Creating an end-to-end cloud operational experience – Under-the-hood series part 2

[Ronak_Chokshi]

This is the second in the “Under-the-hood” series of blogs that I started in Oct 2021. In this series, I offer readers an in-depth view of how HPE InfoSight approaches difficult infrastructure problems with artificial intelligence (AI) and machine learning (ML), as well as learn how we make infrastructure invisible for customers. My first blog in this series focused on best practices that we follow in data science. Today’s blog focuses on our approach to user experience. It’s written in partnership with our user experience (UX) team.

It’s no secret that enterprise IT environments have gotten more complex over time than we would like them to be. Have you ever wondered if you could get visibility of your entire infrastructure stack and operations?

Well, you are not alone. In fact, this IDC study reports that 39% of enterprises consider improving visibility into their data center operations and costs among their top 5 initiatives.

These days, infrastructure is viewed as a means to an end – the end being to support the business in deploying new applications and moving your digital transformation initiatives forward. It’s a journey, really. Businesses want to be agile and stay competitive in their markets. IT administrators want to be able to provision their workloads on the right infrastructure without worrying about, for instance the capacity headroom. That said, they also want to be in the know of any potential infrastructure issues. They want predictive capabilities to address any impending issues with the individual infrastructure components.

We call this an end-to-end cloud operational experience

This is an experience in which IT operations are transformed from infrastructure-centric to application-centric. An experience that makes infrastructure invisible, and operations autonomous.

HPE InfoSight has played a critical role here for more than a decade for our customers, leading the industry in bringing autonomous technologies to their IT operations. HPE offers an industry-leading cloud operational experience with the HPE GreenLake edge-to-cloud platform and HPE InfoSight powers that experience.

As we continue to innovate with HPE InfoSight, in September 2021 we announced App Insights to improve the visibility by providing analytics and insights into the application workloads running on HPE infrastructure. With HPE InfoSight App Insights, we establish that adjacent layers of the stack almost always have clues about each other which can be used to “stitch” them together and infer exact footprint for all application entities.

Next-gen visibility

IT administrators have relied on a combination of application performance monitoring and infrastructure monitoring tools for visibility into their infrastructure. In addition, correlation among the dashboards and deriving meaningful insights is often done manually and has required domain expertise. For instance, if a medical imaging application is slowing down, the performance analysis and troubleshooting of this issue would require the application expert, IT administrator, VM admin and storage administrator to work together and manually correlate the events occurring in their respective domains. Such an approach with siloed tools and manual troubleshooting is inefficient and results in firefighting.

HPE InfoSight App Insights is our attempt to change that by offering a new AI-driven approach for next-gen visibility into your IT stack with deep insights never seen before. It starts with a new user experience (UX) built with simplicity, automation, and ease-of-use.

Let’s consider this example to help visualize what App Insights offers: Say you ran a Microsoft SQL server with eight databases on a bunch of storage and server systems.

As you can tell, the above topology covers the entire stack from the SQL server application down to the storage volume. Imagine if you had the visibility into each of these layers along with performance analytics and predictive insights as data moved up and down the stack, in near-real-time. Imagine if you were presented with accurate correlations between these individual layers in a unified dashboard, so you could accelerate root-cause analysis and attribute the problem area quicker than ever before.

Here is another look at detailed time series charts showing memory usage, transaction rates, wait times, IO read/write and more, that the user can access.

How we are making such visibility possible

The new user experience of App Insights has come to fruition as a result of a 3-step process:

1. Configuration extraction

Using a combination of plugins (e.g. VMware APIs/SDK) and an on-premises agent, we first collect the configuration values for each of the domains at regular intervals. This includes attributes across application, OS, virtualization software, compute, and storage. This data is then transformed to conform to a standardized schema format for faster analytics in the cloud.

2. Topology stitching

This is where attributes from the different entities of the topology are brought together into a data processing pipeline. Using a combination of open-source streaming and database technologies, the new data arriving every hour is merged with the existing configuration data.

3. Topology representation

This is arguably the most important step, as the entire resulting topology must be made available in varying time-windows as well as for quick querying for data science and machine-learning purposes. As the number of users using App Insights potentially scale, this must be done without straining the resources, hence avoiding delayed responses.

User experience is key to differentiation

This experience delivers three distinct advantages to our customers.

First, HPE InfoSight App Insights offers the end-to-end visibility that is second to none. With this visibility, event correlations and insights, admins can attribute infrastructure problems to areas of concern faster than ever before. Secondly, we’re removing the barriers arising from domain expertise that many of the other tools require to manage and operate. And lastly, this allows administrators to transition to an AI-driven, non-siloed approach for their application landscape, alleviating the need for constant monitoring of their application workloads.

To experience the AI-driven infrastructure and HPE InfoSight App Insights, visit HPE InfoSight or log onto HPE InfoSight portal.

---

Ronak Chokshi
Hewlett Packard Enterprise

twitter.com/HPE_Storage
linkedin.com/showcase/hpestorage/
hpe.com/storage