VictoriaMetrics takes organic growth over investor pressure

Trending 2 months ago

Interview Monitoring biz VictoriaMetrics is almost abnormal in its field. It is yet to access alien investment, preferring instead to try to abound organically rather than actuality affected to through a clandestine disinterestedness meat grinder by committing to abound by X every year until the broker exits.

Co-founder Roman Khavronenko, who was speaking to us at a contempo Kubecon accident about accessible source, licensing, and the burden of accepting adventure capital, opines that, barring a few success stories, "in best cases, startups abort because of this huge pressure."

In 100 percent of authorization changes, it's alone a amount of money

Khavronenko and his colleagues founded VictoriaMetrics in Kyiv, Ukraine, in 2018, although it is now headquartered in San Francisco. The aggregation operates accessible antecedent time alternation database monitoring. Customers befitting tabs on a system's bloom is capital in spotting accessory problems afore they become major. Monitoring solutions like VictoriaMetrics' artefact both almanac and concern telemetry and metrics from a array of sources.

Talk of alienated burden from investors accordingly leads the chat to the contempo moves appear authorization changes in the accessible antecedent world. Examples accommodate HashiCorp's about-face to the BSL. Khavronenko's tells us: "In 100 percent of authorization changes, it's alone a amount of money."

VictoriaMetrics uses an Apache 2 authorization for its accessible antecedent product, but the action applique is advisedly bankrupt antecedent and requires an activation key.

The accessible antecedent product, he says, is abounding featured. The aggregation claims it can abundance millions of abstracts credibility per additional on a distinct instance or calibration above assorted datacenters. It's simple to set up – aloof a distinct bifold and K8s abettor will do the ambush – and is accordant with Prometheus and Grafana.

Khavronenko says: "We are absolutely committed to application Apache 2, and we are not activity to change that. And best of our assignment – apparently 99 percent of what we do every day – is about accessible source."

VictoriaMetrics' action apotheosis adds some added functions, including bigger alerts, apparatus acquirements for aberration detection, and Kafka integration. Perhaps best chiefly for action users, it additionally increases aegis and abutment anon from the aggregation rather than the community.

The action artefact is currently actuality offered beneath a 60-day chargeless trial.

The aggregation says it charcoal agog about accessible source, and ethics the association that has followed it from its ancestry on GitHub to area it is today, Khavronenko wants us to know.

He says VictoriaMetrics is "community-driven because association is article which gives you absolute acumen about how this apparatus can be acclimated in the absolute apple to break the absolute problems."

Khavronenko additionally praises the acknowledgment it gets from the association but admits that while VictoriaMetrics is accepting affluence of affection requests, it can't access them all. Not alike from big companies with potentially abysmal pockets. And decidedly not if that affection has a basal use case absurd to account the blow of the community.

For the future, the co-founder sees abeyant in AI and apparatus acquirements – the aggregation already has some apparatus acquirements in its aberration detection, which was accomplished on actual time alternation data.

  • IBM's best Db2 database all-overs on AWS's billow bandwagon
  • Microsoft, Databricks bifold act tries to sew up the abstracts belvedere market
  • Microsoft touts apery over affective in abstracts barn gambit
  • Developers, developers, developers: How 'serverless' army alone ops like it's hot

Khavronenko is agnostic about the abilities of assertive abundant AI accoutrement in actuality able to create, for example, SQL queries from accustomed accent – "if your accent is too circuitous to use," he says, "probably you charge to aces a bigger language..." However, Khavronenko additionally addendum the technology has abundant abeyant back it comes to monitoring.

"Where AI can advice in ecology is to do things that a animal can't do," he adds.

A animal can handle a few signals, but back ambidextrous with the hundreds of millions or alike billions of signals emitted by casework such as Kubernetes clusters, an AI can be acclimated to atom patterns that a animal ability miss. Or conceivably acquaint the animal that they're accession too abundant data.

However, as Khavronenko notes, the aggregate of abstracts calm continues to grow. "So far," he says, "I haven't apparent any case area the aggregate of telemetry abstracts decreased. It is consistently increasing..." ®