Detecting and Tolerating Byzantine Faults in Database Systems
This thesis describes the design, implementation, and evaluation of a replication scheme to handle Byzantine faults in transaction processing database systems. The scheme compares answers from queries and updates on multiple replicas which are off-the-shelf database systems, to provide a single data...
| Main Author: | |
|---|---|
| Other Authors: | |
| Published: |
2008
|
| Online Access: | http://hdl.handle.net/1721.1/41873 |
| _version_ | 1811085420422758400 |
|---|---|
| author | Vandiver, Benjamin Mead |
| author2 | Barbara Liskov |
| author_facet | Barbara Liskov Vandiver, Benjamin Mead |
| author_sort | Vandiver, Benjamin Mead |
| collection | MIT |
| description | This thesis describes the design, implementation, and evaluation of a replication scheme to handle Byzantine faults in transaction processing database systems. The scheme compares answers from queries and updates on multiple replicas which are off-the-shelf database systems, to provide a single database that is Byzantine fault tolerant. The scheme works when the replicas are homogeneous, but it also allows heterogeneous replication in which replicas come from different vendors. Heterogeneous replicas reduce the impact of bugs and security compromises because they are implemented independently and are thus less likely to suffer correlated failures. A final component of the scheme is a repair mechanism that can correct the state of a faulty replica, ensuring the longevity of the scheme.The main challenge in designing a replication scheme for transaction processingsystems is ensuring that the replicas state does not diverge while allowing a high degree of concurrency. We have developed two novel concurrency control protocols, commit barrier scheduling (CBS) and snapshot epoch scheduling (SES) that provide strong consistency and good performance. The two protocols provide different types of consistency: CBS provides single-copy serializability and SES provides single-copy snapshot isolation. We have implemented both protocols in the context of a replicated SQL database. Our implementation has been tested with production versions of several commercial and open source databases as replicas. Our experiments show a configuration that can tolerate one faulty replica has only a modest performance overhead (about 10-20% for the TPC-C benchmark). Our implementation successfully masks several Byzantine faults observed in practice and we have used it to find a new bug in MySQL. |
| first_indexed | 2024-09-23T13:09:19Z |
| id | mit-1721.1/41873 |
| institution | Massachusetts Institute of Technology |
| last_indexed | 2024-09-23T13:09:19Z |
| publishDate | 2008 |
| record_format | dspace |
| spelling | mit-1721.1/418732019-04-11T03:10:12Z Detecting and Tolerating Byzantine Faults in Database Systems Vandiver, Benjamin Mead Barbara Liskov Programming Methodology This thesis describes the design, implementation, and evaluation of a replication scheme to handle Byzantine faults in transaction processing database systems. The scheme compares answers from queries and updates on multiple replicas which are off-the-shelf database systems, to provide a single database that is Byzantine fault tolerant. The scheme works when the replicas are homogeneous, but it also allows heterogeneous replication in which replicas come from different vendors. Heterogeneous replicas reduce the impact of bugs and security compromises because they are implemented independently and are thus less likely to suffer correlated failures. A final component of the scheme is a repair mechanism that can correct the state of a faulty replica, ensuring the longevity of the scheme.The main challenge in designing a replication scheme for transaction processingsystems is ensuring that the replicas state does not diverge while allowing a high degree of concurrency. We have developed two novel concurrency control protocols, commit barrier scheduling (CBS) and snapshot epoch scheduling (SES) that provide strong consistency and good performance. The two protocols provide different types of consistency: CBS provides single-copy serializability and SES provides single-copy snapshot isolation. We have implemented both protocols in the context of a replicated SQL database. Our implementation has been tested with production versions of several commercial and open source databases as replicas. Our experiments show a configuration that can tolerate one faulty replica has only a modest performance overhead (about 10-20% for the TPC-C benchmark). Our implementation successfully masks several Byzantine faults observed in practice and we have used it to find a new bug in MySQL. 2008-07-02T06:00:10Z 2008-07-02T06:00:10Z 2008-06-30 MIT-CSAIL-TR-2008-040 http://hdl.handle.net/1721.1/41873 Massachusetts Institute of Technology Computer Science and Artificial Intelligence Laboratory 174 p. application/pdf application/postscript |
| spellingShingle | Vandiver, Benjamin Mead Detecting and Tolerating Byzantine Faults in Database Systems |
| title | Detecting and Tolerating Byzantine Faults in Database Systems |
| title_full | Detecting and Tolerating Byzantine Faults in Database Systems |
| title_fullStr | Detecting and Tolerating Byzantine Faults in Database Systems |
| title_full_unstemmed | Detecting and Tolerating Byzantine Faults in Database Systems |
| title_short | Detecting and Tolerating Byzantine Faults in Database Systems |
| title_sort | detecting and tolerating byzantine faults in database systems |
| url | http://hdl.handle.net/1721.1/41873 |
| work_keys_str_mv | AT vandiverbenjaminmead detectingandtoleratingbyzantinefaultsindatabasesystems |