In a virtual environment, if a VM has performance issues, it can affect the performance of other components within the virtual infrastructure. The converse is applicable too. For example, you can collectively map all your VMs and, depending on the location of your bottleneck, you could find that the issue is related to storage performance. When you want to view storage performance, you typically look at storage IOPS to measure performance of a storage system. Measuring IOPS throughput will give you the amount of data transferred per second, and measure the number of operations per second that a storage system can process.
Whether you’re using VMware or Hyper-V storage, you should look at storage performance across the hosts and clusters, and periodically monitor for issues. One of the ways you can assess VM performance is by looking at critical performance metrics. To do this, you need to drill down further into a VM and check which VMs are using the most storage resources. For example, if you have three VMs with 30 GB allocated, and two of the VMs are hogging resources, then the third will have performance issues.
As virtual and storage admins, you should determine the number of IOPS your applications use on a daily basis. The best way to go about doing this is to monitor your current application performance to see how your servers are performing and to ensure they are healthy. The next thing you want to look at is usage. Knowing how your applications perform at any given time is key. This will tell you the average IOPS value an application uses. The IOPS value will vary for different applications. For example, Exchange Server may be widely used as opposed to someone using a database server or a multi-media application.
When you’re looking at IOPS and storage performance, you should keep the following in mind:
- Map all the VMs against how much IOPS each one is consuming and determine if the IOPS values are higher than normal
- Look at IOPS against latency, I/O size, read/write values, etc. to make meaningful insights on why storage performance issues occur
- Proactively monitor applications or VMs to determine if they are consuming too much of storage resources
- Have proper baselines in place to ensure high IOPS values aren’t affecting your storage performance
- A significant amount of I/O means there could be potential storage issues
Watch this short video where virtualization vExpert Scott Lowe shows you the steps involved to identify the drivers of storage I/O for a datastore. Learn how to drill down from a datastore to see which VMs are tied to that datastore, which VMs are driving storage I/O, spikes in I/O, and more using a virtualization management software.