Self-service Kafka access for troubleshooting – without giving up control

Empower teams to safely discover and troubleshoot streaming applications, protected by fine-grained access controls and data masking.

Kafka is a black box for engineering teams

An issue can start anywhere: a breaking schema change, config drift, an incorrect quota limit. But developers rarely have self-service Kafka access to troubleshoot them. Every request routes through the platform team, and every minute of that handoff is downtime.

Why observability tools aren’t enough
Observability tools offer infrastructure logs and metrics. But they don’t show you a bad schema, malformed payload, a topic misconfigured for the load it’s carrying. Troubleshooting is easier when you can see entire records flowing through a system.

 

The cost lands on platform teams too
Every data access request turns into tickets: retrieve a schema, check a consumer offset, confirm a config. As Kafka adoption grows, platform teams spend more time being a lookup service, and less time scaling and governing the platform.

 

Locking down Kafka doesn’t remove the demand for access - just redirects it
A locked down Kafka is a secure Kafka, and an unusable resource for app teams. So the need for access doesn’t disappear, but goes underground, into in-house scripts and community UI tools. Neither is built for enterprise self-service.

Kafka is a blackbox for engineering teams
Kafka is a blackbox for engineering teams

Three issues that developers could have caught

Catch it in Staging

Schema changes breaking an app

A schema in staging has drifted from production, leading to an incident after the release of the application.

 

With Lenses: An engineer with the correct permissions can compare the staging and production schemas for the application side-by-side in Lenses Studio.

Schema changes in Kafka
Schema changes in Kafka
Query the Live Stream

Poison pill in a topic

A new rogue producer starts publishing events with null values to a Kafka partition causing lag on a specific consumer instance.

 

With Lenses: An engineer gets alerted to high consumer lag by Lenses. They use the SQL Snapshot engine to identify the last committed offset. Followed by the ISNULL() function to identify the partitions holding null values for a field.

Posion pill
Posion pill
Trace the Topology

Quota causing data processing delays

An application team complains of latency in their data processing, yet the consumer seems perfectly healthy.

 

With Lenses: Using the Lenses Topology an engineer can see the upstream producer applications for the consumer. The number of messages generated by one of the producers seems to be throttled by a Quota that has been incorrectly set.

Processing delays
Processing delays

Safely open up streaming data to developers

Give engineers governed self-service access to discover and troubleshoot issues across every Kafka cluster.

Issues resolved

Issues Resolved

Developers can catch bugs, discrepancies and changes before they become incidents.

Fewer war rooms

Fewer war rooms

Issues get triaged and resolved by the team that owns the app, not escalated.

Faster MTTR

Faster MTTR

The engineer, team or agent investigating can access the data and context without filing a ticket.

Accelerated development

Accelerated Kafka development

Fewer issues means engineers can stay focused on modernizing systems.

Platform team time

Platform Engineering team time

Fewer tickets and more capacity to support teams without adding headcount.

Kafka Troubleshooting features

Discover instantly

Multi-Kafka global catalog

A single secure screen showing topics, data flows, and schemas across Confluent, AWS MSK, Aiven, and open source Kafka clusters.

See the pipelines

Topology View

Discover your streaming applications. Isolate producer-to-consumer data dependencies and visual lag offsets in real time.

Explore data

SQL Studio

Query Kafka directly with SQL - no copying data to a laptop or a separate store just to look at it.

Catch issues early

Real-time metrics and alerts

Alerts for data-flow latency, producer rate spikes, and critical failures.

Delegate with control

Governance and self-service

Apply fine-grained privileges and data masking strictly based on your organization’s roles and permission-levels.

Scale with agentic operations

With an MCP Server

Safely connect AI agents to your Kafka environments, protected with short-lived and scoped OAuth tokens.

What makes Lenses different?

Enterprise governance-first architecture

What makes Lenses Different Enterprise governance

Powerful permission, data masking and auditing model build for enterprise environments

Vendor agnostic

What makes Lenses different Vendor agnostic

Lenses works on top of any Kafka flavor including MSK,  Confluent, Aiven, and Apache Kafka setups.

No data movement

No data movement

Query events with SQL syntax with data in its place; nothing is materialized in a separate store.

Scales to an enterprise footprint

What makes Lenses different Scales to an enterprise footprint

Designed to scale to hundreds of clusters across different networks and distributions.

Customer Stories

Fix problems before customers know they exist

"With how quickly our engineering teams build streaming apps, our systems proactively dispatch engineers and fix problems before customers even know there's an issue," Adam Astle, Head of Software Development at CityFibre.

Adam Astle, Head of Software at CityFibre

See an incident resolved - backwards

Join us for a live demonstration tracing a P1 payload issue from the production alarm all the way to its root origin in development - and solve it without a war room.

Tun Shwe and Jeremy Frenay
Tun Shwe and Jeremy Frenay
  • Lenses tracks and monitors all Consumer Groups connected to connected Kafka clusters to calculate offset-based and time-based lag into a unified monitoring view. Drilldowns are made available to instantly view the events for committed offsets using the Lenses SQL Studio. Alerts can be configured to be alerted when lag is detected.

  • A poison pill is a malformed message that stalls a consumer group instance. With Lenses SQL snapshot queries, developers can explore topics with SQL functions and conditional clauses to pinpoint the exact bad record, inspect its payload, and choose to skip or replay it - no blind consumer restarts or manual offset scrolling.

  • Yes. Lenses gives developers scoped, self-service access to only the topics, schemas and consumer groups they own - governed by RBAC, data policies and a full audit trail - so the platform team never gives up control.

  • Most time investigating incidents is lost waiting on the platform team who have enough visibility into Kafka, usually through tickets, escalations and war rooms. When app teams can investigate their own resources directly, incidents that took hours or days resolve in minutes - CityFibre cut service restoration from 3 days to 3 hours.

  • No. It complements them. Infrastructure observability tools tell you something at the broker or network level. Lenses tells you what’s wrong in at the “data layer” itself: the schema, the payload, the topic config and more, this sort of information simply isn’t available in Observability tools.

  • Yes, it’s common for customers to connect their AI Agent to Lenses via MCP and authorized via the OAuth 2.1 protocol. This allows a human to delegate access to Kafka to troubleshoot on their behalf with short-lived and scope credentials, backed by Lenses’ IAM model.