A trio of cloud business leaders have launched a brand new firm with the mission of modernizing software program configuration information administration.
Rising from stealth at the moment with $4 million in funding, ConfigHub is the handiwork of CEO Alexis Richardson, founding father of cloud-native container administration platform Weaveworks; CTO Brian Grant, former Google software program engineer and unique lead architect of Kubernetes; and CPO Jesper Joergensen, who led in numerous product roles at Salesforce (together with Heroku) earlier than becoming a member of Twilio to steer its voice, video and platform groups.
Richardson instructed TechCrunch that their new firm is getting down to “get individuals out of config hell.”
If this simply made you blurt out, “huh?” allow us to clarify. In July final 12 months, CrowdStrike issued an awry configuration replace to its Falcon Sensor safety software program, inflicting widespread havoc and more than $5 billion in losses for Fortune 500 firms. Delta Airways mentioned it lost $500 million on account of flight disruptions, kickstarting what will be a prolonged authorized battle to recoup its losses.
That occasion demonstrated that software program is now essential infrastructure, held along with an online of interdependent elements linked by APIs to create highly effective methods and functions. Nonetheless, if any a part of that net turns into compromised, intentionally or in any other case, it could actually carry down not simply the entire home, however the complete city.
This brings us to configuration information, which ensures all these disparate components play properly with one another and that the software program and system perceive how they’re alleged to behave. However these configuration recordsdata may be an unwieldy mess, and ConfigHub is getting down to tackle simply that.
“The issue is that configuration information is scattered in every single place — it’s change into a complete sprawl,” Richardson mentioned.
Fixing DevOps
A lot has modified in the way in which software program is deployed and managed over the previous 30 years. Within the days of yore, enterprise software program was largely probably deployed through CD-ROMs on localized {hardware}, and configuration was restricted to a couple textual content tiles that instructed the working system or functions the place to seek out the issues they wanted to run correctly.
Right now, it’s not fairly so easy. Gargantuan configuration recordsdata with hundreds of strains of code is required to make the software program run in what are sometimes dynamic, large-scale environments.
Joergensen says he discovered a precious lesson at Heroku (a Salesforce-owned platform-as-a-service): Code and configuration information are usually not the identical, and require very totally different approaches. Configuration information defines system settings and behaviors that may’t be debugged in the identical means as software program code can. However understanding this information — whether or not with instruments like TerraForm or Kubernetes — is important for avoiding expensive misconfigurations that may trigger outages or delays.
“We now have made nice progress on evolving collaborative coding practices with instruments like GitHub,” Joergensen mentioned. “However the configuration of stay infrastructure wants a distinct method. It isn’t only a bunch of recordsdata. Our objective is to carry the elegant app developer expertise, pioneered by Heroku, to each sort of stay manufacturing operations.”
As such, ConfigHub guarantees to “unify configuration administration with fashionable, automated improvement workflows and compliance.” As an alternative of getting to go trying to find the precise piece of configuration that corresponds to a given error, on ConfigHub, every thing will probably be held in a single database, making configs simple to seek out, replete with a stay view that exhibits what the system is definitely doing.
“So in case your buyer can’t entry the system, you need to have the ability to see and expertise what they’re seeing and experiencing,” Richardson added. “The configuration information in our system will probably be an expression of the stay manufacturing system, which permits the staff not solely to determine the info, but in addition to vary it, in order that they will repair the shopper’s downside instantly.”
The product, which will probably be served through a SaaS mannequin, will focus initially on Kubernetes DevOps tooling comparable to Helm, Argo, Flux, Terraform and its open supply fork Opentofu. Richardson says that ConfigHub is already working with some “medium to massive” enterprise design companions, although he didn’t reveal names.
The product remains to be some months from formal launch, so at the moment’s announcement is extra about ConfigHub introducing itself to the world and asserting its $4 million funding. The funding comes from some notable VC and angel traders, together with Crane Enterprise Companions, Encoded Ventures, Pear VC, and Poolside CEO Jason Warner.
The AI issue
Whereas the CrowdStrike episode was one of the expensive configuration clangers in latest instances, it wasn’t an remoted incident by any stretch.
“Configuration modifications have, for a very long time, been often called the main explanation for cloud system failures,” Grant instructed TechCrunch.
Final 12 months, for instance, Australian superannuation fund UniSuper was hit with an outage that left some half-a-million members with out entry to their accounts, after a Google misconfiguration led to UniSuper’s Google Cloud VMware Engine (GCVE) personal cloud being deleted.
And again in January, a configuration error on GitHub’s cloud meant all Git operations had been unavailable for as much as two hours.
The issue goes to worsen — AI will compound issues at a time when firms are racing to embed AI into the material of their software program.
“AI utterly upends how the complete world writes, makes, and operates software program totally, not simply code writing,” Warner mentioned in an announcement. “The subsequent few years are far more than simply vibing about small modifications to greenfield code bases. Enterprises want mature options to arduous issues. ConfigHub is a lacking piece to permit this rewrite to happen safely.”
Corporations can, after all, already construct their very own options from a concoction of instruments comparable to Terraform, Kubernetes, Prometheus and others. However these may be overly complicated to arrange and handle.
There are big-name incumbents, too, comparable to ServiceNow and Atlassian that provide companies to handle configuration points. However such firms had been based greater than 20 years in the past, earlier than cloud computing had actually taken off. Furthermore, many such firms have long-expanded into the realms of HR, CRMs, staff collaboration and challenge administration.
“These sorts of instruments are usually not well-suited to the fashionable stack of dynamic, cloud-native, AI-powered, containerized functions,” Richardson added. “You possibly can’t have any person attempting to roll out AI if a easy misconfiguration implies that the AI will see their firm information and publish it on the web.”
Three’s a cloud
Previous to ConfigHub, Richardson based an enterprise-focused cloud messaging firm known as RabbitMQ, which was acquired by VMWare subsidiary SpringSource in 2010. Then in 2014, he based Weaveworks, which went on to boost greater than $60 million in funding from a who’s who of traders together with Accel, Amazon Internet Companies and Google Ventures. At Weaveworks, Richardson additionally developed the GitOps framework, which is now broadly utilized in cloud-native and Kubernetes environments for managing infrastructure and utility deployment.
Weaveworks formally shut store final February, the results of “lumpy” gross sales and lack of recent capital, Richardson wrote on the time. So in mid-2024, he teamed up with Grant and Joergensen to begin ConfigHub.
Grant and Richardson met by their work on the Cloud Native Computing Basis (CNCF), which counted Kubernetes as its inaugural challenge after Google donated it in 2015. Round 2020, Grant began giving up his numerous management roles across the Kubernetes challenge, and over time, he developed an concept for a brand new enterprise whose focus probably wouldn’t align with Google’s priorities. This would finally align, although, with Weaveworks coming to an premature finish final 12 months.
Making the leap from one of many planet’s largest know-how firms to the world of startups looks as if an enormous transition. Nonetheless, Grant labored at two startups earlier than Google — for one, he was chief architect at a small startup known as PeakStream, which Google acquired in 2007.
“I used to be the third engineer there — after we began, we didn’t actually have a title, we didn’t have an workplace, we didn’t have a marketing strategy,” Grant mentioned. “So I’ve finished very early startups earlier than.”
Whereas Google continued to develop exponentially in his 17 years on the firm, Grant’s involvement with Kubernetes, which Google open sourced in 2014, was additionally just a little akin to working in a startup.
“Kubernetes began as simply 4 engineers and a product supervisor; we didn’t actually have a supervisor in widespread, all of us reported to totally different individuals,” Grant continued. “It was very very similar to a startup in its personal proper.”
Richardson, in the meantime, met Joergensen across the time VMware acquired RabbitMQ greater than 10 years in the past. After serving in numerous senior roles at among the world’s largest tech firms, Joergensen too has now elected to step into the world of startups.
“You will get a lot extra finished now than ever earlier than — we’re leaping proper into the rip-current of AI, and I in all probability wouldn’t wish to be every other place proper now,” Joergensen mentioned. “In a startup, we get to actually see with readability what may be finished.”