How Claritev cut $150,000 in Kafka support costs with Kpow

A new adopter learning Kafka in a high-stakes environment

Claritev is a healthcare technology, data, and insights company. It sits between payers, providers, and members in the US healthcare system, using data analytics to help each side get a fairer price. As Dave Gale, who leads Claritev’s middleware team, put it: “Our objective is to help the payers, providers, and members all save money. The payers are the insurance companies, the providers are the doctors and hospitals, and the members are you and me.”

Dave’s team runs Claritev’s middleware stack, including Kafka, Kong, WebLogic, Tomcat, and Solr, supporting claims processing workloads where mistakes carry real financial consequences and Kafka throughput runs into the millions of messages a day. As Dave put it: “If you make a mistake with somebody’s claims processing, you cost people millions of dollars.” That’s why upgrades move slowly, soaking for a month at a time across development, SQA, and CIT before reaching production.

Gaining visibility into Kafka as a new adopter

Claritev was new to Kafka when it first stood it up, and Kpow was there from day one, included as part of an early Kafka support engagement. “We’d never seen Kafka before,” Dave said. “They said, ‘By the way, with our offering you get this Kpow tool.’” What started as a given turned out to shape how the team worked with Kafka from the very beginning.

That changed once developers started working with the platform day to day. It became apparent that Kpow was critical to their use of it. It allowed them to see their consumers and producers, and they could manually produce and consume messages within the tool. The alternative was the Kafka CLI: workable, but slow and fiddly for anything beyond the basics.

The stakes got real when Kpow was later separated from the original support bundle, and Dave had to justify a net-new software purchase to his executives. His developers pushed back hard at the prospect of losing it. “When we told the developers there was a good chance they wouldn’t have Kpow anymore, they were very upset. They’re the ones that helped us come up with the justification for why we needed to purchase it.”

That justification came down to time. “It’s the amount of time and effort a developer would have to spend managing those Kafka clusters without Kpow. When your project is a week behind because every time you wanted to do something you had to write custom scripts or go through the CLI, it’s a fairly simple way to justify the cost.”

Claritev also compared Kpow against a per-user competitor. With six teams of three to four developers each, plus a ten-person middleware team, that’s around forty users, well past the point where per-user licensing tips into enterprise pricing. “Kpow is priced by the cluster, and the competitor charges by the user. As soon as you’re at forty users, the cost from a per-user model starts to go up.” Claritev runs six clusters, and the math favored Kpow.

Kpow as the backbone of a self-managed Kafka operation

Day to day, Kpow is where Claritev’s developers go to work directly with Kafka: inspecting messages in a topic, checking whether a consumer is behaving as expected, and manually producing test messages without touching the CLI. That visibility spans four production clusters and two pre-production, handling millions of messages a day, and it’s part of a disciplined upgrade rhythm, with Kpow patched quarterly alongside the rest of Claritev’s middleware stack and soaked for a month in each environment before reaching production.

That cluster-based model is paying off again as renewal pricing on Claritev’s third-party Kafka support contract rose toward $150,000 a year, for a service the team called on only a handful of times annually. Because Kpow already gave the middleware team deep, real-time visibility into cluster health, Claritev had the confidence to bring that support in-house rather than renew. “Kafka is open source, and with the visibility Kpow gives us, it’s fairly straightforward for us to manage ourselves,” Dave said.

That confidence is also carrying Claritev through a migration of its self-managed Kafka from Rancher Kubernetes to Oracle Kubernetes (OKE). Factor House worked directly with the team on the move, providing setup instructions for running Kpow on OKE. Kpow’s role-based access controls also govern who can do what across that environment, separating elevated, auditable admin access for the middleware team from read-only access for developers in production and the client-facing CIT environment.

Lower costs and faster fixes

Bringing Kafka support in-house is on track to save Claritev roughly $150,000 a year, the full cost of the prior support contract, without any loss of operational confidence. That’s a direct result of the visibility and self-service diagnostics Kpow gives the middleware team day to day.

The clearest example of that visibility is diagnostic speed. In one incident, Kubernetes reported every pod healthy and Prometheus raised no alert, yet a Kafka cluster was degraded. “We went into Kpow and could see the brokers were out of sync. It showed us exactly which broker wasn’t running. We restarted it, and everything came back.” Without that visibility, the team would have had to restart every broker in turn, guessing at the cause.

A similar pattern played out when a team struggled to connect Kafka to Informatica. Kpow showed instantly that the application wasn’t part of the expected consumer group, something the CLI could only reveal as a single point-in-time snapshot rather than Kpow’s live, historical view. Dave estimates Kpow cuts troubleshooting time by about half: “It’s easily a 50 percent time saving for us.”

Taken together, the visibility and control Kpow gives Claritev’s middleware team has let the company retire a $150,000-a-year support contract with confidence, keep a consistent operational view of Kafka as it migrates between infrastructure providers, and cut troubleshooting time in half, freeing developers to spend less time diagnosing and more time shipping.