Israel-based serverless performance monitoring company Epsagon said it recently completed a $4.1 million seed funding round that it plans to use to bolster its position in what it sees as a rapidly growing serverless computing market.
The funding round was led by Lightspeed Venture Partners, StageOne Ventures, and Ariel Maislos, CEO and founder of Stratoscale. Epsagon was founded last August by a team that had previously worked in the Israel Defense Forces’ cyber intelligence unit.
Nitzan Shapira, co-founder and CEO of Epsagon, said the funding will go toward the company’s marketing and development efforts.
“We decided to close on a large seed round because this space is exploding,” Shapira said. “Looking back even over the past six months, it has grown very fast and we expect it to grow even faster. … We think that two years from now, serverless will be the default way to build in the cloud.”
Epsagon’s platform uses artificial intelligence (AI) for performance monitoring of serverless architectures. This allows users to uncover potential performance issues before they occur.
Shapira said that understanding what’s happening in a serverless production environment is “not impossible, but pretty much impossible with current tools.”
“The general idea of performance monitoring in serverless is that if you see a service running, it’s working as expected,” Shapira explained. “That can be sort of hard to define sometimes, and you want to make sure that if a problem does crop up, you can fix it as quickly as possible.”
Serverless architectures are a challenge to monitor because unlike more traditional virtual machines (VMs) or even containers, serverless does not rely on a single, known server to house running applications. Serverless instead runs applications on the most readily available — and ideally cheapest — host to lower the cost and increase the speed of deployment.
This architecture removes a lot of the control and management from DevOps and research and development teams, making it more difficult to monitor performance.
Shapira said most performance issues with serverless computing happen at the server level, where there are “moving” parts that can lead to an application having a performance issue.
“Serverless does take away some of the control, but the issues that typically occur go wrong on the server side,” Shapira said. “Mistakes there are possible and can cause an application to break. It’s usually not something like a server being too loaded, but instead more of a logistics issue.”
Epsagon is currently focused most on Amazon Web Services’ (AWS) Lambda platform, which Shapira said accounts for nearly 90 percent of all serverless efforts. He said the company is looking to add broader support once its platform hits general availability, which he expects to happen over the next six months.
While it’s not unexpected for Epsagon to tout the market potential for serverless computing, the segment has indeed seen some positive signs toward growth.
The Cloud Native Computing Foundation (CNCF) conducted a survey late last year that found 41 percent of its community members were dabbling with serverless technology. Vendor Platform9 conducted its own survey at last year’s KubeCon event that found functions-as-a-service (FaaS) was the third most popular use case among the community.
Redmonk analyst Fintan Ryan noted in a recent blog post that growing uptake by some large enterprise customers is legitimizing serverless as an option.
“None of these use cases are trivial – they are at significant scale, and with real revenue attached to them,” Ryan wrote.
Ryan noted that from conversations, a majority of serverless application workloads are being run on Lambda. Microsoft’s Azure Functions is also getting some attention with Google’s Cloud Functions “running a distant third.”
As an interesting side note, Ryan wrote that he has also heard of enterprise developers looking to bypass work on container platforms and moving straight to serverless architectures “for suitable applications.”
“We expect this trend to pick up pace over the coming 12 to 18 months, but with various teething problems as developers adopt to this new paradigm and reach decisions on what programming languages and frameworks they find to be productive,” Ryan explained.
Epsagon’s Shapira echoed Ryan’s comments that it’s possible for organizations to at some point skip directly to serverless to run some of their applications. But, he said it was more likely that containers and serverless would continue to operate side by side due to differing use cases and performance needs.
Photo copyright: sudok1 / 123RF Stock Photo