解决了

文件服务器的虚拟机的最大卷大小

  • 2021年3月5日
  • 1回复
  • 317的浏览量

你好,我们在我们的Nutanix的一个节点上安装了带有文件服务器的虚拟机。目前磁盘的大小是3.5 TB,但是我们希望将这个大小扩展到4.5TB。有人知道这么大的磁盘,VM是否能正常工作吗?

图标

最佳答案伊万诺夫2021年4月4日11:04

Hi SergeySephora,<\/p>

If the hypervisor is AHV, the virtual disk size is almost limitless. The theoretical maximum would be 9 exabytes. So, if the guest operating system supports the desired size, it is good to go.<\/p>

However, i would recommend not to create large disks for performance reasons. In Nutanix we have write cache, which is called oplog and the oplog is always 6Gb in size. It is assigned per vdisk and it means that every vdisk has 6Gb of oplog no matter what size the vdisk is. If you get a very big vdisk and you do write operations towards it very actively, it might be performing slower, because the oplog would get full very quickly. The trick is to create multiple vdisks and present them to the guest OS. If the application cannot work with multiple disks and requires only 1 disk, you can use striped volumes to create a single guest disk made of multiple Nutanix vdisks. For example, if you need 4.5Tb, you can create 5x900Gb disks to get the same size, but that way you will get 5 times write cache compare to asingle large vdisk and the performance will be better. However, if performance is not very important, you can go with 1 large vdisk.<\/p>","className":"post__content__best_answer"}">

查看原始

这个话题已经停止评论

1回复

Userlevel 4
徽章 +5

嗨SergeySephora,

如果管理程序是AHV,那么虚拟磁盘的大小几乎是无限的。理论上的最大值是9艾字节。因此,如果客户操作系统支持所需的大小,就可以使用它。

但是,出于性能考虑,我建议不要创建大磁盘。在Nutanix中,我们有写缓存,叫做oplog, oplog的大小总是6Gb。它被分配给每个vdisk,这意味着每个vdisk都有6Gb的oplog,无论vdisk的大小如何。如果您有一个非常大的vdisk,并且非常积极地对它进行写操作,那么它可能会执行得更慢,因为oplog很快就会被填满。诀窍是创建多个vdisk并将它们呈现给来宾操作系统。如果应用程序不能使用多个磁盘,只需要一个磁盘,可以使用条带卷创建由多个Nutanix vdisk组成的单个客户磁盘。例如,如果您需要4.5Tb,您可以创建5x900Gb的磁盘来获得相同的大小,但这样您将获得5倍于单个大vdisk的写缓存,性能也会更好。但是,如果性能不是很重要,您可以使用1个大的vdisk。

Baidu