The design and implementation of a log-structured file system
ACM Transactions on Computer Systems (TOCS)
TNet: A Reliable System Area Network
IEEE Micro
File system logging versus clustering: a performance comparison
TCON'95 Proceedings of the USENIX 1995 Technical Conference Proceedings
Measuring Real-World Data Availability
LISA '01 Proceedings of the 15th USENIX conference on System administration
Geographically Distributed System for Catastrophic Recovery
LISA '02 Proceedings of the 16th USENIX conference on System administration
Hi-index | 0.00 |
This article describes aspects of the implementation of a project called "Clustered Failover." A Clustered Failover configuration connects two independent file server appliances via a NUMA network to provide file service that can survive hardware faults with minimal disruption to clients. The paper focuses on how the combination of the properties of WAFL (Network Appliance's proprietary log structured file system) and NUMA interconnects minimizes overhead and makes for a relatively straightforward implementation.