120 likes | 262 Views
TCP Server Fault Tolerance Using Connection Migration to a Backup Server. 2003 IEEE Manish Marwah, Shivakant Mishra, Christof Fetzer University of Colorado International Conference on DSN. Presented by JIUN-JAU-CHIOU. Outline. Introduction Overview of ST-TCP UDP channel & receive buffer
E N D
TCP Server Fault Tolerance Using Connection Migration to a Backup Server 2003 IEEE Manish Marwah, Shivakant Mishra, Christof Fetzer University of Colorado International Conference on DSN Presented by JIUN-JAU-CHIOU
Outline • Introduction • Overview of ST-TCP • UDP channel & receive buffer • Initialization • Failure-free period • Failure detection • Performance • Conclusion
Introduction • ST-TCP (Server fault-Tolerant TCP) • A Primary/Backup approach • ST-TCP is transparent to clients • Changes only required on server side
Overview of ST-TCP Client Primary UDP communication channel Backup
UDP channel • Heartbeat message • ACK message • ACKs from backup to primary • NextByteExpected – LastByteAcked >= X • A fixed time interval • Packet retransmission • Backup sends request to primary to get lost packets.
Primary’s receive buffer • Size of buffer is doubled • It was logically divided into 2 part NextByteNeed LastByteAcked LastByteRead
Initialization. • UDP channel • It created when servers started • Sequence number • Backup server makes its seq. number match primary’s
Failure-free period • Consistence • ACK strategy • Modified receive buffer • Using multicast • Heartbeat message Client Primary Gateway Backup
Failure detection • Both servers monitor HB messages • Timeout mechanism • Timeout = 3 consecutive HB • Guess is always right • Switch off power of suspect
Conclusion • Low performance overhead during failure-free period • ST-TCP is transparent to clients • Fast failover