By: Lora Johnson
From bottlenecks to latency issues, problems with your storage performance can wreak havoc across your IT stack. Data is the backbone of operations for today’s organizations, as they rely on it to drive their businesses forward, improve the end-user experience and even provide a competitive advantage.
The right visibility into your storage resources helps you transform your data management and improve performance across the IT stack. After all, tracking performance can be challenging given the impact of the overall environment or even applications. However, these three performance parameters offer good indication into how your storage resources are performing.
Often times, your read/write ratio serves as the first clue to potential performance issues. This ratio offers insight into how much of your operations are focused on reads versus writes. Most times, the average is approximately 70 percent reads to 30 percent writes, but can vary based on the workloads of your server. For instance, if your operations are focused more on backups, cloud or clustered applications, your read/write ratio may be more of an even split.
Insight into your read/write ratio can help you understand top-level performance issues with your storage. If you are unable to complete simple read/write requests, it can indicate that there are deeper-level problems that you need to uncover — like lack of capacity or even the inability to process essential transactions.
Insight into how many transactions your system handles each second gives you an indication of how well your servers are able to find and access blocks of data from your storage systems, and how that translates into the overall performance of your Dell EMC VMAX resources.
When reviewing IOPS, however, it’s important to take into consideration the number of transactions in the queue, and the size of each transaction. IOPS simply looks at the number of transactions per second — so if you have low queue, latency can look higher than it does in reality.
On the flipside, a large queue can make the numbers look better than they actually do. The system is still processing a large amount of transactions per second, but it’s taking more time to process each one. Because of this, note that IOPS does offer some imperative insights — particularly if you are interested in evaluating your performance based on transactions.
IOPS gives you enough insight to detect some minor performance problems, but for true visibility, extending that monitoring to include latency will give you the best indication if and when your performance is starting to lag.
Latency builds on IOPS — giving you the key metrics around how many transactions your system can handle — but also the amount of time that your storage system needs to process a data request or a single storage transaction.
Within a flash device like Dell EMC VMAX, following read latency gives you indication of how long it takes for it to access the network connectivity, as well as the amount of time to access the specific data blocks and preparing to transfer that data. Because IOPS tends to only focus on the short and small processes, insight into latency can help you take it one step further and understand how long-term performance may be impacted by current issues.
The Blue Medora Dell EMC VMAX Management Pack extends monitoring so you can monitor these three key performance parameters, among others, to optimize your Dell EMC VMAX resources. The management pack collects more than 500 metrics, including:
In addition, it offers nine comprehensive monitoring dashboards — like overview, front end director detail and IOPS — so you can dive deeper into the metrics that matter to simplify troubleshooting and elevate your performance.
Figure 1: The Health Investigation Dashboard from Blue Medora
To learn more about the Dell EMC VMAX Management Pack from Blue Medora or to download a free trial, please visit the True Visibility Suite for VMware vRealize Operations page on Blue Medora’s website.
This blog post first appeared on the VMware Cloud Management blog. Read the full post here.