[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

# ![@brankopetric00 Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::1157414015666184192.png) @brankopetric00 Branko

A recent AWS outage was caused by a DNS problem that hit the AWS DynamoDB endpoint, making it unavailable for services to access. The issue was fixed in about X hours, but then another issue arose with EC2, the system that creates virtual servers. This incident highlights the importance of having a robust disaster recovery plan and being prepared for such outages.

### Engagements: XXXXXX [#](/creator/twitter::1157414015666184192/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1157414015666184192/c:line/m:interactions.svg)

- X Week XXXXXXX +71%
- X Month XXXXXXXXX -XX%
- X Months XXXXXXXXXX +1,753%
- X Year XXXXXXXXXX +2,468%

### Mentions: XX [#](/creator/twitter::1157414015666184192/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1157414015666184192/c:line/m:posts_active.svg)

- X Week XX -XX%
- X Month XXX -XX%
- X Months XXX +371%
- X Year XXX +1,947%

### Followers: XXXXXX [#](/creator/twitter::1157414015666184192/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1157414015666184192/c:line/m:followers.svg)

- X Week XXXXXX +5.20%
- X Month XXXXXX +13%
- X Months XXXXXX +254%
- X Year XXXXXX +1,185%

### CreatorRank: XXXXXXX [#](/creator/twitter::1157414015666184192/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::1157414015666184192/c:line/m:influencer_rank.svg)

### Social Influence

**Social category influence**
[technology brands](/list/technology-brands)  [finance](/list/finance)  [stocks](/list/stocks)  [social networks](/list/social-networks)  [countries](/list/countries) 

**Social topic influence**
[environment](/topic/environment) #481, [$1200month](/topic/$1200month) #3, [not for](/topic/not-for) #640, [money](/topic/money), [math](/topic/math) #1696, [$18kmonth](/topic/$18kmonth), [slack](/topic/slack), [fine](/topic/fine), [business](/topic/business), [tomcat](/topic/tomcat)

**Top accounts mentioned or mentioned by**
[@prmptvault](/creator/undefined) [@willvincent](/creator/undefined) [@0xraghuboi](/creator/undefined) [@duck4i](/creator/undefined) [@cordasfilip](/creator/undefined) [@kelvinjay1](/creator/undefined) [@realbirdman85](/creator/undefined) [@echamudi](/creator/undefined) [@cpufxrndmv](/creator/undefined) [@abez95](/creator/undefined) [@bdesmet_](/creator/undefined) [@garyfung](/creator/undefined) [@thegrimscalper](/creator/undefined) [@thecolemanthe](/creator/undefined) [@g4mi_](/creator/undefined) [@hanghuang_](/creator/undefined) [@philkellr](/creator/undefined) [@xsuraj08](/creator/undefined) [@digitalcolmer](/creator/undefined) [@sudoferraz](/creator/undefined)

**Top assets mentioned**
[New Relic, Inc. (NEWR)](/topic/$newr)
### Top Social Posts
Top posts by engagements in the last XX hours

"Your Dockerfile: - FROM ubuntu:latest - Installs XX packages - Runs as root - 2.3GB image size - Rebuild takes XX minutes - Security scan shows XX vulnerabilities Your app: - A Python script - X dependencies - Could run on Alpine in a 50MB image You're not using containers. You're building VMs with Dockerfiles. This is what happens when people learn Docker from random Medium articles"  
[X Link](https://x.com/brankopetric00/status/1996581511236030874)  2025-12-04T14:05Z 14.7K followers, 357.9K engagements


"You split your monolith into XX microservices. Now you have: - XX deployment pipelines to maintain - XX different ways to handle logging - Distributed tracing you don't understand - Network calls that used to be function calls - A debugging nightmare when anything breaks Your "scalability problem" It was a single slow SQL query. You spent X months building a distributed system to avoid adding an index. This is what happens when architects read Medium articles instead of profiling code"  
[X Link](https://x.com/brankopetric00/status/1996702563568120199)  2025-12-04T22:06Z 14.7K followers, 44.6K engagements


"Job posting: "DevOps Engineer" Actual job: - Manually deploy WAR files to Tomcat - SSH into servers to check logs - Restart services when they crash - Update firewall rules via tickets - No infrastructure as code - No CI/CD - No cloud You're not hiring a DevOps Engineer. You're hiring a sysadmin who knows Docker exists. Just be honest about the role. Sysadmins are valuable. Stop pretending you're doing DevOps when you're not"  
[X Link](https://x.com/brankopetric00/status/1996748115257290831)  2025-12-05T01:07Z 14.7K followers, 44.5K engagements


"Implemented API rate limiting by IP address. 1000 requests per hour per IP. Perfect protection against abuse. Next week: Legitimate customer hit rate limit. Complained their service was broken. They were behind corporate NAT. XXX employees sharing one IP address. Hit limit in XX minutes. Changed to rate limit by API key instead. Problem solved. Week after: Bot attack. 50000 requests per hour. Different API keys. Free tier abuse. Changed to rate limit by both IP and API key. Problem solved. Week after: DDoS from 10000 different IPs using 10000 different free tier API keys. Rate limiting was"  
[X Link](https://x.com/brankopetric00/status/1982799551246266586)  2025-10-27T13:20Z 14.7K followers, 355.8K engagements


"We chose PostgreSQL over MongoDB for our analytics platform. The context: - 50GB of time-series data daily - Complex queries with joins across multiple dimensions - Team had more SQL experience than NoSQL MongoDB seemed obvious for scale but: - Query complexity made aggregation pipelines unwieldy - Horizontal scaling wasn't needed yet - PostgreSQL's JSON support gave us flexibility - TimescaleDB extension handled time-series perfectly XX months later: PostgreSQL handles 2TB with sub-second queries. Sometimes boring technology wins"  
[X Link](https://x.com/brankopetric00/status/1995624711716229421)  2025-12-01T22:43Z 14.7K followers, 204.4K engagements


"Shared SSH keys across the team means nobody knows who did what and ex-employees still have access. Implement certificate-based SSH with short-lived certs. Audit logs are useless when everyone uses the same key"  
[X Link](https://x.com/brankopetric00/status/1998116402213925306)  2025-12-08T19:44Z 14.7K followers, 10.4K engagements


"Secrets in environment variables is security theater with extra steps. Use AWS Secrets Manager or Parameter Store. Environment variables are logged visible in process lists and stored in plain text. This isn't secure it's just slightly less obvious"  
[X Link](https://x.com/brankopetric00/status/1998623219411005471)  2025-12-10T05:18Z 14.7K followers, 4314 engagements


"CDN (Content Delivery Network) in simple terms: Your server is in Virginia. User is in Singapore. Without CDN: Request travels 9000 miles. Takes 300ms. With CDN: Request goes to nearest edge server in Singapore. Already has your content cached. Takes 50ms. CDN is a globally distributed cache. Best for: images videos static files APIs with cacheable responses. Not for: dynamic content that changes per user. It's about putting data closer to users"  
[X Link](https://x.com/brankopetric00/status/1987769697845035321)  2025-11-10T06:30Z 14.7K followers, 49.3K engagements


"Reduced Docker image size from 2.1GB to 180MB. Deployments 8x faster. The original Dockerfile: - Started with ubuntu:latest - Installed everything via apt - Included dev dependencies - Copied entire project directory - Left build artifacts - No layer optimization The problems: - Pull time: 6-8 minutes - Registry storage costs high - Deployment took forever - Security scan found XX vulnerabilities - Most from unnecessary packages What we optimized: X. Base image - ubuntu:latest (2.1GB) alpine:latest (5MB) X. Dependencies - Removed dev dependencies - Multi-stage build - Only production packages"  
[X Link](https://x.com/brankopetric00/status/1989294089150361654)  2025-11-14T11:27Z 14.7K followers, 256.4K engagements


"Security tools were costing more than our infrastructure. Here's how we optimized. The bill breakdown: - Snyk: $3200/month - Datadog Security: $4800/month - AWS GuardDuty: $1200/month - AWS Security Hub: $600/month - Aqua Security: $2400/month - GitHub Advanced Security: $2100/month - Total: $14300/month - Actual infrastructure: $11000/month The problem: - Paying for overlapping features - Using XX% of each tool's capabilities - No one owned the security stack - Tools bought by different teams - No coordination The audit: - Listed every security tool - Identified feature overlap - Measured"  
[X Link](https://x.com/brankopetric00/status/1989621034715103587)  2025-11-15T09:06Z 14.7K followers, 29K engagements


"Application handled 10K concurrent websocket connections perfectly. Scaled to 50K connections and Redis started throwing OOM errors. The investigation: - Redis configured with 2GB memory - Each connection stored session data - Session data: 50KB average per user - Math: 50K users 50KB = 2.5GB - Redis couldn't fit all sessions in memory - Started evicting old sessions - Users randomly logged out The architecture problem: - Used Redis as cache with eviction policies - Treated it like persistent storage - Didn't account for memory per connection The solution: - Moved session storage to DynamoDB"  
[X Link](https://x.com/brankopetric00/status/1993776800602771606)  2025-11-26T20:20Z 14.7K followers, 39.3K engagements


"Cut our AWS bill from $28k to $9k per month with X simple changes. The audit revealed: - XX stopped EC2 instances still had EBS volumes attached ($4.2k/month) - NAT Gateway in every AZ but only using one ($1.8k/month waste) - RDS instances running 24/7 for dev/staging ($8k/month) What we did: - Lambda to cleanup orphaned EBS volumes - Consolidated to single NAT Gateway with route table updates - Scheduled dev/staging RDS shutdown (7pm-7am weekends) Bonus optimizations: - Switched to Graviton instances: XX% cheaper - Moved logs to S3 Glacier after XX days - Reserved instances for production"  
[X Link](https://x.com/brankopetric00/status/1994876798648099327)  2025-11-29T21:11Z 14.7K followers, 16.9K engagements


"Chose serverless for a batch processing job. Worst decision for our use case. The requirement: - Process XX million records daily - Each record: X seconds of processing - Run during off-peak hours (2am-6am) - Cost-sensitive project Why we chose Lambda: - Pay per execution - Auto-scaling - No infrastructure to manage - Seemed perfect The reality: - Lambda max timeout: XX minutes - Needed to batch records carefully - Cold starts killed performance - Concurrent execution costs added up The costs: - Lambda: $4200/month - DynamoDB for state tracking: $800/month - CloudWatch Logs: $300/month -"  
[X Link](https://x.com/brankopetric00/status/1994900203124523507)  2025-11-29T22:44Z 14.7K followers, 14.8K engagements


"Switched from Jenkins to GitHub Actions. Team productivity doubled. Our Jenkins setup: - Self-hosted on EC2 - XX plugins installed - Groovy-based pipelines - One person knew how it worked The problems: - Server maintenance took XX hours/month - Plugin conflicts broke builds randomly - Groovy syntax scared developers - No one wanted to touch pipeline code Why we migrated: - That one person quit - Jenkins became a black box - Developers avoided CI/CD changes - Innovation stalled The transition: - YAML-based workflows easier to understand - Native GitHub integration - No server maintenance -"  
[X Link](https://x.com/brankopetric00/status/1995153027032903695)  2025-11-30T15:28Z 14.7K followers, 31K engagements


"Built a Lambda function to automatically stop non-production instances. - Identified instances by environment tags - Scheduled shutdowns at X PM startups at X AM - Added weekend shutdown for development environments - Implemented Slack notifications for manual overrides Reduced non-prod costs by XX% with XX lines of Python. Sometimes the best optimization is just turning things off when nobody's using them"  
[X Link](https://x.com/brankopetric00/status/1996054560264036823)  2025-12-03T03:11Z 14.7K followers, 5587 engagements


"@PrateekJainDev This could be super intuitive if integrated with AI so you can ask questions"  
[X Link](https://x.com/brankopetric00/status/1996198255844274439)  2025-12-03T12:42Z 14.7K followers, 1959 engagements


"Our AWS bill had a $3400 charge labeled 'Data Transfer Out.' We barely had any traffic. Dug into it: - CloudWatch Logs was shipping to S3 - S3 bucket was in us-west-2 - Application was in us-east-1 - Cross-region transfer: $XXXX per GB - We were logging everything at DEBUG level - 170TB of logs transferred in one month Moved S3 bucket to same region. Bill dropped to $XX. Lesson: Data transfer costs hide in places you'd never think to look"  
[X Link](https://x.com/brankopetric00/status/1996207541575901196)  2025-12-03T13:19Z 14.7K followers, 45.2K engagements


"Your "Internal Developer Platform": - XX microservices - XX API endpoints developers don't use - Documentation from X months ago - X people who understand it - Slack channel full of "how do I." Developers still: - SSH into production - Deploy via bash scripts - Ask ops for access You built a platform nobody uses. This is what happens when you build platforms for platforms not for users"  
[X Link](https://x.com/brankopetric00/status/1996598623933476920)  2025-12-04T15:13Z 14.7K followers, 4371 engagements


"@willvincent Amazing"  
[X Link](https://x.com/brankopetric00/status/1996632098371710985)  2025-12-04T17:26Z 14.7K followers, 31.9K engagements


"Your Kubernetes cluster has: - XX namespaces - XX broken monitoring sidecars - X people who understand the networking - X documentation - A $18K/month AWS bill Your monolith on a VM has: - X server - X person who can deploy - Actual documentation - A $140/month bill - XXXX% uptime But sure K8s is "simpler" and "more reliable." Keep telling yourself that at X AM when nobody knows which pod is failing"  
[X Link](https://x.com/brankopetric00/status/1996660031903658119)  2025-12-04T19:17Z 14.7K followers, 51K engagements


""We need Kubernetes" - said the team with X microservices and X users. Why this drives me insane: - You're spending XX hours/week managing K8s - Your app runs on a $X VPS just fine - 6-month learning curve for zero business value - Your real bottleneck is the database you're ignoring What you actually need: - Docker Compose on a t3.medium - A managed database - Time to build features customers want Stop cosplaying as Netflix. You're not Netflix. Your ego is more expensive than your AWS bill"  
[X Link](https://x.com/brankopetric00/status/1996679410049515679)  2025-12-04T20:34Z 14.7K followers, 10.3K engagements


"Your environments: - dev - dev2 - dev-test - test - qa - qa2 - staging - staging-prod - pre-prod - prod-test - prod Your actual testing: - Everything in dev - Skip to prod - Fix in prod You maintain XX environments that cost $31K/month. You use X of them. This is infrastructure hoarding. Every extra environment is tech debt with a monthly subscription"  
[X Link](https://x.com/brankopetric00/status/1996909416432435451)  2025-12-05T11:48Z 14.7K followers, 21.9K engagements


"Moved from a startup to enterprise and learned that scale changes everything. The differences: - 5-minute changes became 5-week approval processes - Simple deployments required XX different teams - Security wasn't optional it was the starting point - Documentation wasn't nice-to-have it was survival DevOps principles are universal but implementation is context-dependent"  
[X Link](https://x.com/brankopetric00/status/1997966026051359096)  2025-12-08T09:46Z 14.7K followers, 7917 engagements


""Reserved Instances always save money" - not for bursty workloads. Compare: EC2 t3.medium running X hours/day - On-Demand: $0.0416/hour X XX = $99.84/month - Reserved (1-year): $0.0270/hour XX XX = $194.40/month Reserved pricing assumes 24/7 usage. Burstable credit model doesn't change the hourly billing. For XX% utilization workloads On-Demand beats Reserved Instances every time"  
[X Link](https://x.com/brankopetric00/status/1998108097475170689)  2025-12-08T19:11Z 14.7K followers, 1661 engagements


"@eiselems docker compose has health checks so you can postpone containers going down unless new app is started"  
[X Link](https://x.com/brankopetric00/status/1998681771559252271)  2025-12-10T09:10Z 14.7K followers, 1185 engagements


"Datadog bill progression: Month 1: $XXX Month 3: $XXX Month 6: $2340 Month 9: $4180 Month 12: $6730 What happened: Started with XX hosts Added custom metrics (charged per metric) Went from XXX metrics to 4200 metrics Each metric: $0.05/month ($210/month for metrics alone) Added APM ($31 per host) Added log management (holy shit expensive) Migrated to self-hosted: Prometheus + Grafana + Loki Setup time: X week Monthly cost: $XXX (EC2 + storage) Saved $6550/month. Managed services are convenient until they're not"  
[X Link](https://x.com/brankopetric00/status/1991828961798521223)  2025-11-21T11:20Z 14.7K followers, 11.6K engagements


"Burstable instances (t3) worked better than we thought for production. - Worried about CPU credit exhaustion under load - Monitored credit balance obsessively for X months - Never dropped below XX% credits even during spikes - Saved $1200/month vs m5 instances Turns out our 'high-performance' workload had predictable burst patterns. Don't let instance class prejudice cost you money"  
[X Link](https://x.com/brankopetric00/status/1995935867509088325)  2025-12-02T19:19Z 14.7K followers, 3213 engagements


"@PasqualePuzio What would be your take"  
[X Link](https://x.com/brankopetric00/status/1996880932360610130)  2025-12-05T09:54Z 14.7K followers, 13.7K engagements


"Lambda cold starts cost money through increased duration charges. Warm execution: 200ms $XXXXXXXXX Cold start: 800ms $XXXXXXXXX (4 cost) High-frequency function (1M invocations/month XX% cold start rate): - Warm: 800K $XXXXXXXXX = $XXXX - Cold: 200K $XXXXXXXXX = $XXXX Cold starts double Lambda costs for high-traffic functions. Provisioned concurrency eliminates cold starts but adds $24/month per concurrent execution"  
[X Link](https://x.com/brankopetric00/status/1998060784857784651)  2025-12-08T16:03Z 14.7K followers, 2526 engagements


"Step Functions cost more than Lambda for simple orchestration. Workflow: X Lambda functions in sequence 100K executions/month Direct Lambda chaining: $XXXX (compute only) Step Functions: $XXXX state transitions + $XXXX compute = $XXXX X cost premium for visual workflow and error handling. Complex workflows justify the cost. Simple chains favor direct Lambda integration"  
[X Link](https://x.com/brankopetric00/status/1998099792958751096)  2025-12-08T18:38Z 14.7K followers, 2092 engagements


"Kubernetes migration almost killed our startup. Where we were: - X EC2 instances - Ansible for deploys - Boring but working - $1200/month AWS bill Why we migrated: - New investor wanted 'cloud-native' - Engineers wanted K8s experience - Competitors were using it - Seemed like the future X months later: - X engineers spending full-time on K8s - AWS bill at $4500/month - Deploys took longer than before - More outages not fewer - Product development stalled We rolled back: - Moved to ECS Fargate - X week migration - Back to $1800/month - Engineers back on features K8s is amazing for scale. We"  
[X Link](https://x.com/brankopetric00/status/1986276714364731478)  2025-11-06T03:37Z 14.7K followers, 1.4M engagements


"We went serverless to reduce costs. Lambda API Gateway DynamoDB. Clean architecture. Modern stack. Six months later: - Cold starts were killing user experience - DynamoDB was 3x more expensive than RDS would have been - Debugging required piecing together logs from XX different services - Local development became nearly impossible Serverless saved us from managing servers. It didn't save us money or complexity. Pick your architecture for your actual constraints not the latest conference talks"  
[X Link](https://x.com/brankopetric00/status/1996031874997911922)  2025-12-03T01:40Z 14.7K followers, 38K engagements


"Your Docker image is 2.5GB because you installed the entire Ubuntu desktop environment to run a Node.js API and your container has more attack surface than a traditional server"  
[X Link](https://x.com/brankopetric00/status/1996077964757184746)  2025-12-03T04:44Z 14.7K followers, 217.7K engagements


""Go serverless" they said. "No infrastructure to manage" they said. Now you have: - XX Lambda functions - Cold start issues you can't fix - Timeouts you can't control - Debugging that's a nightmare - CloudWatch logs that cost $800/month - Vendor lock-in so deep you can't leave And your bill Higher than running everything on EC2. Serverless doesn't mean no servers. It means no control over servers you're paying premium prices for"  
[X Link](https://x.com/brankopetric00/status/1996566485720797282)  2025-12-04T13:05Z 14.7K followers, 193.7K engagements


"Month 1: "Cloud is so flexible" Month 3: "Why is our AWS bill $8K" Month 6: "It's fine we'll optimize later." Month 12: "$47K/month is just the cost of doing business." Month 18: "How did we get to $89K" You're running: - XX oversized RDS instances - NAT Gateways costing $1200/month - EBS volumes from 2019 nobody uses - CloudWatch logs you never read You're not scaling. You're bleeding. Every month you ignore it costs more than the month before"  
[X Link](https://x.com/brankopetric00/status/1996636374443671726)  2025-12-04T17:43Z 14.7K followers, 22.8K engagements


""Our database won't scale". Database: - 40GB total data - XX queries per second - X indexes on query columns - N+1 queries everywhere - 200ms average query time Solution: - Shard across XX databases - Add read replicas - Implement caching layer - Switch to "web scale" NoSQL Actual solution: - Add X indexes - Fix the N+1 queries - 5ms query time - $40/month Postgres You don't have a scaling problem. You have a competence problem"  
[X Link](https://x.com/brankopetric00/status/1996873680849092811)  2025-12-05T09:26Z 14.7K followers, 244.5K engagements


"You put an API Gateway in front of your API. Now you have: - Rate limiting (that blocks real users) - Authentication (that adds 200ms latency) - Request transformation (that breaks in production) - Monitoring (that duplicates what you had) - Caching (that serves stale data) - $890/month Kong/Apigee bill Your original problem: - Wanted to add CORS headers You spent X weeks and $10K/year to add "Access-Control-Allow-Origin: *" A 5-line nginx config would have done the same thing"  
[X Link](https://x.com/brankopetric00/status/1997114517738684586)  2025-12-06T01:23Z 14.7K followers, 75.7K engagements


"The best monitoring tool is the one the entire team actually checks during incidents. Standardize on one observability platform. Splitting metrics across X tools means nobody knows where to look when production burns. Consolidation beats coverage"  
[X Link](https://x.com/brankopetric00/status/1997976195603059069)  2025-12-08T10:27Z 14.7K followers, 1461 engagements


"Your base Docker image is 2GB. You are shipping an entire operating system every time you change a line of CSS. This isn't just a minor inefficiency. This is a systemic failure that causes: - Excruciatingly slow CI/CD pipelines. - Massive ECR/storage bills. - A giant attack surface filled with vulnerabilities you'll never patch. Learn how to use multi-stage builds. Use a distroless or Alpine base image. Scan your images for vulnerabilities. A container should contain your app not the entire Debian repository. It's called a container not a warehouse"  
[X Link](https://x.com/brankopetric00/status/1998153397598359563)  2025-12-08T22:11Z 14.7K followers, 13.4K engagements


"Cost optimization without performance measurement leads to over-optimization. Scenario: API response time SLA: 200ms Current: m5.xlarge 150ms average response Downsized: m5.large 190ms average response Cost savings: $70/month Performance margin: 10ms from SLA breach Optimization reduced cost XX% but eliminated performance buffer. Monitor performance metrics during cost optimization to avoid invisible degradation"  
[X Link](https://x.com/brankopetric00/status/1998396718874853419)  2025-12-09T14:18Z 14.7K followers, 2584 engagements


"Running open source to avoid vendor fees while paying X engineers to maintain it costs more than the SaaS. Buy unless building it is your competitive advantage. Operating Elasticsearch costs $300K/year in engineering time. Elastic Cloud costs $40K. Do the math"  
[X Link](https://x.com/brankopetric00/status/1998525816968200265)  2025-12-09T22:51Z 14.7K followers, 4521 engagements


"You have XX monitoring dashboards XX alerts and X observability platforms. You still found out about the outage from Twitter. Your observability stack: - Datadog: $5000/month for metrics no one looks at - New Relic: $3000/month for traces that confuse everyone - PagerDuty: $1000/month for alerts everyone ignores - Grafana: Because pretty dashboards solve everything Real observability: knowing your system is broken before your customers do. You have expensive noise not signal"  
[X Link](https://x.com/brankopetric00/status/1998878159756763244)  2025-12-10T22:11Z 14.7K followers, 4541 engagements