Variable capacity arrays, also known as “All Flash” storage arrays (like Pure Storage, EMC XTREMIO, and HP Nimble) are popular in the enterprise storage marketplace. Thanks to space reduction technologies like compression and deduplication, customers can buy 20TB of storage while actually storing 80TB of data. Customers love these arrays because they improve the cost-effectiveness of all-flash storage, delivering excellent performance over time (with a simple set-up-and-run).
While these arrays are simple to implement and manage at an operational level, there are some unique features of these arrays that can cause management-related issues if not properly maintained.
Variable capacity arrays are one of those unique features. These are arrays in which data reduction occurs automatically, meaning the overall capacity of an array can, and often does, change over time. Whereas the capacity of a traditional array is fixed, variable capacity arrays are like moving targets, shifting up or down based on changes in data reduction ratios.
In some cases, the reduction ratios can go up, allowing storage arrays to last longer. For example, a variable capacity array with 10 TB of raw storage might actually be able to hold 40 TB of effective storage in this scenario. In other cases, however, these ratios can go down, causing arrays to reach capacity sooner than expected.
As data changes, so does effective capacity.
The challenge with this environment is not managing to the effective capacity, but rather predicting the effective capacity’s changes over time. This may not sound so difficult, but when your effective capacity changes by 20% over a 1-2 week period it can create a variety of unexpected or even “emergency” events.
Monitoring a Moving Target
Most customers monitor storage growth and assume that capacity remains constant. This is not true with variable capacity arrays. For variable capacity arrays you must monitor both “effective capacity” and storage consumption – tracking the relative relationship between these two numbers and setting warnings/alerts for when the trends reach prespecified thresholds.
Easier said than done! These kinds of arrays are only cost-effective if you can dedicate the time and energy needed to analyze and anticipate changes – or, better yet, if something can do it for you.
Pitch-Perfect Forecasting in a Fraction of the Time
Wouldn’t it be nice if you could model and forecast your effective capacity, used storage trends, and data reduction changes – without sidelining other projects in order to find the necessary time? Visual Storage Intelligence enhances the work of IT professionals by doing all of this, even using trends to project across the next six months.
In the example below, you can see both data reduction (light green) and used storage increases (orange) modeled for another VSI client.
As of August 2021, the client’s data reduction ratios have dropped from approximately 3.4-to-1 to 3.0-to-1. That might not seem like much, but it equates to a significant 60-70 TB on this array. At this rate, the client may need additional storage sooner than anticipated.
How soon? Look at the chart again. Using the trends in the data, Visual Storage Intelligence could forecast that the array will reach 80% used storage sometime in Q4. This is critical information – especially if the client was waiting until Q1 or a new budget cycle to make additional storage purchases.
But what if the array capacity does not change and remains constant? It would be nice to have the ability to model this as well. VSI provides this capability and you can see the conclusions are very different from the previous chart.
Now the array is not projected to even reach 70% until months later in July 2022.
Variable capacity arrays are only cost-effective if you can easily analyze and anticipate changes. With Visual Storage Intelligence, you can.
Which scenario should you plan for? It’s best to monitor and plan for both, which is why VSI provides multiple styles of modeling plus a six-month predictive analytics trend analysis.
Moreover, there is one other piece of information that can help you predict which scenario is more likely: the array’s data reduction ratio historical trends.
Below are the data reduction trends for this array:
Now we can see that the client historically has bottomed out around 3:1 – the same level they are at currently. This helps the client predict that Scenario #2 (constant array capacity) is more likely than Scenario #1 (changing array capacity).
Still, both scenarios should ideally be reviewed monthly so that future changes don’t result in avoidable emergency purchases.
With single-pane-of-glass analyses, Visual Storage Intelligence provides automated and human intelligence you can’t get anywhere else. Let us show you how.