home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!walter!qualcom.qualcomm.com!network.ucsd.edu!ucsbcsl!ucsb.edu
- From: mliu@ucsb.edu (Mei-Ling Liu)
- Newsgroups: comp.databases.ingres
- Subject: Detecting Failure by Timeout
- Message-ID: <6719@ucsbcsl.ucsb.edu>
- Date: 16 Nov 92 16:55:13 GMT
- Sender: root@ucsbcsl.ucsb.edu
- Organization: University of California, Santa Barbara
- Lines: 18
-
- In a distributed DBMS, if you are worried about site failures and communication
- failures, you are supposed to detect these failures by using timeouts: Site
- A sends a message to Site B and waits for a reply within a predetermined
- period of time t called the timeout period. This timeout period is a
- function of many, many things, including the system load at each site,
- the message propagation time, and clock skews, all of which are pretty
- hard to quantify. As far as I know, no analytical study has been made
- on how one would arrive at a value for the timeout period.
-
- My question is, how is such a quantity determined in practice?
- If you have implemented such a system, I would appreciate hearing
- from you. Thanks in advance.
-
- Mei-Ling Liu
- Distributed Systems Lab
- Computer Science Department, UCSB
- mliu@cs.ucsb.edu
-
-