怎么使用Nvidia Grid M10将VMWare 6.0升级到6.5

你好
任何已知问题都将出现在特定平台的GRID驱动程序包的发行说明中。
但据我所知，那里没有阻挡剂。
我已经在6.0和6.5平台上部署了GRID，对于我通常使用的VM类型（大内存），由于6.0的性能问题，建议使用6.5。
问候
本



以下为原文

Hi

Any known issues will be in the release notes of the GRID driver bundle for your specific platform. But to my knowledge, there are no show stoppers out there. I've deployed GRID on 6.0 and 6.5 platforms, and for the type of VMs I typically work with (large memory), 6.5 is a recommended upgrade due to performance issues with 6.0.

Regards

Ben

谢谢本。
您遇到的6.0性能问题是什么？



以下为原文

Thanks Ben. What were the performance issues with 6.0 that you have experienced?

我的客户遇到的主要问题是启动时间。
但仅此一点就可以真正影响平台的管理和可用性......
在分配大内存的情况下，VM需要很长时间才能从冷启动或关闭开启。
重新启动很好，因为内存已经分配给VM。
我们使用的虚拟机每个都有128GB RAM，只需要15分钟就可以启动vGPU。
迁移到ESXi 6.5，启动时间减少到90秒。
当其中一个VM正在启动或关闭时，该主机上的任何其他设备都无法启动或关闭，直到该VM完成其任务。
因此，您必须等待（在我们的情况下为15分钟）才能在该主机上执行任何其他操作。
因此，在主机上有4个VM，启动时间是最后一个用户能够登录到VM之前的1小时。
它在生产平台上几乎无法使用。
VMware认识到了这个问题，并通过6.5解决了这个问题。



以下为原文

The main issue my customer was having was startup times. But this alone can really impact administration and usability of the platform ...

With large memory assigned, the VMs take ages to power on from cold boot or shutdown. Restarts are fine as the memory has already been allocated to the VM.

The VMs we were using each had 128GB RAM and took 15 minutes just to power on with a vGPU assigned. Moving to ESXi 6.5, the startup time was reduced to 90 seconds.

When 1 of those VMs is powering on or shutting down, nothing else on that host can start or shutdown until that VM has finished its task. So you have to wait (in our case 15 minutes) before you can do anything else on that host. So with 4 VMs on the host, the start up time was 1hr before the last user was able to log on to the VM. It's pretty much unusable in a production platform.

VMware recognized the issue and it was resolved with 6.5.