Blog
Blog
What we are seeing in the market and in production, written by the people who run these platforms.
- Blog4 min
Cloudera on-premises support dates: 7.1.7 is already out of support, and 7.3.1 ends in December
Cloudera on-premises end-of-support dates: 7.1.7 ended August 2025, 7.3.1 ends December 2026, 7.1.9 runs to October 2028, 7.3.2 to March 2032.
Read - Blog5 min
Redis vs Valkey: the license history in three dates, and what it means for your team
Redis 7.4 moved from BSD to RSALv2/SSPLv1 (20 March 2024). Valkey forked from 7.2.4 under BSD on 28 March 2024; Redis 8 added AGPLv3 on 1 May 2025.
Read - Blog5 min
Mule 4.6 LTS is out of standard support: the Java 17 move, and why 4.10 is a manual upgrade
Mule 4.6 LTS standard support ended August 2026; extended support ends February 2027. Mule 4.9+ runs only on Java 17, and 4.10 is a manual upgrade.
Read - Blog5 min
Confluent vs Apache Kafka in 2026: three licences, one core, and a new owner
Apache Kafka is Apache 2.0; Confluent adds a Community License and an Enterprise License. IBM completed its Confluent acquisition on 17 March 2026.
Read - Blog5 min
Airflow 2 reached end of life in April 2026: what the move to Airflow 3 actually breaks
Airflow 2 reached end of life on 22 April 2026 and gets no fixes. Airflow 3 needs 2.7 or later and removes SubDAGs, SLAs and direct database access.
Read - Blog5 min
RabbitMQ Khepri is now the default metadata store: what changes when a majority of nodes is gone
Khepri replaced Mnesia as RabbitMQ's default metadata store in 4.2.0. It is Raft-based, so metadata changes need a node quorum. How to enable it and plan.
Read - Blog5 min
ServiceNow Brazil for Discovery teams: the relationship-type change that breaks custom scripts quietly
ServiceNow Brazil, early availability: Discovery teams get a findOrCreateRelationshipType change, PowerShell 7 MID support and credential-less scans.
Read - Blog5 min
ServiceNow Brazil CMDB changes: Dynamic IRE by default, and the role change that hides incidents from service owners
In ServiceNow Brazil, Dynamic IRE is the default on new instances, Service Graph Workspace merges into CMDB Workspace, and itil leaves app_service_admin.
Read - Blog4 min
Pulsar subscription types: which one keeps your message ordering, and which one quietly drops it
Pulsar's four subscription types trade ordering for parallelism differently. What each does, which preserves order, and why Shared surprises teams.
Read - Blog4 min
Core NATS vs JetStream: the delivery guarantee you are actually getting
Core NATS is at-most-once and stores nothing; JetStream adds streams and at-least-once delivery. What each guarantees in production and when to use which.
Read - Blog4 min
Solace direct vs guaranteed messaging: when 'reliable' is not 'guaranteed'
Solace Direct messaging is reliable but not guaranteed; Guaranteed messaging spools to persistent storage. Where messages can drop, and how to choose.
Read - Blog5 min
From Kubernetes 1.35 the kubelet will not start on cgroup v1: what to check before you upgrade
From Kubernetes 1.35, failCgroupV1 defaults to true and the kubelet won't start on cgroup v1 nodes. What breaks, the temporary override, what v2 changes.
Read - Blog5 min
The CMDB says 94 percent complete and the service dashboard is still wrong. What we ask ITOM candidates
Our ServiceNow ITOM screen: why 94% CMDB completeness can sit alongside a wrong executive dashboard, and how a senior candidate explains it.
Read - Blog6 min
SQS visibility timeouts, dead-letter queues and duplicates: what a managed queue does not do for you
Amazon SQS removes broker operations, not delivery semantics. Visibility timeouts, at-least-once duplicates and the dead-letter retention trap, per AWS.
Read - Blog4 min
WSO2 API Manager support dates: 3.2.x has passed its listed end-of-life date, and 4.2.0 is deprecated
WSO2 API Manager 3.2.0 and 3.2.1 reached listed support EOL in September 2026, 4.2.0 is deprecated until March 2027, and 4.3-4.7 run to 2027-2029.
Read - Blog5 min
Redis persistence, RDB vs AOF: what each one loses when the server stops
Redis can snapshot with RDB, log every write with AOF, or both. What each loses when the server stops unexpectedly, and why the Redis docs recommend both.
Read - Blog5 min
Thirty system APIs and a slow experience API. What we ask MuleSoft candidates
Our MuleSoft screen: thirty system APIs, a slow experience API and a team that wants to add caching. What a senior answer asks first.
Read - Blog4 min
Why we publish our screening questions
Most staffing firms keep their screen private; we publish ours on every role page. Why a rubric you can read is evidence, why preparation doesn't break it.
Read - Blog5 min
IBM MQ channel authentication records: what they control, and where they do not apply
IBM MQ channel authentication records block clients by IP or user ID and map asserted identities. What they control, how to enable them, where they skip.
Read - Blog5 min
Gateway API v1.5 and v1.6: what moved to Standard in 2026, and the upgrade trap with Experimental routes
Gateway API v1.5 promoted ListenerSet, TLSRoute and four more features to Standard; v1.6 did TCPRoute and UDPRoute. Why Experimental TLSRoutes can break.
Read - Blog5 min
An event-driven rebuild for a team that has never run a broker. What we ask enterprise architects
Our enterprise architecture screen: an event-driven rebuild, a team with no broker experience, a regulator demanding replay. Why not rebuilding may win.
Read - Blog5 min
Pod disruption budgets: the Kubernetes setting that decides whether a node upgrade breaks your quorum
A PodDisruptionBudget limits how many replicas are down at once during node drains. What it protects, what it can't, and how to set one for quorum systems.
Read - Blog5 min
Upgrading CDP 7.1.7 to 7.1.9 when the OS goes end of life first. What we ask data engineering candidates
Our data engineering screen: sequencing a CDP Private Cloud Base 7.1.7 to 7.1.9 upgrade when the operating system reaches end of life first.
Read - Blog4 min
Discovery runs every night and the service map is still stale. Why ServiceNow needs both schedules
Horizontal Discovery finds the CIs; Service Mapping's own schedules keep service maps current. Why both must run, and what it means for your CMDB.
Read - Blog4 min
Kafka 4.3: the deprecations to plan for before Kafka 5.0, starting with the classic rebalance protocol
Apache Kafka 4.3 shipped 25 KIPs and began retiring the classic consumer rebalance protocol. What is marked for removal in 5.0 and what to change now.
Read - Blog5 min
70 percent of the error budget is gone with five weeks left. What we ask SRE candidates
Our SRE screen: what a senior engineer does when the error budget is 70% spent and product wants to ship to checkout, and where a mid-level answer stops.
Read - Blog5 min
Kafka tiered storage is production-ready. What it changes, and what it doesn't
Kafka tiered storage is production-ready since 3.9. What offloading to object storage changes about retention and disk sizing, and what it leaves alone.
Read - Blog5 min
What nearshore overlap actually buys you
Nearshore is sold on rate, but same-day working-hour overlap is what makes it work. Cases where overlap doesn't matter are more common than vendors say.
Read - Blog5 min
Kafka 4.0 removed ZooKeeper. What the upgrade actually involves
Kafka 4.0 runs only in KRaft mode: ZooKeeper clusters must migrate first, brokers need Java 17 and old message formats are gone. What the upgrade involves.
Read - Blog6 min
Ingress NGINX is retired: what the March 2026 deadline means, and why the migration is not a YAML rename
Kubernetes retired Ingress NGINX in March 2026: no more releases or security fixes. How to check you're affected and why no alternative is drop-in.
Read - Blog7 min
A RabbitMQ quorum queue loses its leader. What we ask, and why most answers stop early
Our published DevOps screen: what a senior answer contains when a RabbitMQ quorum queue loses its leader mid-upgrade, and where a mid-level answer stops.
Read - Blog5 min
Kafka share groups are production-ready in 4.2: does Kafka replace your queue now?
Queues for Kafka (KIP-932) went from early access in 4.0 to preview in 4.1 to production-ready in 4.2. What share groups change, and don't replace.
Read - Blog5 min
RabbitMQ 4.0 removed classic mirrored queues. What the migration actually involves
RabbitMQ 4.0 removed classic mirrored queues. What migrating to quorum queues involves, why blue-green is usual, and the one case they don't cover.
Read - Blog5 min
In-place Pod resize is stable in Kubernetes 1.35: what it changes, and where it still restarts you
In-place Pod resize is GA in Kubernetes 1.35: CPU and memory become mutable in a running Pod. What changed since beta and the limits that remain.
Read - Blog4 min
ActiveMQ and Artemis are now separate Apache projects. What it means if you run either
Apache split ActiveMQ and Artemis into two top-level projects in November 2025. What it changes for teams running either, and for ActiveMQ job specs.
Read
Tell us the role. A practitioner scopes it, not a recruiter.
Send the requirement — or the job description you are not happy with — and an engineer who runs this technology will tell you what is missing before we source anyone.