By: David Wang, Director, Product Marketing, Nimble Storage
InfoSight has forever changed how storage infrastructure is managed. Before Nimble Storage, IT had the displeasure of spending countless hours dealing with unexpected problems and pulling out their hair talking to vendor support. And now with InfoSight, problems are predicted and prevented and support has been transformed into an experience customers actually like.
Competitors have seen the impact of our platform and “…have attempted to replicate the benefits”, as noted by IDC. But, our capabilities are unique and that has led IDC to recently write that InfoSight is “the industry-leading cloud-based predictive analytics platform”1 .
And what does being industry-leading do for you? With InfoSight, you get over 99.9999% measured availability from your storage infrastructure and automatic prediction and resolution to 86% of problems, including those outside of storage. So, if ensuring data reliably gets to your apps and spending less time managing storage are important to you, here are four key differences that separate InfoSight from the rest.
Architected for analytics
We consciously built predictive analytics into our architecture right from the start. This required taking a fundamentally different approach to data collection and analysis to expand the telemetry analyzed well beyond standard logs and metrics.[img src=”/wp-content/uploads/sites/1706/2017/06/architect.png” class=”aligncenter”]
Our architecture embeds deep diagnostic sensors into every module of code in our operating system (NimbleOS). And, we’ve been adding sensors since the first line of code was written. Since we have connected our installed based in the cloud, InfoSight continuously correlates data from over 10,000 customers. So today, InfoSight collects and analyzes millions of sensor data points every second. That’s over 300 trillion pieces of data analyzed to date.
Our expansive network of globally connected systems and rich telemetry forms the foundation for InfoSight to deliver predictive insights about your environment. And, these are not just insights for storage.
Awareness beyond storage
Ensuring data reliably gets to your apps requires understanding and resolving all the issues that can impact the I/O path. More than half the issues that cause app disruptions are caused by factors outside of storage. So, vendors with visibility limited to their arrays are missing more than half the problems.[img src=”/wp-content/uploads/sites/1706/2017/06/awareness.png” class=”aligncenter”]
InfoSight’s visibility spans your infrastructure stack to resolve non-storage issues too. Not only are we collecting data from the arrays, but we’re also collecting data across the stack and correlating it to provide a single source of truth from storage to the VMs. InfoSight’s analytics, in turn, help you rapidly root cause and even prevent problems that are outside of storage. Here are two examples:
- Performance Diagnostics: Everyone knows how hard it is to pinpoint performance bottlenecks. Because of the complexity in infrastructure, you have to spend time collecting and stitching log files and performance data from each layer just to understand what’s happening. InfoSight’s correlations tell you the primary factor causing the problem across the host, network and storage layers. It can also diagnose “noisy neighbor” VMs, misconfigurations in the infrastructure, and hosts that are not configured correctly. With InfoSight, complex performance issues get resolved in a fraction of the time.
- Problems You Don’t Ever See: InfoSight has over 800 predictive signatures that are constantly being monitored in every system – predicting and preventing issues. These signatures even include problems outside of storage that are not only difficult to pinpoint, but have big impact. For example, a customer experienced a 10X performance drop after updating to a new version of NimbleOS. InfoSight diagnosed that the hypervisor was sending the array an illegal iSCI response that caused excessive retries. We quickly developed a fix for the customer, then applied a signature pattern matching algorithm to see who else could be susceptible. InfoSight found over 600 systems and blacklisted that version of NimbleOS from each system’s update path, preventing them from experiencing the issue.
Clearly, InfoSight sees what others can’t and makes problems go away across the stack.
Applied data science and machine learning
At Nimble, data science isn’t an experiment. InfoSight applies advanced techniques to create statistically accurate predictions and insights into your environment. This enables InfoSight to take the guess work out of managing infrastructure from planning to expanding.[img src=”/wp-content/uploads/sites/1706/2017/06/applieddata.png” class=”aligncenter”]
Planning for new infrastructure is a cinch because InfoSight’s app-centric modeling correctly sizes infrastructure needs based on your applications, users, and requirements. And, as new infrastructure is deployed, machine learning algorithms monitor performance and refine the models.
You won’t hit a ceiling with Nimble arrays. InfoSight accurately predicts capacity, performance and bandwidth needs. Our capacity forecasting models use autoregressive and Monte Carlo simulations taking into account irregular patterns that would appear unpredictable to untrained eyes. And, InfoSight applies an eigenvalue decomposition algorithm to automatically identify the primary cause of latency problems and predict the right solution.
InfoSight is constantly learning from the collective insights and experiences of the installed base. It’s learning I/O signatures and performance patterns, it’s learning from past support cases, and it’s learning to predict problematic behavior in your infrastructure. As it learns, the predictive algorithms continue improving and every system gets smarter and more reliable.
Just showing charts and graphs isn’t enough anymore. It’s too difficult and takes too long trying to interpret them, so you end up with more questions than answers. For example, what if you see spikes when looking at a latency graph – should you be concerned? Are they indicative of a problem? If so, what’s causing them? And, how do you fix them?[img src=”/wp-content/uploads/sites/1706/2017/06/spikes.png” class=”aligncenter”]
Instead of scratching your head, InfoSight delivers prescriptive, clear recommendations. For performance, InfoSight understands the underlining I/O signatures and the impact different workloads have on users. Applications are much more sensitive to latency when they are performing many random small-block vs large block operations. So, InfoSight predicts when performance is problematic and highlights those time periods in the UI. And based on the analytics, performance improvements include applying QoS controls on IOPS and throughput, optimizing resources, and upgrading components.
InfoSight constantly monitors your environment and automatically gives you recommendations to make every day easier. It can even tell you what bandwidth is required for replication and what volumes are not being protected based on factors like data change rate, compressibility, etc. And with proactive alerts and case automation, InfoSight automatically opens support cases and resolves them even before you know there’s an issue. As mentioned above, over 86% of issues are identified and resolved by InfoSight before you even know there is a problem.
Unique and getting even better
So, the next time you hear another vendor say “We can do what InfoSight does too”, just ask them how they compare across these four dimensions. InfoSight is in a league of its own and the business benefits are invaluable.
And now that Nimble Storage has joined HPE, it’ll be even harder to catch up to InfoSight. Just imagine the predictive insights that can come from having the largest database of infrastructure telemetry and connected systems in the industry. Stay tuned for the exciting future for InfoSight at HPE.
1 IDC, “HPE Hits a Home Run with the Acquisition of Midmarket Enterprise Storage Vendor Nimble Storage”
Author: David Wang