| « | October 2025 | » | | 日 | 一 | 二 | 三 | 四 | 五 | 六 | | | | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | | |
|
公告 |
| 本博客在此声明所有文章均为转摘，只做资料收集使用。并无其他商业用途。 |
| Blog信息 |
|
blog名称:
日志总数:210
评论数量:205
留言数量:-19
访问次数:923395
建立时间:2007年5月10日 |
| |
|
 [内容管理框架]Jackrabbit 笔记：Session and server state
文章收藏, 网上资源, 软件技术, 电脑与网络
李小白 发表于 2008/4/8 22:07:53 |
| I think session creation is not expensive, so #2 is perhaps simplest. I
guess the session can act as a cache, but perhaps a cache targeted more
specifically at an application would better serve performance then the
Session itself.
Thoughts?
On 2/8/07, Pruitt, Steve <[EMAIL PROTECTED]> wrote:
I am interested in managing server state in regards to repository
sessions. Specifically, the scenario of a load balancer distributing
user requests across multiple servers. There are a number of questions
that come to mind and I was wondering how others have approached the
problem.
In no particular order...
1) To avoid recreating a new session per user request, the repository
session can be cached until the user "logs off".
2) Or, if session creation is not expensive then maybe the session is
committed and then destroyed per user request?
3) If the session is persisted across user requests, how can the session
be serialized to sync up the other servers.
4) Or, the session cannot be serialized and server affinity must be
enforced.
There may be other and obvious issues, but these are forefront for me.
Any comments and ideas are appreciated.
-S
Re: Session and server state Michael Neale |
|
|