Services

Resources

Company

Our Work

Blog

SRE

Back to Tags

Case Studies

Cover image for the case study Modernising Observability for a Billion Dollar Revenue Gaming Platform

Read Case Study

Modernising Observability for a Billion Dollar Revenue Gaming Platform
How One2N modernised observability for a major gaming platform: 10x faster incident detection and tracing across 200+ services with OpenTelemetry.

Blogs

Illustration of a laptop and browser windows on a deep crimson background crossed with red circuit-like lines

Read Blog

How to fix the eMudhra CA compromise on macOS
Portrait of Saurabh Hirani, Principal SRE at One2N

Saurabh Hirani

Principal SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

How to fix the eMudhra CA compromise on a Mac: clean the CA bundles curl reads, update Node, Python, Go and Java, and prove each fix with a probe.

Illustration on a navy background of a terminal running exec 3<>/dev/tcp/host/443, cabled into a network socket, next to a dashed folder labelled /dev/tcp that does not exist

Read Blog

How to test TCP connectivity without curl using bash /dev/tcp
Portrait of Saurabh Hirani, Principal SRE at One2N

Saurabh Hirani

Principal SRE @One2N

No curl, wget or nc on a locked-down box? Use bash /dev/tcp to check if a port is open and read HTTP endpoints, and learn why it fails in zsh.

Cover illustration for Why your EKS Auto Mode NodeClass keeps reverting to default

Read Blog

Why your EKS Auto Mode NodeClass keeps reverting to default
Portrait of Ananya Dasgupta, SRE at One2N

Ananya Dasgupta

SRE @One2N

EKS Auto Mode reverted our default NodeClass after Argo CD synced secondary CIDR pod subnets. Why Auto Mode owns that object and what to change instead.

Cover illustration for How to evaluate AI SRE agents for production?

Read Blog

How to evaluate AI SRE agents for production?
Portrait of Hemant Kumar, Staff SRE at One2N

Hemant Kumar

Staff SRE @One2N

How we evaluate AI SRE agents before they touch a production incident, applied to an OpenSRE v0.1 PoC and scored on how fast it helps isolate a fault.

Cover illustration for Conway's Law as a lever to fix legacy architectures

Read Blog

Conway's Law as a lever to fix legacy architectures
Portrait of Chinmay Naik, CEO at One2N

Chinmay Naik

CEO @One2N

Conway's law says architecture mirrors team communication. A shared policy database showed it; the Reverse Conway Maneuver helps refactor legacy systems.

Cover illustration for Hardening Kubernetes security with Istio ambient mode on flat networks

Read Blog

Hardening Kubernetes security with Istio ambient mode on flat networks
Portrait of Harshwardhan Mehrotra, SRE at One2N

Harshwardhan Mehrotra

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Istio Ambient mode on a flat EKS Hybrid network: how we rebuilt workload isolation with mTLS, SPIFFE identity, egress control and NetworkPolicies.

Cover illustration for Git was not built for multiple AI agents. Worktrees fix that.

Read Blog

Git was not built for multiple AI agents. Worktrees fix that.
Portrait of Srivatsa RV, SRE at One2N

Srivatsa RV

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Git worktrees for agentic development: give each AI coding agent its own folder and branch on one repo, so parallel chats stop fighting over one checkout.

Cover illustration for How one parameter stalled Postgres replication with Debezium for 3 weeks

Read Blog

How one parameter stalled Postgres replication with Debezium for 3 weeks
Portrait of Harsh Mishra, SRE

Harsh Mishra

SRE

A story about WAL segments, missing events, and the one line of SQL that should have been there from the start.

Cover illustration for My new laptop broke Slack. That's how I ended up on Bluefin.

Read Blog

My new laptop broke Slack. That's how I ended up on Bluefin.
Portrait of Harsh Mishra, SRE

Harsh Mishra

SRE

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

A kernel patch broke Slack on my Dell XPS. Here's how that led me to Bluefin, immutable Linux, bootc, and OCI images and why it solved my problem for good.

Cover illustration for Spring Kafka batching: how we shot ourselves in the foot

Read Blog

Spring Kafka batching: how we shot ourselves in the foot
Portrait of Mahesh Aravind, SDE at One2N

Mahesh Aravind

SDE @One2N

Spring Kafka batch listeners cost us non-blocking retries for a small throughput gain. How listener batching differs from producer batching, and the fix.

Cover illustration for Why your Architecture should start with Questions, not boxes

Read Blog

Why your Architecture should start with Questions, not boxes
Portrait of Chinmay Naik, CEO at One2N

Chinmay Naik

CEO @One2N

Architecture should start with questions about volume, SLAs, budget, team skills and growth before anyone draws Kafka and microservices boxes on a diagram.

Cover illustration for GitOps for Kafka: how we governed 78 clusters without breaking production

Read Blog

GitOps for Kafka: how we governed 78 clusters without breaking production
Portrait of Harshwardhan Mehrotra, SRE at One2N

Harshwardhan Mehrotra

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

GitOps for Kafka with Jikkou: how a small platform team governed 78 clusters, 10,000+ topics and Kafka 0.8.2 to 4.x, replacing 3 to 5 day topic tickets.

Cover illustration for Prayogshala: the engineering laboratory at One2N

Read Blog

Prayogshala: the engineering laboratory at One2N
Portrait of Chinmay Naik, CEO at One2N

Chinmay Naik

CEO @One2N

Prayogshala is One2N's internal engineering lab, where engineers build real-world POCs with Principal SRE mentorship and learn to ask why before how.

Cover illustration for The gotchas of OTel Collector processors in Kubernetes

Read Blog

The gotchas of OTel Collector processors in Kubernetes
Portrait of Sanket Rajgiri, SRE at One2N

Sanket Rajgiri

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

OTel Collector processor gotchas in Kubernetes: batching to stop 429s, crashes in spikes, noise filters, k8s enrichment, sampling and two-tier collectors.

Cover illustration for How queueing theory makes systems reliable

Read Blog

How queueing theory makes systems reliable
Portrait of Srivatsa RV, SRE at One2N

Srivatsa RV

SRE @One2N

Queueing theory for SREs: the M/M/1 model shows why 70% utilisation feels calm and 90% collapses under spikes, and why headroom is insurance, not waste.

Cover illustration for How to read SRE graphs without lying to yourself

Read Blog

How to read SRE graphs without lying to yourself
Portrait of One2N, Team

One2N

Team

How to read SRE graphs honestly: averages that hide pain, sampling windows that erase spikes, log scales and heatmaps, plus a checklist for incidents.

Cover illustration for Error budget calculation: downtime minutes for every SLO

Read Blog

Error budget calculation: downtime minutes for every SLO
Portrait of Srivatsa RV, SRE at One2N

Srivatsa RV

SRE @One2N

Error budget calculation from your SLO: 99.9% over 30 days is 43 minutes of downtime, 99.95% is 21m 54s. Use burn rate to decide when to ship or pause.

Cover illustration for Percentiles in SRE: Why averages lie about latency

Read Blog

Percentiles in SRE: Why averages lie about latency
Portrait of Srivatsa RV, SRE at One2N

Srivatsa RV

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Percentiles in SRE: 50 slow requests in 1,000 can leave average latency near 200 ms. How p50, p95 and p99 expose tail latency and which ones to monitor.

Cover illustration for Latency vs throughput: why high utilisation hurts reliability

Read Blog

Latency vs throughput: why high utilisation hurts reliability
Portrait of Srivatsa RV, SRE at One2N

Srivatsa RV

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Latency vs throughput explained for engineers: how the two interact with utilisation, and why running at 90 percent capacity quietly wrecks your SLOs.

Cover illustration for Deploying a scalable NATS cluster part 2: hands-on demo

Read Blog

Deploying a scalable NATS cluster part 2: hands-on demo
Portrait of Barun Debnath, SRE

Barun Debnath

SRE

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Deploy a four-node NATS cluster with Docker, JetStream stream replication, dead letter queues and Prometheus and Grafana dashboards, then load test it.

Cover illustration for SRE math every engineer should know: a practical guide

Read Blog

SRE math every engineer should know: a practical guide
Portrait of Srivatsa RV, SRE at One2N

Srivatsa RV

SRE @One2N

SRE math for on-call engineers: percentiles that averages hide, Little's Law, error budgets as downtime minutes, queueing theory and honest graph reading.

Cover illustration for From aircraft to cloud: my experience with the SRE Bootcamp

Read Blog

From aircraft to cloud: my experience with the SRE Bootcamp
Portrait of Pankaj Goriya, SRE at One2N

Pankaj Goriya

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

An aircraft maintenance engineer's review of One2N's SRE Bootcamp: how the hands-on exercises built DevOps and SRE fundamentals, and advice for newcomers.

Cover illustration for DORA metrics: useful or fluff

Read Blog

DORA metrics: useful or fluff
Portrait of Jaideep Khandelwal, CTO at One2N

Jaideep Khandelwal

CTO @One2N

DORA metrics without context invite Goodhart's Law gaming. One team shipped 2,500 deploys a month with long lead times. How to tie DORA to business goals.

Cover illustration for A pragmatic guide to get started with OpenTelemetry

Read Blog

A pragmatic guide to get started with OpenTelemetry
Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

A pragmatic OpenTelemetry guide for SREs: monitoring vs observability, logs, metrics and traces, OTel instrumentation, and strategies to migrate backends.

Cover illustration for DARE to question your alerts?

Read Blog

DARE to question your alerts?
Portrait of Saurabh Hirani, Principal SRE at One2N

Saurabh Hirani

Principal SRE @One2N

Alert fatigue grows when alerts are set and forgotten. DARE adds alerts per unit time to RED metrics, with fixes like muting CPU alerts during backups.

Cover illustration for Deploying a scalable NATS cluster, part 1: core architecture

Read Blog

Deploying a scalable NATS cluster, part 1: core architecture
Portrait of Barun Debnath, SRE

Barun Debnath

SRE

NATS cluster architecture, part 1: pub-sub and queue groups, JetStream persistence, cluster, super cluster and leaf node topologies, and what to plan for.

Cover illustration for Alerting with GitOps: automating Elasticsearch alerts

Read Blog

Alerting with GitOps: automating Elasticsearch alerts
Portrait of Srivatsa RV, SRE at One2N

Srivatsa RV

SRE @One2N

Automating Elasticsearch alerts with GitOps: we replaced manual 30-minute dashboard checks on the ELK stack with alert definitions a NOC team can act on.

Cover illustration for GitHub runners fundamentals and self-hosted runner setup

Read Blog

GitHub runners fundamentals and self-hosted runner setup
Portrait of Rajesh Jangid, SRE

Rajesh Jangid

SRE

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

What GitHub runners are, how GitHub-hosted and self-hosted runners differ, and how to set up a self-hosted GitHub Actions runner.

Cover illustration for EKS upgrade best practices: how to upgrade EKS clusters safely

Read Blog

EKS upgrade best practices: how to upgrade EKS clusters safely
Portrait of Aashi Rathore, SRE at One2N

Aashi Rathore

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

EKS upgrade checklist from our v1.23 to v1.30 move: control plane checks, VPC CNI, CoreDNS and kube-proxy add-ons, Velero backups and node group upgrades.

Cover illustration for AWS SSM port forwarding to private EC2 without SSH or bastions

Read Blog

AWS SSM port forwarding to private EC2 without SSH or bastions
Portrait of Harsh Mishra, SRE

Harsh Mishra

SRE

AWS SSM port forwarding to EC2 in private subnets with Session Manager, with no open SSH port or bastion. Covers the IAM role, SSM Agent and endpoints.

Cover illustration for Gotchas to avoid when using AWS spot instances in production

Read Blog

Gotchas to avoid when using AWS spot instances in production
Portrait of Aashi Rathore, SRE at One2N

Aashi Rathore

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

AWS spot instances in production on EKS: two-minute termination notices, ignored PodDisruptionBudgets, slow replacement capacity and a real outage we had.

Cover illustration for Accessing private EC2 Instances without keys or passwords

Read Blog

Accessing private EC2 Instances without keys or passwords
Portrait of Adarsh Raj, SRE

Adarsh Raj

SRE

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Access private EC2 instances without SSH keys or passwords using AWS Systems Manager Session Manager: no public IPs, no inbound rules, and no shared keys.

Cover illustration for Using UTM and Vagrant for dev environments on Apple Silicon

Read Blog

Using UTM and Vagrant for dev environments on Apple Silicon
Portrait of Srivatsa RV, SRE at One2N

Srivatsa RV

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

VirtualBox no longer supports M-series Macs. How to run Vagrant with UTM to spin up reproducible local dev environments on Apple Silicon, tested on an M3.

Cover illustration for Karpenter vs Cluster Autoscaler for Kubernetes node scaling

Read Blog

Karpenter vs Cluster Autoscaler for Kubernetes node scaling
Portrait of Rahul H Bhatia, SRE

Rahul H Bhatia

SRE

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Karpenter vs Cluster Autoscaler: how each scales Kubernetes nodes, from predefined node groups and priority expanders to provisioning and cost trade-offs.

Cover illustration for Exploring RAG: enhancing LLMs with external knowledge

Read Blog

Exploring RAG: enhancing LLMs with external knowledge
Portrait of Harsh Mishra, SRE

Harsh Mishra

SRE

Portrait of Saurabh Hirani, Principal SRE at One2N

Saurabh Hirani

Principal SRE @One2N

Retrieval augmented generation (RAG) for engineers: LLM basics, where standalone models fall short, and the parts of a RAG app such as embeddings.

Cover illustration for Setting up CloudWatch alerts for EKS clusters: insights and solutions

Read Blog

Setting up CloudWatch alerts for EKS clusters: insights and solutions
Portrait of Mihir Bhagwat, SRE at One2N

Mihir Bhagwat

SRE @One2N

Portrait of Saurabh Hirani, Principal SRE at One2N

Saurabh Hirani

Principal SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

CloudWatch alerts for EKS node readiness with Container Insights: per-node alarms for static nodes, then Metric Insights and MIN(MIN) for autoscaled ones.

Cover illustration for How we used Fly.io to improve a Pipecat-AI chatbot

Read Blog

How we used Fly.io to improve a Pipecat-AI chatbot
Portrait of Ajinkya Wadekar, SRE at One2N

Ajinkya Wadekar

SRE @One2N

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

A Pipecat-AI chatbot on ECS hit limits at 10 to 15 chat rooms per task, and 6GB images slowed scale-up. How Fly.io machines spun from containers fixed it.

Cover illustration for Understanding MLOps from a software engineers' perspective

Read Blog

Understanding MLOps from a software engineers' perspective
Portrait of Harsh Mishra, SRE

Harsh Mishra

SRE

Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Portrait of Chinmay Naik, CEO at One2N

Chinmay Naik

CEO @One2N

Portrait of Jaideep Khandelwal, CTO at One2N

Jaideep Khandelwal

CTO @One2N

MLOps for software engineers is DevOps plus data engineering and ML. The MLOps lifecycle, knowledge silos, automation, versioning and experiment tracking.

Cover illustration for Database schema migrations: a practical guide to managing schema changes

Read Blog

Database schema migrations: a practical guide to managing schema changes
Portrait of Spandan Ghosh, Content at One2N

Spandan Ghosh

Content @One2N

Database schema migrations without manual steps: versioned migrations, running them in every deploy, and tools like Flyway, Liquibase and ORM migrations.

Cover illustration for Achieving CI/CD with previews for your static website hosted on S3

Read Blog

Achieving CI/CD with previews for your static website hosted on S3
Portrait of Drasti Kaurav, SRE at One2N

Drasti Kaurav

SRE @One2N

CI/CD for a static website on S3 with preview environments: automated builds and deploys without developer AWS keys, plus CloudFront cache invalidation.

Cover illustration for Kubernetes HPA on external metrics with Prometheus Adapter

Read Blog

Kubernetes HPA on external metrics with Prometheus Adapter
Portrait of Akshay Srivastava, SRE at One2N

Akshay Srivastava

SRE @One2N

Kubernetes HPA on external metrics: scaling pods on Redis queues from a server outside the cluster with Prometheus Adapter, plus CPU and memory metrics.

Talks

Thumbnail for the talk The implications of AI in the SRE world | One2N Bits Ep.5

Watch Talk

The implications of AI in the SRE world | One2N Bits Ep.5
Portrait of Jaideep Khandelwal, CTO at One2N

Jaideep Khandelwal

CTO @One2N

Portrait of Saurabh Hirani, Principal SRE at One2N

Saurabh Hirani

Principal SRE @One2N

Saurabh Hirani and Jaideep Khandelwal break down what it takes for AI to be genuinely useful in SRE, beyond incident bots and code completion.

Thumbnail for the talk AI Enabled SRE Practices | One2N Cursor Meetup Pune

Watch Talk

AI Enabled SRE Practices | One2N Cursor Meetup Pune
Portrait of Saurabh Hirani, Principal SRE at One2N

Saurabh Hirani

Principal SRE @One2N

Saurabh Hirani's lightning talk from the first Cursor Meetup in Pune, hosted at One2N, on what AI-enabled SRE practices look like in daily work.

Thumbnail for the talk A beginners guide to understand Kubernetes using Restaurant Analogy | DevOpsDays Kerala 2024

Watch Talk

A beginners guide to understand Kubernetes using Restaurant Analogy | DevOpsDays Kerala 2024
Portrait of Chinmay Naik, CEO at One2N

Chinmay Naik

CEO @One2N

Kubernetes is 10+ years old but still hard to learn. This talk uses a restaurant analogy to make K8s approachable and drive better adoption.

‹ SpringKafka

SRE-Math ›