问题

VM节流

  • 2019年1月3日
  • 2回答
  • 1376的浏览量
E
是否可能在每个VM级别上实现节流?

这个问题背后的想法/场景:
有几次，我们发现一个VM(每次都不一样)丢失了存储控制器的IOPs。当然，我们的目标是解决违规VM的实际问题，但我们希望能够实现节流，以确保当这种情况发生时，集群中的其他VM不会受到影响。

到目前为止，磁盘I/O一直是一种场景，但在某些时候也可能需要限制CPU、内存和/或网卡吞吐量。
    \r\n
    \r\nThe thoughts\/scenario behind the question:
    \r\nWe have had on a couple occasions where we found a VM (different in each occasion) that was running away with storage controller IOPs. Of course the goal is to get the actual issue resolved for the offending VM, but we would like to be able to implement throttling to be able to assure other VMs in that cluster don't suffer when this does occur.
    \r\n
    \r\nDisk I\/O has been the scenario so far, but could also see the need to throttle CPU, memory, and\/or NIC throughput at some point.","quoteUsername":"erobbbins","translations":{"Common":{"like":"Like","unlike":"Unlike"},"Forum":{"Quote":"Quote","Share":"Share"}}}">

    2回答

    ddubuque
    Userlevel 3
    徽章 + 10
    如果您指的是相当于资源池的AHV，以及您如何控制CPU/RAM共享，那么不幸的是，您就不走运了。我也不记得存储端存在任何限制存储I/O的东西。这绝对是一个很好的特性请求。对于异步PD复制，您可以限制带宽，但我从未在其他地方看到过这种情况。

    最好的办法是进入操作系统，限制特定应用程序的每秒字节数。

    你可以购买一些免费的工具和第三方来在应用程序级别限制应用程序。
    这样做的风险由你自己承担。在将任何东西投入生产工作负载之前，我肯定会进行测试，以验证应用程序的稳定性。

    https://www.google.com/amp/s/www.howtogeek.com/347711/how-to-limit-any-applications-bandwidth-on-windows/amp/
    David Dubuque || BKD || Nutanix NTC 2019 & 2020 || Nutanix技术咨询委员会成员|| https://darklinkrising.com
    \r\n
    \r\nYour best bet is to go into the OS and throttle the bytes per second for the specific application.
    \r\n
    \r\nThere are free tools and some third party you can buy to throttle applications at the application level.
    \r\nDo so at your own risk. I would definitely test before throwing anything into a production workload to verify application stability.
    \r\n
    \r\nhttps:\/\/www.google.com\/amp\/s\/www.howtogeek.com\/347711\/how-to-limit-any-applications-bandwidth-on-windows\/amp\/","quoteUsername":"ddubuque","translations":{"Common":{"like":"Like","unlike":"Unlike"},"Forum":{"Quote":"Quote","Share":"Share"}}}">
    E
    @ddubuque，谢谢你的回复。我们将把它移动到一个特性请求，并与我们的客户经理一起工作。

    回复


    Baidu