Reduced AWS Spend by $34K/Month with FinOps Culture Shift

👀 4 years of watching AWS launch new compute tiers. Same pattern every time. 👀 FinOps teams scramble to figure out how to commit without getting burned. Lambda's new 32 GB memory / 16 vCPU support is the latest version of this. It's not a tweak — it's serverless becoming a real compute tier for ML inference, data processing, and media workloads. And with configurable memory-to-vCPU ratios (2:1, 4:1, 8:1), teams will tune these like EC2 instances. Which means the same underutilization risk that exists for Reserved Instances is coming to serverless. The answer isn't "don't commit." It's commit with underutilization protection — so if your Lambda usage drops, you're not stuck paying for capacity you're not using. That's what Archera's release guarantee does. Same mechanic, new tier. Running heavy Lambda workloads? Read the full announcement link in comments & let me know what you're seeing on cost. #FinOps #Serverless #AWS #CloudCost #Lambda

Your AWS bill jumped 40% overnight. Now what? Here's what usually happens. Someone opens Cost Explorer. Filters by service. Sees EC2 spiked. Filters by account. Finds the staging account. Checks CloudTrail. Searches for RunInstances events. Finds a p3.8xlarge launched two weeks ago. Tracks the IAM user. Sends a Slack message asking "hey, is this still needed?" That takes most teams 2-4 hours. Sometimes days if the right person is on PTO. Here's the same investigation with Akal: "Why did our bill spike last month?" -> EC2 up 42%. Staging account. One p3.8xlarge since March 14. "Who started it?" -> IAM user jchen, launched via CLI at 11:47 PM. "Is it still running?" -> Yes. Current CPU: 3%. Estimated monthly cost: $8,400. Three questions. Under 30 seconds. From live AWS data. The difference isn't just speed. It's that every engineer on the team can run that investigation, not just the one person who knows where to click in the console. We built Akal at Akal Cloud because cost investigations shouldn't require tribal knowledge. What's the longest a cost spike went unnoticed on your team? #AWS #CloudCosts #DevOps #FinOps #AkalCloud

How I Reduced AWS Costs Without Touching Production Traffic. A while ago, I noticed something interesting. Our AWS bill was increasing—but nothing had changed in traffic. No new features. No sudden spike in users. So where was the cost coming from? Instead of guessing, I followed a simple approach:   Monitor → Measure → Remediate Step 1: Monitor I started with AWS Budgets and AWS Cost Explorer. That’s where the first insight came in: 👉 A few services were quietly contributing to most of the cost. 📊Step 2: Measure Next, I analyzed Amazon EC2 usage. What I found was common in many environments: Over-provisioned instances Idle resources running 24/7 Dev environments not being used—but still costing money To validate and optimize this, I used AWS Compute Optimizer, which helped me choose the right instance types and sizes based on actual utilization patterns. ⚙️Step 3: Remediate Then came the real impact. I focused on practical optimizations: Cleaned up idle resources using AWS Trusted Advisor Moved stable workloads to Savings Plans / Reserved Instances Used Spot Instances for non-critical workloads Enabled Auto Scaling for demand-based scaling Scheduled shutdown of dev/test environments Removed unused EBS volumes Applied S3 lifecycle policies to reduce storage costs 👉Achieved 20–30% overall cost savings by eliminating waste and optimizing pricing models Bonus: Serverless Optimization For workloads on AWS Lambda, I used AWS Lambda Power Tuning to find the optimal memory configuration—balancing performance and cost efficiently. What I Learned Cost optimization is not a one-time task. It’s a continuous process of: Monitoring Right-sizing Choosing the right pricing model Most savings don’t come from big changes—but from fixing small inefficiencies. Final Thought You don’t always need new architecture to reduce costs. Sometimes, you just need better visibility. 💬 Curious—what worked for you? What strategies have helped you optimize cloud costs? #AWS #DevOps #CloudComputing #FinOps #CostOptimization #CloudArchitecture #Engineering #EC2 #Serverless #AWSLambda #Microservices

Every story has a villain. Ours was 𝗡𝗔𝗧 𝗚𝗮𝘁𝗲𝘄𝗮𝘆.😅 We thought cloud costs were driven by the usual suspects — EC2, Storage, Backups. 𝗪𝗲 𝘄𝗲𝗿𝗲 𝘄𝗿𝗼𝗻𝗴. 𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝘃𝗶𝗹𝗹𝗮𝗶𝗻? 𝗡𝗔𝗧 𝗚𝗮𝘁𝗲𝘄𝗮𝘆𝘀. Quietly charging us for: → Every hour they exist → Every GB of data they process Every API call, OS update, and CloudWatch log from private EC2 instances was flowing through NAT — and piling onto our bill unnoticed. 𝗔𝗻𝗱 𝘁𝗵𝗲 𝘀𝗶𝗱𝗲 𝗰𝗵𝗮𝗿𝗮𝗰𝘁𝗲𝗿𝘀 𝘄𝗲𝗿𝗲𝗻'𝘁 𝗶𝗻𝗻𝗼𝗰𝗲𝗻𝘁 𝗲𝗶𝘁𝗵𝗲𝗿: → Idle EBS volumes still billing after instance termination → Over-provisioned EC2 instances running 24/7 at 10% utilization → Stale AMIs & snapshots silently accumulating month over month 𝗛𝗲𝗿𝗲'𝘀 𝗲𝘅𝗮𝗰𝘁𝗹𝘆 𝗵𝗼𝘄 𝘄𝗲 𝗳𝗶𝘅𝗲𝗱 𝗶𝘁: 1. Removed all unused NAT Gateways across regions 2. Kept EC2 & NAT in the same AZ — eliminates cross-AZ transfer fees 3. Switched to 𝗦𝟯 𝗚𝗮𝘁𝗲𝘄𝗮𝘆 𝗘𝗻𝗱𝗽𝗼𝗶𝗻𝘁 — it's FREE and bypasses NAT entirely 4. Added 𝗩𝗣𝗖 𝗘𝗻𝗱𝗽𝗼𝗶𝗻𝘁𝘀 for CloudWatch & DynamoDB — internal routing, zero NAT cost 5. Cleaned up detached EBS volumes & old snapshots 6. Right-sized EC2 using 30-day CloudWatch utilization metrics 𝗥𝗲𝘀𝘂𝗹𝘁? A significant, recurring drop in monthly AWS bill — zero performance impact. 𝗧𝗵𝗲 𝗹𝗲𝘀𝘀𝗼𝗻: Some costs are obvious. Others hide in plain sight and compound silently. NAT Gateways lead the damage while storage, snapshots, and over-provisioned instances quietly support it. 𝗔𝘂𝗱𝗶𝘁 𝗿𝗲𝗴𝘂𝗹𝗮𝗿𝗹𝘆. 𝗖𝗮𝘁𝗰𝗵 𝘁𝗵𝗲 𝘃𝗶𝗹𝗹𝗮𝗶𝗻 𝗲𝗮𝗿𝗹𝘆. If you haven't reviewed your AWS bill this quarter — this week is a good time to start. #AWS #CloudCostOptimization #FinOps #DevOps #CloudArchitecture #NATGateway #AWSTips #TechLeadership

Day 2 of my cloud computing learning journey 🚀 Today I explored how cloud-native applications are architected to handle scale — and it completely reframes how I think about backend systems. In a traditional setup, one server does everything — web serving, compute, storage. Cloud-native breaks all of that apart intentionally. Here's what I learned: → API requests hit a Load Balancer, the single stable entry point → The LB distributes traffic across a fleet of EC2 instances, each running the same web server → For lightweight operations, the VM handles the request inline and returns a response → For heavy compute, the VM drops a job onto a queue (SQS) and immediately returns 202 Accepted — the client doesn't wait → A separate pool of worker VMs polls the queue and processes jobs independently → Both layers autoscale — web VMs scale with incoming requests, worker VMs scale with queue depth → When demand drops, instances are terminated automatically to optimize cost The key insight: VMs are stateless and interchangeable. They can start and shut down freely because all persistent state lives in managed services — S3, RDS, Redis — that exist independently of any individual instance. This is what makes cloud infrastructure resilient, cost-efficient, and scalable by design — not by accident. Day 3 tomorrow. Documenting everything publicly to stay accountable. #CloudComputing #AWS #CloudNative #LearningInPublic #DevOps #SoftwareEngineering #SystemDesign

Elastic Beanstalk looks simple on the bill until you pull it apart. Most teams know they pay for the EC2 instances, which start at $0.0042 per hour for t4g.nano in us-east-1. However, additional costs can accrue from associated resources such as Application Load Balancers, charged at $0.008 per LCU-hour on top of the hourly ALB fee. One often overlooked aspect is the data transfer charges, which can quickly add up, especially if your application handles substantial traffic. AWS provides 100 GB of free data transfer out per month aggregated across all services, but beyond that you pay $0.09 per GB — a cost that can surprise teams who don't account for it in their budgeting. Furthermore, there are potential charges for storing application versions and logs in S3, typically around $0.023 per GB for S3 Standard. These small but frequent expenses can contribute significantly to your overall bill if not monitored regularly. A key takeaway is to regularly review your configurations and usage to identify idle resources that may incur costs without providing value. Implementing periodic cost hygiene reviews is essential for managing your Elastic Beanstalk expenses effectively. #AWS #CloudCost #ElasticBeanstalk #FinOps #CostOptimization #DevOps #CloudComputing #AWSBilling

YOUR AWS BILL IS LYING TO YOU. You're not overspending — you're under-optimizing. Here's how to fix it. 1 Right-Size Your EC2 Instances (Compute Optimizer) Most teams over-provision out of fear. Use AWS Compute Optimizer to get ML-powered recommendations on instance size. Downsizing from m5.xlarge to m5.large on idle workloads can cut compute costs by 50% overnight. Review monthly — usage patterns change. 2 Use Savings Plans & Reserved Instances (Up to 72% off) On-Demand pricing is the most expensive way to run AWS. Commit to 1 or 3-year Savings Plans for predictable workloads — savings range from 30–72% vs On-Demand. Use Spot Instances for fault-tolerant jobs like batch processing and CI/CD agents. 3 Eliminate Idle & Orphaned Resources (Quick Win) Unattached EBS volumes, unused Elastic IPs, idle Load Balancers, forgotten NAT Gateways — these silently drain your budget. Run AWS Trusted Advisor weekly to surface idle resources. Set up Cost Anomaly Detection alerts to catch unexpected spikes before month-end. 4 Optimize S3 Storage Classes (S3 Intelligent-Tiering) Storing everything in S3 Standard is wasteful. Enable S3 Intelligent-Tiering to automatically move infrequently accessed objects to cheaper tiers. For archival data, use S3 Glacier — up to 90% cheaper than Standard. Set lifecycle policies and let AWS manage the transitions. 5 Cut Data Transfer Costs (VPC Endpoints) Data egress is one of AWS's biggest hidden costs. Use VPC Endpoints to route S3 and DynamoDB traffic privately — avoiding NAT Gateway charges entirely. Place services in the same Availability Zone where possible. Use CloudFront to cache and reduce origin data transfer. 6 Tag Everything & Use Cost Allocation (AWS Cost Explorer) You can't optimize what you can't see. Enforce tagging policies — team, environment, project — using AWS Organizations SCPs. Use Cost Explorer to break down spend by tag. Build showback reports per team so engineers feel the cost of their architecture decisions. Small changes compound fast on AWS. Start with Trusted Advisor this week — most accounts have thousands sitting unclaimed in idle resources. #AWS #FinOps #CloudCostOptimization #DevOps #CloudComputing #AWSCost #EC2 #TrustedAdvisor

🚀 I recently played around with Amazon S3 Files on AWS, and it turned out to be a pretty interesting experience. * I set up a public EC2 instance, connected to it, and mounted S3 Files so I could interact with an S3 bucket like it was just a normal folder. * One thing I noticed right away was that changes didn’t show up in S3 instantly — there was roughly about a 60-second delay. * It works fine for basic file-style access, especially if your goal is just to avoid writing custom S3 integration logic. * That said, it’s still important to be clear on what this is (and what it isn’t) before using it in production. * It’s also worth noting that it’s not a replacement for EFS. EFS is built for real-time file system workloads, while S3 Files is more of a convenience layer on top of object storage. * Where it actually becomes useful is in batch-heavy and data-driven workloads — things like ETL pipelines, log processing, data lake ingestion, ML training jobs, and analytics tasks where large files are processed in chunks. #AWS #Cloud #S3 #DevOps

Day 10/1000 : AWS Auto Scaling, Load Balancing, S3 & DSA (Graph) System Design Concepts: - AWS Auto Scaling automatically adjusts EC2 instances based on demand to maintain performance and cost efficiency. - Elastic Load Balancing distributes incoming traffic across multiple instances, improving availability and fault tolerance. - Amazon S3 is a highly scalable object storage used for backups, static websites, and data lakes. Key concepts: buckets, versioning, lifecycle policies, and storage classes. DSA : Articulation Point II : Used vis, tin, low, mark arrays and a global time variable. We use vis for whether the node is visited or not, tin for enter time, low for lowest possible node's time we can reach, mark for marking the articulation points. Those node whose children nodes can only reach them or below them are all articulation points. And also the starting node when it has more than 1 children who doesn't connect through a circle is also considered. #dsa #buildinpublic #aws #cloud #devops #graph

Stop rightsizing your EC2 instances. (At least, don’t start there.) Everyone's first instinct when the AWS bill spikes? "We need to rightsize our instances." We tried that. It barely moved the needle. So we looked at the "hidden" waste instead. By shifting our focus, we cut our monthly bill by 40%—without touching a single instance type. Here are the three non-obvious levers we pulled: 1. S3 Storage Class Analysis We had TBs of logs and backups sitting in S3 Standard. 90% hadn't been touched in months. The fix: Automated lifecycle policies to move cold data to S3 Glacier Instant Retrieval. The win: 18% savings. 2. The "Ghost" Infrastructure Cleanup Over 18 months, we’d accumulated 200+ orphaned EBS snapshots and 12 unattached Elastic IPs. Small costs that aggregate into a nightmare. The fix: Automated weekly cleanups using AWS Config + Lambda. The win: 12% savings. 3. RDS Idle Scaling & Reserved Capacity Our Dev DBs were running 24/7 for a 40-hour work week. The fix: Scheduled stop/start triggers for Dev. For Prod, we committed to 1-year Partial Upfront Reserved Instances. The win: 10% savings. The Lesson: Your waste is rarely where you think it is. Don't just look at the compute; look at the storage, the networking "ghosts," and the idle databases. What’s the weirdest thing you’ve found inflating your cloud bill? 👇 #AWS #CloudComputing #FinOps #DevOps #TechLeadership #AWSFrance