On-The-Fly Reading of Entire Databases

A common database need is to obtain a global-read, which is a consistent read of an entire database. To avoid terminating normal system activity, and thus improve availability, we propose an on-the-fly algorithm that reads database entities incrementally and allows normal transactions to proceed concurrently. The algorithm assigns each entity a color based on whether the entity has been globally read, and a shade based on how normal transactions have accessed the entity. Serializability of execution histories is ensured by requiring normal transactions to pass both a color test and a shade test before being allowed to commit. Our algorithm improves on a color-only-based scheme from the literature; the color-only scheme does not guarantee serializability. >

[1]  Irving L. Traiger,et al.  The notions of consistency and predicate locks in a database system , 1976, CACM.

[2]  Hector Garcia-Molina,et al.  Altruistic locking , 1994, TODS.

[3]  Divyakant Agrawal,et al.  Locks with constrained sharing , 1990, PODS 1990.

[4]  Rudolf Bayer,et al.  Parallelism and recovery in database systems , 1980, TODS.

[5]  Rudolf Bayer,et al.  Consistency of transactions and random batch , 1986, TODS.

[6]  K. V. S. Ramarao,et al.  On the management of long-living transactions , 1990, J. Syst. Softw..

[7]  Calton Pu,et al.  Performance Evaluation of Global Reading of Entire Databases , 1988, Proceedings [1988] International Symposium on Databases in Parallel and Distributed Systems.

[8]  Divyakant Agrawal,et al.  Performance characteristics of protocols with ordered shared locks , 1991, [1991] Proceedings. Seventh International Conference on Data Engineering.

[9]  Sushil Jajodia,et al.  Distributed timestamp generation in planar lattice networks , 1993, TOCS.

[10]  D. Elliott Bell,et al.  Secure Computer System: Unified Exposition and Multics Interpretation , 1976 .

[11]  Meichun Hsu,et al.  Partitioned two-phase locking , 1986, TODS.

[12]  Sushil Jajodia,et al.  A two snapshot algorithm for concurrency control in multi-level secure databases , 1992, Proceedings 1992 IEEE Computer Society Symposium on Research in Security and Privacy.

[13]  Hamid Pirahesh,et al.  Efficient and flexible methods for transient versioning of records to avoid locking by read-only transactions , 1992, SIGMOD '92.

[14]  Divyakant Agrawal,et al.  Locks with constrained sharing (extended abstract) , 1990, PODS '90.