With the advent of cloud computing and open-source software, small-scale companies can now operate large-scale production environments. But while a nimble operation can punch far above its weight, it may struggle to manage the large amount of data its systems generate. Even if there are monitoring tools to catch errors, teams may not know what, where, or when to search. That’s where Coralogix comes in. By deploying machine learning to crunch millions of log records, it helps companies understand their log data and work proactively to maintain their systems. But it’s not just for tiny startups—its clients include giants like Lufthansa and Campaigner.

Coralogix co-founder and CPO Ariel Assaraf started the company after completing his service in the Israeli army. He was working as the Quality Architect of Variant Systems when he noticed that there was a high error rate in what they were delivering to customers, even though they had multiple tools for production delivery and monitoring. He began to ask his teams why the tools to catch these errors weren’t enough, and discovered that even though the data was there, it was so overwhelming that the most basic parameters of what to look for weren’t clear.

“So we decided to build Coralogix,” Assaraf says, to “automatically figure out insights out of massive amounts of data and actually pinpoint issues so that people can understand what’s wrong, where to search, and when to actually approach the system and analyze problems.”

With Coralogix, companies receive analytics that automatically cluster millions of log records back into actionable patterns, helping engineers see baseline flows and error rates more swiftly, and empowering companies to innovate more and more. This is of special use to small startups with large cloud-enabled production environments.

“Suddenly five people can have a hundred, two hundred, three hundred servers online,” Assaraf explains. “We saw that, as CI/CD is evolving as companies release more and more versions, it’s becoming extremely hard to understand and impact a version release.” With Coralogix software, “we help them figure out problems and analyze their new versions automatically so that they can fearlessly release.”

Coralogix uses machine learning to achieve this, by dividing its activity into two moves. One is to “organize data so that people can get better observability and understand their data better,” explains Assaraf. “The other thing is to generate out-of-the-box automated insights.” Coralogix automatically figures out flows and sequences of logged data—which logs arrived together at a certain time frame, ratio, and order, so users get pinpoint insights on specific components or cross components.

“We understand patterns and behaviors of each and every component in a customer’s environment and generate automated insights that tell them when they have a bad trend of errors, of response codes, or whatever,” Assaraf continues. “Also, we cluster the data automatically. We tie data into versions, we understand security threats automatically, and these are tools that allow people to do their span of work, only better.”

However, it’s not just error logs that Coralogix is able to parse. “More than just generating insights and proactive insight, companies need help in managing the work flows and policies,” Assaraf says. “So we’re going to make efforts this year to combine the insights that we deliver into policies that R&D managers can actually apply to their workforce. Let’s say zero broken windows, no errors in specific components, and better CI/CD maintaining their SLAs.”

As Coralogix scales up, it has in its sights the whole log analytics market. It’s a huge market—worth billions of dollars a year. “As a younger company, it’s been a challenge to educate the market that machine learning is actually needed,” Assaraf says. But just in the past year, they’ve been able to grow the business 15x in terms of revenue. It doesn’t take machine learning to see that that’s no error.

from AWS Startups Blog


0 Comments

Leave a Reply

Your email address will not be published. Required fields are marked *