JBOSS 5.1 EAP可用的任何生产Ready JMS故障转移（HA）吗？
我们在 JBoss 5.1 EAP 中使用 HornetQ 作为 JMS 提供者，总的来说，我们对 HornetQ 的性能非常满意。
HornetQ 在 2.2.2 中默默地放弃了数据复制支持。现在，我们在使用 JBoss 和 HornetQ 创建 HA 解决方案时遇到了问题，因为当前的 HornetQ 故障转移策略并不可靠。例如，在主服务器重新上线后，客户端重新连接到 HornetQ 时会遇到问题。使用循环策略时，消息存储在存储转发队列中，这会导致节点关闭时消息未传送。
HornetQ 正在不断发展，但如果没有数据复制，就无法为生产环境提供严肃的解决方案。
哪个 JMS 提供商支持真正的生产安全 HA 和故障转移功能以及良好的性能、监控，当然还有 JBoss 5 互操作性？
We are using HornetQ as JMS provider in JBoss 5.1 EAP and in general we are quite satisfied with HornetQs performance.
HornetQ dropped data replication support in 2.2.2 silently. Now, we have problems creating a HA solution with JBoss and HornetQ as the present HornetQ failover strategies aren't reliable. For example, clients have problems to reconnect to the HornetQ after the master is back online. Messages are stored in a store-and-forward queue when using a round-robin strategy and this leads to undelivered messages in a case of a node shutdown.
HornetQ is evolving, but without data replication no serious solution for a productive environment.
Which JMS provider support real production safe HA and failover capabilities in conjunction with good performance, monitoring and of course JBoss 5 interoperability?
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）
我们一直在支持使用 GFS 的用户，这比复制更可靠，因为它保证单例实例在生命和备份之间随时运行。
我们正在努力复制。我们的代码已经可以运行，应该很快就会发布。但共享存储已被证明是非常可靠的。
我们将支持复制，但恕我直言，共享存储是大多数情况下的最佳解决方案。唯一不符合这种情况的情况是，无法投资支持 GFS 的磁盘（或任何其他共享且高可用的磁盘）。
我们有大型银行在生产环境中使用此解决方案。正如你所说，我们正在取得进展。即使已经支持复制，这些银行实际上也会使用硬件解决方案（包括灾难恢复数据中心）。
We have been supporting our users using GFS, which would be way more reliable than replication since it guarantees a singleton instance running between life and backup at any point.
We are working towards replication. We have the code working and it should be released shortly. But shared storage has been proved to be very reliable.
We are going to support replication, but IMHO shared storage is the best solution for most of the cases. The only scenario where this is not true is when there's no way to invest on the disk that will support GFS (or any other shared and high available disks).
We have big banks using this solution in production environments. As you said we are making progress. These banks actually would use the hardware solution (including disaster-recovery datacenter) even if replication was already supported.