Virtualization Technology News and Information
Article
RSS
VMblog Expert Interview: Selector Talks Network Observability Challenges, Making Sense of Network Data, Skills Needed in the New Era of AIOps

interview-selector-antich 

The purpose of observability is to provide network engineers, DevOps and IT teams the ability to understand what’s going on across various environments and systems so they can diagnose and troubleshoot issues faster and more efficiently.  One of those companies helping to make sense of that data is Selector.  They use artificial intelligence and machine learning to provide actionable insights to network, cloud, and application operators.

To find out more about the challenges in this space, and what organizations need to know, VMblog reached out to industry expert, Javier Antich, Product Manager at Selector

VMblog:  What are the three main challenges in network observability today?

Javier Antich:  The network is highly dispersed, naturally siloed, and constantly producing a broad range of data such as logs, metrics, events, configuration, and flows. Each type of data is inherently different from the next, requiring multiple disparate tools to reliably collect and store it. This data fragmentation creates a significant obstacle to achieving proper network observability.

Additionally, there is a lack of tooling to support a comprehensive analysis of this data, preventing cross-domain correlation. Enhancing this data with the correct contextual information enables downstream algorithms to "connect the dots" across multiple domains and surface meaningful insights.

Lastly, teams find it challenging to interrogate the data in real-time. Network operation teams must quickly resolve incidents and issues, often under pressure from customers and management. These teams often struggle with leveraging dozens of tools, navigating disjointed dashboards, and executing complex SQL-like queries to obtain the needed insights to triage a given problem.

VMblog:  Monitoring network infrastructure is fundamental to ensuring the continuous delivery and operation of IT services. How has the ecosystem evolved over the last few years?

Antich:  Digital transformation initiatives have made the performance and reliability of the network and IT infrastructure critical to the business. Traditional monitoring is no longer sufficient to meet these needs, with increasing pressure to detect and remediate issues faster and more efficiently.

A new approach, called Observability, offers deeper analysis and contextualization of data. However, Observability may be insufficient to surface more complex issues or anomalies within enterprise-class environments.

A new class of tools, AIOps, is now being defined to address these evolving requirements. Within AIOPs, different AI/ML techniques are leveraged to analyze and correlate data across one or multiple domains to achieve a comprehensive view of the environment.

VMblog:  There are multiple types of data a network infrastructure can produce. Examples include SNMP data, log messages, and/or events. Which type of data is the most important?

Antich:  The most important type of data is the one most often not collected -- metadata. Metadata is the data of the data. It is the information used to contextualize the data, allowing for proper analysis and observability.

In the case of Selector, the platform treats data and metadata as first-class citizens, leading to a broad range of insights. This results in a fundamental difference in terms of the quality and quantity of surfaced insights.

VMblog:  Network engineers oftentimes become overwhelmed with the deluge of data coming into the network. How does Selector Analytics make sense of all this data to eliminate data fatigue and enable engineers to make better decisions?

Antich:  Network engineers oftentimes become overwhelmed with the deluge of data coming into the network. How does Selector Analytics make sense of all this data to eliminate data fatigue and enable engineers to make better decisions?A key challenge for network operations engineers -- most of the time,  is not the lack of data but too much of it. The volume of metrics, events, and logs can quickly overwhelm operations teams, leaving teams to struggle to make sense of all the data.

A key challenge for network operations engineers -- most of the time,  is not the lack of data but too much of it. The volume of metrics, events, and logs can quickly overwhelm operations teams, leaving teams to struggle to make sense of all the data.

Selector directly addresses this challenge through two complementary strategies. First, enrichment mechanisms ensure the contextualization of data before ingestion into the platform. Second, Selector consolidates and prioritizes alerts, ensuring operations teams aren't deluged with unimportant or duplicate alerts.

VMblog:  What is the role of Machine Learning in facilitating better network observability?

Antich:  Machine learning significantly enhances network observability through various methods, such as identifying anomalies, predicting trends, detecting outliers, grouping logs, and recognizing entities.

While Selector's platform incorporates multiple techniques, it's crucial to remember that high-quality, relevant data is the foundation for utilizing advanced ML. This data-centric approach provides a great deal of flexibility and allows us to deliver outcomes currently unseen within the market.

VMblog:  What skills should today's operation engineers possess to be successful in this new era of AIOps and network observability?

Antich:  Network and cloud technologies evolve rapidly, requiring a near-constant knowledge refresh and update. As a result, network engineers and operators are constantly experimenting, learning, and innovating. 

This community, out of necessity, has embraced automation and ramped up its software development skills. As AIOP solutions become more pervasively deployed, it will also be important to become to understand what is behind the "ML black box," along with becoming comfortable with closed-loop automation.

Trust in AIOps solutions can only be built if network engineers have a clear insight into what type of decisions ML algorithms are taking, why, and how. Recently, I read a statement that said "AI will not take your job, but someone that uses AI will" -- I believe this is on point. Embracing AI/ML will empower individuals and teams, helping them deliver value for their respective businesses, along with becoming more competitive in the general marketplace.

VMblog:  We are entering into an "All as a Service" era. How will this impact Selector Software?

Antich:  Digital technologies are increasingly delivered through cloud-based models, enabling new business models such as subscription-based services and event consumption.

At Selector, we embrace the concept of "Observability as a Service" to simplify the process for customers and reduce complexity. We believe that customers should pay for outcomes rather than technology. Our Platform is designed to be cloud-native, allowing for in-the-cloud, on-premises, or even hybrid options.

VMblog:  What would Selector recommend to an enterprise organization planning to develop an in-house observability tool?

Antich:  Often, we find custom, bespoke solutions put together by internal software development teams to help bridge their observability requirements. While this can be effective if it aligns with the organization's core competencies, but can also be costly, time-consuming, and difficult to maintain over the long term.

Selector offers an alternative for organizations that want control over the outcome but do not want to invest heavily in building a platform. Our engagement model allows customers to tailor the solution around their respective use cases by leveraging the flexibility of our platform and architecture. Customers can either work with our solution engineering team or use their resources to build specific use cases that meet their requirements while avoiding the cost of building and maintaining generic platform building blocks.

##

Published Tuesday, January 31, 2023 7:35 AM by David Marshall
Filed under: ,
Comments
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
Calendar
<January 2023>
SuMoTuWeThFrSa
25262728293031
1234567
891011121314
15161718192021
22232425262728
2930311234