添加架构或创建新实例？
不像往常那样是一个复杂的问题。
我刚刚被要求就在我们当前运行的生产系统的同一个 Oracle (10g) 实例中填充更多模式可能产生的影响提供一些意见。
这不完全是一个问题。服务器没有超载或任何其他情况，它将获得支持新模式可能需要的适当的额外资源。
我的观察是“好吧，您将共享分配给该实例的资源”，“当一个备份启动时，两个系统都会关闭”。
想知道我错过了什么。绝对建议何时获取单独的实例。对此的正常约定是什么？或者你们可能有的其他建议。
想让你们知道，我们谈论的是内存和处理器密集型应用程序，长时间内几乎没有 I/O。简而言之，我们一直在比较庞大的数据表。
非常感谢您的想法。
f.
Not to much a complex question as usual.
I've been just asked to give some opinion on possible impacts of stuffing more schemas in the same Oracle (10g) instance we have our production system running currently.
That's not exactly a problem. The server isn't overloaded or anything and it will get the proper additional resources it might need for supporting the new schemas.
My observations are of the sort "well, you are going to share resources assigned to that instance", "when one backup kicks in, both systems will be down".
Wonder what I'm missing. When would be absolutely recommended to get separated instances. What would be a normal convention for that? or other tips you guys might have.
Just so you guys know, we are talking about a memory and processor intensive application, with pretty much no I/O for long periods. In a nutshell, we are comparing huge data tables all the time.
Highly appreciate your thoughts.
f.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
如果一台服务器上有多个实例，那么您将很难解决性能问题，因为 Oracle 工具将无法看到彼此的性能日志。
这就是为什么在繁忙的生产系统上我的建议是“每台服务器最多一个实例”。我在生产中的虚拟化方面没有太多经验，但如果您有资源来处理虚拟化环境，则可以为每个逻辑服务器拥有一个实例。
另一个不相关的问题是，为什么要使数据库脱机来执行备份，特别是如果它不是 IO 密集型系统？
If you have more than one instance on a single server, you will have some hard time troubleshooting performance issues because the Oracle tools won't be able to see each other's performance logs.
This is why on busy production systems my advice would be "one instance at most per server". I don't have much experience with virtualization in production but if you have the resource to deal with a virtualized environment, you could have one instance per logical server.
On an unrelated note, why would you take the DB offline to perform a backup, especially if it is not an IO intensive system?