i**h 发帖数: 17 | 1 even though failover is not J2EE standard, every vendor pretty much
implemented it in the same way. after all, there are well-known computer
science research on how this can and should be done. and i don't think porting
to another container is a big deal because most likely you are going to have
container specific configure file for clustering (e.g. weblogic.xml,
jboss.xml, etc) and if you need to port to another container, you know where
you should change. (unless you use things like xdoclet, | m******t 发帖数: 2416 | 2
That's probably why almost all the container vendors have a "not recommended
for production environment" note attached to hot deployment.
【在 i**h 的大作中提到】 : even though failover is not J2EE standard, every vendor pretty much : implemented it in the same way. after all, there are well-known computer : science research on how this can and should be done. and i don't think porting : to another container is a big deal because most likely you are going to have : container specific configure file for clustering (e.g. weblogic.xml, : jboss.xml, etc) and if you need to port to another container, you know where : you should change. (unless you use things like xdoclet,
| r*t 发帖数: 34 | 3 Thanks a lot for your answer, it is really helpful for me. But I still have
some puzzles to be clarified, thanks for your advice very much!
1. You talked about failover related to 'clustering', however, my project
focused more on 'failover' itself. All we need is just failover, not
clustering. In this case, can we still have the similar expection as for
clustering? Correct me if I am wrong, for 'failover' related to clustering, we
rely more on hardware. However, here we means to implement 'failo
【在 i**h 的大作中提到】 : even though failover is not J2EE standard, every vendor pretty much : implemented it in the same way. after all, there are well-known computer : science research on how this can and should be done. and i don't think porting : to another container is a big deal because most likely you are going to have : container specific configure file for clustering (e.g. weblogic.xml, : jboss.xml, etc) and if you need to port to another container, you know where : you should change. (unless you use things like xdoclet,
| i**h 发帖数: 17 | 4 the only way to support failover is via redundancy, of software processes
(e.g. you can start two weblogic servers on one machine) and hardware (e.g.
you have 2 machines with weblogic running on them, for true failover).
clustering refered to in this context is at software level, not at hardware
level. for instance, there are very complex communication schemes on how to
synchronize changes among multiple servers (see jboss jgroups as an example).
as for classloader issues, i cannot give you much
【在 r*t 的大作中提到】 : Thanks a lot for your answer, it is really helpful for me. But I still have : some puzzles to be clarified, thanks for your advice very much! : 1. You talked about failover related to 'clustering', however, my project : focused more on 'failover' itself. All we need is just failover, not : clustering. In this case, can we still have the similar expection as for : clustering? Correct me if I am wrong, for 'failover' related to clustering, we : rely more on hardware. However, here we means to implement 'failo
|
|