Imagine an application that could take all rote tasks out of a job and leave only the bits that one loves doing. Such an application may not exist today, but the building blocks are here already.
The Selector platform presented recently at the Cloud Field Day event in February does something close for enterprise administrators.
The platform, enabled with artificial intelligence (AI), is designed to identify sources of network incidents and drive automated troubleshooting on them – a work that has administrators working around the clock in a spiral.
In hybrid multi-cloud environments where data packets run through multiple hops touching scores of platforms between the source and destination, following the path is notoriously difficult.
It’s a “needle in a stack of needles” situation, says Sachin Natu, VP of product. “What is driving the complexity is you’re dealing with multitude of different technologies which are built across generations in last 30 years and multiple administrative domains. Each of them have their own siloed view.”
A day in the life of an enterprise administrator explains this best. An application has gone down and the administrator is tasked with the job of finding where the issue is. It could be anywhere on premises, in the cloud or with the ISPs.
Typically, they would start looking in the corporate office stacks where all the familiar technologies live. But if they are not lucky, they need to cover the vast expanse of ISPs and cloud vendors that lie beyond the corporate offices.
This is the reason why the network operating center (NoC) looks such a hodge podge of dashboard, application and tooling.
“You’re dealing with different types of vendors, different types of technologies, and depending on what information you are pulling in, you are talking different methods,” Natu highlighted.
It is easy to go down the rabbit hole but never find the bottom. But the administrator needs to act fast. They must skim through all potential data sources, consume relevant information, quickly make connections, and come up with an answer that can enable a prompt troubleshooting. It’s a job that spans multiple individuals and entities across multiple domains.
Gray failures are even more notorious to debug, Natu pointed out. “If something is down, it’s really cut and dry…But if the link is dumping a lot of traffic, that’s where the challenge becomes difficult..because for one location, things are great, for another, things are not.”
The Selector AI platform essentially performs and augments the steps of investigation and troubleshooting. A visibility and automation solution that can be deployed on both public cloud and on-prem locations, and is also available as SaaS, its goal is to replace the labor of root cause analysis and troubleshooting, no matter where the problem, before they impact users’ experience.
“Our goal really is to be proactive. We want to be filing a ticket and solving the problem before somebody even notices that,” Natu said.
Three key capabilities make this possible – the platform’s inherent ability to ingest data from all databases, monitoring tools and other sources in the environment, a causal ML event correlation mechanism and a digital twin.
“We can ingest config, events, metrics, alerts, logs, topology information…or any other information that the customer may have,” Deba Mohanty, VP of Solutions, said.
“If you look at existing platforms or monitoring tools, they are primarily focused on a single type of data source whether it’s metrics, logs or events. So primarily, when someone is looking at the whole infrastructure, they are looking at multiple different data sources and devices to figure out what’s happening,” he added.
With focus on eliminating the inefficiencies in the process, Selector throws together various combinations of these data, whittle it down to events and make correlations between them. Working ceaselessly in the background, it derives intel on “what is happening and why it is happening”, pushing out summaries to the waiting administrators at the other end via a host of apps.
“We correlate the information and file an actionable insight and a ticket. That ticket can go to a ServiceNow or PagerDuty or ITSM tool. That alert can go into Slack and Teams.”
The corollary is instant reduction in the volume of tickets. But the real advantage, Mohanty says, is that the operation happens proactively.
To provide a future perspective of the network health and state, Selector uses a digital twin that combines all operational information about the systems to anticipate problems in the future. The best use of the simulation of failures is for analysis of what-if scenarios, Mohanty said.
Users can access information on the platform in two ways – through a copilot that gives back answers to questions in near-real time, much like ChatGPT, Gemini or any AI chatbot, or through an SQL interface that can be interacted with via queries.
In order to keep pricing predictable, Selector charges customers by use cases and number of devices monitored only.
Selector’s target customers are companies in the telco space, financial firms, retailers and media networks.