Putting the Intelligent in Intelligent Infrastructure — Part II — Metrics and Analytics
by Rob Girard, Principal Technical Marketing Engineer
In my previous blog post, Putting the Intelligent in Intelligent Infrastructure — Part I — the File System, I explored the concept of intelligence and discussed how it manifests in our VMstore product family through our unique file system with its inherent knowledge of virtual machines (VMs) and databases and how machine learning (ML) drives its experiential capabilities to predict workload behavior. In this blog post, I continue with the other unique capabilities that VMstore affords and how we leverage granular metrics and predictive analytics to deliver a fundamentally different user experience, one that we believe you would enjoy.
What else can the brain (that I discussed in my previous blog post) do? I don’t know. What can your brain do? Just as you can learn new things that may not have been considered in the original design of the human brain (such as rockin’ at Beat Saber in VR), we can also leverage this powerful building block to do more.
What is the result of all this industry-leading granularity? Automated decision making in the moment. Decisions that adapt as the circumstances do; in real-time and without human involvement. Where the does the human involvement come in? Metrics! Lots and lots of metrics, at an incredibly granular and accurate level. These metrics enable us to show our industry-leading latency breakdowns per-VM and per-vDisk, visualizing where bottlenecks may be: host, network or storage. For your business this means smoother predictable operations without continuous human intervention.
We’ve got an incredible amount of granular metrics that we make available for you to see. These include all kinds of information about your VMs and their related files, and their behavior. So, what are you going to do with all these metrics? If you’re an expert, you’ll leverage them to learn more about your environment and take actions on those insights. Perhaps you (as a human) won’t analyze them, but instead, let your AIOps systems have a go at them.
But what if you’re simply overwhelmed by too much data? Or more likely, you have great data, but you’re unsure how to interpret that data. Like we do everywhere else in our lives when we encounter something we don’t understand, we seek an expert on the topic. You’re in luck; we’ve got you covered!
I’ve been fortunate enough to have the pleasure to work with hundreds of customers, thousands of systems, and hundreds of thousands of VMs over the years. While I’m typically not one to boast and brag, I’m somewhat of an expert in the art of interpreting system performance data. I’ve been obsessing over graphs for decades and have been performed countless root cause analyses on systems.
Now, let’s take my intelligence and apply it to the data. In some cases, I can look at a graph and within seconds, tell you with 90% certainty what the issue is. The problem is, I don’t scale, no one does. Fortunately, computing does, so this is an incredibly exciting time for all of us in this age of big data. We can apply my learned interpretations across a large data set to find matches, then provide advice to alleviate the system. Even cooler, as we get better at this, AI and ML can not only replicate my knowledge and thus what I can do, but they’ll find things that I singularly never could!
There is a whole other level of intelligence that can be derived from analysis of data. These analyses should deliver actionable insights that help you: make your life easier, better understand your data (how it grows and changes), and bring answers to issues that may otherwise be too complex to pinpoint.
At this point, I’d like to point out that standard infrastructure (classic LUNs and Volumes, in the case of storage) can still produce useful metadata. While it might not be able to pinpoint a particular VM as the source of latency on its own, if you correlate other data sources, such as vCenter, you might be able to come to some conclusions. So, when I am calling our competitors' claims BS (as I did in my previous blog), I’d like to give credit where credit is due. They’ve done a great job analyzing the limited and non-granular dataset that they have. The analytics they’ve produced are decent as are some of the insights they provide. But more often than not, their interpretation of data and advice is specific to storage constructs (e.g., LUNs and volumes) that we opted to intelligently abstract out of the equation completely. We’ve relieved you from the burden of having to manage these, and continue to do everything possible to provide meaningful insight and analysis about what’s most important to you: your VMs and applications.
Let’s be clear. I’m not saying our competitors’ products have no intelligence whatsoever, rather I am saying that their products are standard infrastructure not Intelligent Infrastructure. They have some cool features that were designed by some highly intelligent engineers and have some incredibly intelligent marketing personnel; I’m not disputing that.
Of course, I am by no means calling their customers non-intelligent, quite the opposite in fact! It’s much harder to operate our competitors’ standard infrastructure than Tintri VMstore. My hat goes off to the skilled and talented IT professionals who dedicate so much of their time and energy to constantly tuning and analyzing their storage systems to get it just right. At the end of the day, we’re all looking to achieve the desired result: a healthy, problem-free compute environment. If it’s not, then we want to get it healthy as quickly as possible.
Realize that even the best interpretation in the world isn’t going to help if the data isn’t accurate and granular enough. Our metrics are a by-product of intelligent end-to-end design. The intelligence that our analytics deliver is fueled by the insight we gain through larger, more granular datasets.
With granular telemetry of over a million VMs across our install-base running for many years, we’re able to harness this data to derive knowledge, which we apply to further improve our file system and the supporting analytical systems. The benefits to you only increase as we continue to add advice to our Advisor within our analytics engine.
Folks: This is the real deal. With VMstore, we’ve providing you with a choice of a different experience that demonstrably can simplify your environment and substantially reduce your administrative overhead. If you want to see and experience first-hand what intelligence resides in our Intelligent Infrastructure and how it can change your environment for the better, I encourage you to reach out to us for a demo or talk to an existing Tintri customer and ask them about their experience.
We’re not looking to win buzzword bingo; we’re looking to be recognized as the rightful leaders in this space. We’ll be elaborating on many of the topics raised in this blog in future posts and other content.
Thank you for reading; I look forward to the next time. – Rob.
Unique control with VM-level actions for infrastructure functions including snapshots, replication and QoS make protection and performance certain in production, and accelerate test and development cycles.