生产中的 mongodb
我们计划在生产中使用 mongodb 来获取一部分数据。过去，我读到 mongodb 存在阻塞写入和写入持久性问题。 2.0版本解决了吗？在生产环境中部署 mongodb 之前还有什么需要注意的吗？
We are planning to use mongodb in production for a subset of data. In past, I have read that mongodb has issues with blocking writes and write durability. Are they resolved with 2.0 release? Are there anything else which one should be careful of, before deploying mongodb in production?
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）
不存在阻塞写入的问题。原子写操作是MongoDB处理并发性和一致性的策略。这确实意味着，如果您的写入负载很高（使用 mongostat 工具进行监控并密切关注“锁定百分比”，这通常应该保持在非常低的水平），您将必须开始使用分片来最大程度地减少每个实例的写入锁争用
。在 2.0 中通过日志功能进行了改进，但在副本集方面已经相当稳定。基本上，如果您投入资源（实例），那么 MongoDB 的持久性和故障转移就非常可靠。日记比任何事情都更能改善（崩溃）恢复。
长话短说，采取适当的措施 MongoDB 是一个生产就绪的存储解决方案。
There are no issues with blocking writes. Atomic write operations is MongoDB's strategy to deal with concurrency and consistency. This does mean that if your write load is high (monitor using mongostat tool and keep an eye on "locked %", this should typically stay very low) you will have to start using sharding to minimize per-instance write lock contention
Durability actually has been improved in 2.0 with the journaling feature but was already pretty solid with replica sets. Basically, if you invest the resources (instances) then durability and fail-over is pretty solid in MongoDB. Journaling improves (crash) recovery more than anything.
TL;DR with the appropriate measures MongoDB is a production ready storage solution.