The sunburst visualisation is perfect for very quickly summarising capacity and headroom across your estate. In a single visualisation, you can see all your data centres, clusters, hosts, virtual machines, physical servers and more. It’s a fantastically impactful image and behind that visualisation is a vast amount of un-aggregated data, hundreds of millions of data points that can be presented in many ways.
A benefit of being SaaS is that we can easily extend the product for our customers when they have very specific data presentation requirements. Based on these type of requests, we’ve started building a library of reports that will be available to all subscribed users of Sumerian Capacity Planner.
When you log into Capacity Planner now, you’ll find a whole new area of the product available to you, with a new accordion entry called ‘Reports’. Each report has several configuration parameters you can use to narrow down queries. They also export nicely into Excel format if you want to take the output and use it elsewhere.
The release note for this update goes into more detail on the reports and configurations but here’s a bit on some of the reports that you will have available to you.
Highly Inactive VMs
This report will identify those virtual machines across the chosen baseline that are considered ‘highly inactive’ with very low levels of CPU utilisation. All statistical measures are given for each VM and ordered by ‘peak’, smallest to largest and a summary of the disk space used by each VM is also provided. This will allow you to very quickly identify those VMs that can be switched off or those that spend significant amount of time inactive, but with very small times of peak activity.
Host and VM Thresholding Reporting
CPU and Memory demand thresholding is frequently used in regular capacity reporting. These reports usually present a summarisation of these metrics using a chosen statistical measure and a given timeframe. Often, the average value is used for this type of reporting, but using the average really doesn’t provide the visibility you need.
This report also allows you to select the statistical measure to use in combination with a threshold. The measure is an indication of how much total time in the chosen timeframe that metric has been above the threshold.
In the below example, the user has selected a timeframe of ‘Last Month’, the 95th %ile and threshold value of 50% for both CPU and Memory utilisation. Any servers with a red bar are those that have spent a total of 36 hours above 50% utilisation.
A click through on the either the CPU or memory bar will take you to a daily box plot chart.
Clicking on the server name itself will take you to a scatter plot of CPU against Memory for the time-period selected.
Host and VM Time Above Threshold
These are particularly powerful reports that allow the user to really drill down into the behaviour of chosen metrics. They allow you to filter to a group of hosts or VMs then determine how long a metric has been above a threshold, then identify the single longest time-period for which it was above that threshold. This can be further filtered on days of the week and time of day.
For example, using this report a user can analyse CPU utilisation across all VMs or hosts for a given service or application for weekdays only during the working day over the last month.
The report will then present the servers that have crossed this threshold in the last month, the total amount of time that server has been above the threshold and the single longest time-period.
In the below example, server 1 has spent 40 minutes above 70% utilisation and this occurred between 12:15 and 12:50 on the 6th March.
Click through on the ‘length of longest episode’ for a detailed time series of when that occurred.
That’s just a few of the reports that are available to you. We will be continuing to release more reports and, if you’re a SaaS subscriber, you can get in touch with us at any time to discuss your own detailed reporting requirements.