有成千上万的上游用户和频繁的更改,Apache Apisix当前实现如何解决它有什么问题?
如果我现在有成千上万的上游,并且正在发生很多变化,那么我如何使用Apisix更好地解决此问题,是否有任何最佳实践?
If I have tens of thousands of upstream now and it's changing a lot, how can I use Apisix to better solve this, are there any best practices?
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)
由于上游有太多,因此需要在ETCD群集中更新这些更新会引起高CPU使用情况。此外,在ETCD群集内部,在数据库写入的情况下,应该有一个群体状态以保持一致性,这也将增加ETCD CPU使用情况。
为了改善这种情况,我们可以在创建路由时使用上游ID引用上游,以使负载较小,因为无需更新路由配置。
Since there are so many upstreams, updates to those will cause high CPU usage as they need to be updated in the etcd cluster. Moreover, inside the etcd cluster, there should be a quorum status for consistency in case of database writes which will also increase etcd CPU usage.
To improve this situation we can refer upstreams using upstream id when creating routes so that the load is lesser as no need to update route configuration.