Lizrd Bloghttps://www.lizrd.ai/blogLizrd is the engineering excellence platform that shows engineering teams exactly where their infrastructure money goes — and guides quick, low-risk fixes.Cost as a first-class SLO: building a cost-aware culturehttps://www.lizrd.ai/blog/cost-as-a-first-class-slohttps://www.lizrd.ai/blog/cost-as-a-first-class-sloWed, 19 Aug 2026 00:00:00 GMTOptimization that lives in quarterly cleanups always regresses. Treating cost like reliability — owned and continuous — is what makes it stick.Serverless isn't automatically cheaphttps://www.lizrd.ai/blog/serverless-isnt-automatically-cheaphttps://www.lizrd.ai/blog/serverless-isnt-automatically-cheapTue, 18 Aug 2026 00:00:00 GMTServerless removes idle cost but adds a per-invocation price. Past a certain scale, tuning and the crossover point decide the bill.Your non-prod is bigger than you thinkhttps://www.lizrd.ai/blog/your-non-prod-is-bigger-than-you-thinkhttps://www.lizrd.ai/blog/your-non-prod-is-bigger-than-you-thinkMon, 17 Aug 2026 00:00:00 GMTCI, staging, and preview environments run around the clock at production sizes for a fraction of the value. That is reclaimable spend.Cross-account, multi-cloud, one number: the transparency problemhttps://www.lizrd.ai/blog/multi-cloud-one-numberhttps://www.lizrd.ai/blog/multi-cloud-one-numberFri, 14 Aug 2026 00:00:00 GMTFragmented billing across accounts and clouds hides waste in the seams. A single allocated view is what changes decisions.Spot instances for the workloads you were told not tohttps://www.lizrd.ai/blog/spot-instances-beyond-batchhttps://www.lizrd.ai/blog/spot-instances-beyond-batchThu, 13 Aug 2026 00:00:00 GMTSpot is not just for stateless batch. With the right interruption patterns, it safely covers far more of your fleet.You can't optimize what you can't attribute: a tagging playbookhttps://www.lizrd.ai/blog/tagging-playbook-attributionhttps://www.lizrd.ai/blog/tagging-playbook-attributionWed, 12 Aug 2026 00:00:00 GMTUntagged spend is unownable spend. A pragmatic allocation strategy makes "who owns this cost?" an answerable question.Autoscaling that saves money instead of hiding wastehttps://www.lizrd.ai/blog/autoscaling-that-saves-moneyhttps://www.lizrd.ai/blog/autoscaling-that-saves-moneyTue, 11 Aug 2026 00:00:00 GMTAutoscaling on bad requests amplifies over-provisioning instead of fixing it. Get the inputs right and it becomes a real cost lever.Unit economics: cost per request is the metric that mattershttps://www.lizrd.ai/blog/unit-economics-cost-per-requesthttps://www.lizrd.ai/blog/unit-economics-cost-per-requestSat, 08 Aug 2026 00:00:00 GMTA monthly total tells you what you spent, not whether it was worth it. Cost per request ties infrastructure to the business.The Graviton math: when ARM pays for the migrationhttps://www.lizrd.ai/blog/the-graviton-mathhttps://www.lizrd.ai/blog/the-graviton-mathFri, 07 Aug 2026 00:00:00 GMTGraviton offers real price/performance gains, but the migration has a cost. Here is how to decide which fleets are worth moving.Rightsizing databases without a 3 a.m. pagehttps://www.lizrd.ai/blog/rightsize-databases-safelyhttps://www.lizrd.ai/blog/rightsize-databases-safelyWed, 05 Aug 2026 00:00:00 GMTDatabases are the scariest thing to rightsize and often the most over-provisioned. Read the right signals and it becomes routine.Idle and orphaned: the silent tax on every cloud accounthttps://www.lizrd.ai/blog/idle-and-orphaned-the-silent-taxhttps://www.lizrd.ai/blog/idle-and-orphaned-the-silent-taxTue, 04 Aug 2026 00:00:00 GMTZombie volumes, unattached IPs, and idle databases accumulate in every account. A repeatable sweep reclaims them without breaking anything.How to rightsize without breaking productionhttps://www.lizrd.ai/blog/rightsize-without-breaking-productionhttps://www.lizrd.ai/blog/rightsize-without-breaking-productionSat, 01 Aug 2026 00:00:00 GMTRightsizing is the highest-impact, lowest-risk cloud saving — if you do it with evidence and a rollback. Here’s a safe, repeatable way to do it.Your storage bill is mostly forgotten decisionshttps://www.lizrd.ai/blog/your-storage-bill-is-forgotten-decisionshttps://www.lizrd.ai/blog/your-storage-bill-is-forgotten-decisionsFri, 31 Jul 2026 00:00:00 GMTgp2 volumes, orphaned snapshots, and un-tiered S3 compound quietly. Storage waste is the cost of choices nobody ever revisits.Commitments without regret: Savings Plans vs RIs vs Spothttps://www.lizrd.ai/blog/commitments-without-regrethttps://www.lizrd.ai/blog/commitments-without-regretWed, 29 Jul 2026 00:00:00 GMTCommitments cut rates but lock in whatever you commit to. A portfolio approach layers them under a baseline you can actually measure.Kubernetes is a cost black box until you allocate ithttps://www.lizrd.ai/blog/kubernetes-cost-allocationhttps://www.lizrd.ai/blog/kubernetes-cost-allocationMon, 27 Jul 2026 00:00:00 GMTA shared cluster bill hides which team, service, and pod drove the spend. Here is how to turn node cost into per-workload truth.FinOps for engineers: make cost a metric, not a meetinghttps://www.lizrd.ai/blog/finops-for-engineershttps://www.lizrd.ai/blog/finops-for-engineersFri, 24 Jul 2026 00:00:00 GMTFinOps works when cost behaves like any other engineering signal — measured continuously, owned by the team, and acted on in the normal flow of work.The NAT gateway tax: the line item nobody readshttps://www.lizrd.ai/blog/the-nat-gateway-taxhttps://www.lizrd.ai/blog/the-nat-gateway-taxThu, 23 Jul 2026 00:00:00 GMTData-processing and cross-AZ transfer quietly turn NAT gateways into a top-3 bill item. Here is how to see the flows and cut the charge.The 30% problem: cloud waste is an execution failurehttps://www.lizrd.ai/blog/cloud-waste-is-an-execution-problemhttps://www.lizrd.ai/blog/cloud-waste-is-an-execution-problemWed, 15 Jul 2026 00:00:00 GMTMost cloud bills carry 20–30% waste. It persists not because teams can’t see it, but because seeing waste and fixing it are two different jobs.