/DS and VM Shared File System, in the IBM prototypes R*, Starburst and QuickSilver, and in IBM's SNA LU6.2 and DRDA. Mohan is a consultant to IBM's database, transaction and workflow product groups. Mohan is the recipient of several IBM awards: an IBM Corporate Award for his contributions to database support for the S/390 parallel sysplex; an IBM Outstanding Innovation Award (OIA) for his coinvention of the ARIES recovery method which is being used in several IBM products, in Transarc's Encina, and in the University of Wisconsin's Gamma and EXODUS DBMSs, and SHORE persistent object system; an OIA for his inventions (ARIES, ARIES/IM, Commit_LSN) and major contributions to performance, availability and concurrency in DB2/MVS V4; three OIAs for his algorithmic and hardware architectural coinventions for supporting the shared disks transaction environment in S/390 and DB2/MVS; an OIA for his coinvention of the Hybrid Join method which is implemented in DB2/MVS; an OIA for his coinvention of the Presumed Abort commit protocol which has been widely adopted in the industry and which is now part of the ISO-OSI, X/Open and DRDA distributed transaction processing standards; an IBM Research Division Award (RDA) for his work on transaction management in R*; an RDA for his contributions to WDSF/VM; 9th Plateau IBM Invention Achievement Award for his patent activities (28 issued and 3 pending). Mohan was named a leading software inventor of IBM for 1994 and 1995, and a Master Inventor in 1997. Mohan was the Americas Program Chair for the 1996 International Conference on Very Large Data Bases, a Program Vice-Chair for the 1994 IEEE International Conference on Data Engineering and the Program Chair for the 1987 International Workshop on High Performance Transaction Systems. He is an Editor of the VLDB Journal, and Distributed and Parallel Databases - An International Journal. Mohan received a PhD in computer science from the University of Texas at Austin in 1981 and a BTech in chemical engineering from the Indian Institute of Technology, Madras in 1977.