SLA credits, outage runbooks, and cloud-cost recovery — the short version of everything we learn monitoring the big cloud providers.
September 30, 2026
Every credit claim travels through a support case, and every support case travels on a clock of its own. Fifteen minutes, an hour, or a category that is not covered at all: what the big three commit to when the incident is burning, and why the response clock is not the credit clock.
Read moreSeptember 29, 2026
DynamoDB, Cosmos DB, Firestore, and Bigtable carry the highest uptime commitments in the cloud: up to 99.999%. Which deployment shapes earn five nines, what the credit tiers pay, and how to file inside the window.
Read moreSeptember 28, 2026
The services that tell you when everything else breaks have their own SLAs, and the definitions are observability problems in themselves: late alerts, failed metrics, missing log ingestion. What CloudWatch, Azure Monitor and Google Cloud Observability promise, and the compounding irony of claiming on them.
Read moreSeptember 27, 2026
Cloud SLAs measure the resources you already have, not the ones you cannot create. Quotas, scale-ups and capacity constraints sit outside the uptime maths by design, and two Azure reservation products show what it takes to move them inside.
Read moreSeptember 26, 2026
When you run AWS Organizations, multiple Azure subscriptions, or several GCP projects, an outage credit lands in one account, not all of them. Who files, which account gets paid, and the per-provider rules that decide it.
Read moreSeptember 25, 2026
CloudFront, Azure CDN, Cloud CDN and Spaces all promise roughly 99.9% uptime, but each measures downtime on a different meter. Exact credit tiers, deadlines, and the evidence each provider demands.
Read moreSeptember 24, 2026
Security services promise uptime on their own terms: uninspected requests at AWS, error rates above a hundredth of a percent at Google, unmitigated attacks at Azure. The security edge has the most inventive downtime definitions in the cloud, and the credits follow.
Read moreSeptember 23, 2026
Every major cloud SLA says the same thing in different words: service credits are the whole compensation for an outage. The remedy clause, the warranty disclaimer and the damages cap, and what they mean before your next incident.
Read moreSeptember 22, 2026
The private link between your network and the cloud carries the strictest fine print in the SLA world: what you built decides what you can claim. Commitments, credit ladders and money math for Direct Connect, ExpressRoute and Cloud Interconnect.
Read moreSeptember 21, 2026
Warehouse uptime promises range from 99.5% to 99.99% depending on deployment shape, and the measurement rules decide more claims than the headline numbers. Credit ladders, money math and deadlines for BigQuery, Redshift, Synapse and Athena.
Read moreSeptember 20, 2026
Azure App Service and DigitalOcean App Platform carry 99.95% promises with credits attached. App Engine measures errors, not connectivity. Elastic Beanstalk has no SLA of its own. What that means for a claim.
Read moreSeptember 19, 2026
AWS runs two compute SLAs over one VM, Azure pays by disk type, Google splits by zone count, DigitalOcean pays a flat 100%. The commitments, the credit tiers, the money math, and the claim windows.
Read moreSeptember 18, 2026
AWS and Google leave IAM, consoles, organizations, and billing out of their SLA indexes, so failures there earn no credits. Microsoft covers Entra ID but not the portal. What that costs, and how to find your exposure.
Read moreSeptember 17, 2026
The exact SLA commitments, credit tiers, and claim deadlines for SQS, SNS, EventBridge, Azure Service Bus, and Google Pub/Sub, and why these SLAs measure API calls, not your messages.
Read moreSeptember 16, 2026
West US lost inbound and outbound traffic for 4 hours 57 minutes on July 23, 2026. Here is which Azure services pay credits, what the math yields per service, and the 60-day claim window that closes September 21.
Read moreSeptember 15, 2026
All three platforms promise 99.9%, and the fine print decides everything: only HTTP 500s count on Bedrock, single-node deployments earn nothing on Vertex AI, Azure caps model endpoint credits at 25%, and the claim windows run from 30 to 60 days.
Read moreSeptember 14, 2026
Lambda, Azure Functions, and Cloud Run Functions all promise 99.95% uptime. The fine print on measurement decides whether you ever get a credit: request error rates, execution logs, the no-traffic loophole, and payout floors.
Read moreSeptember 13, 2026
How AWS, Azure, Google Cloud and DigitalOcean decide a load balancer was down, the credit schedules for ALB, App Gateway, Front Door and Cloud LB, and the preconditions that decide whether a claim exists.
Read moreSeptember 12, 2026
Spot capacity costs up to 90 percent less and carries no uptime promise. The documented SLA position for AWS, Azure, and Google Cloud, what a reclaim looks like in practice, and the math that decides whether the discount is worth it.
Read moreSeptember 11, 2026
The uptime commitment for a managed database depends on the deployment option, not the product: Multi-AZ or single, zone redundant or not, HA or zonal. Exact thresholds, credit schedules, the tiers where no SLA exists, and the claim route for each.
Read moreSeptember 10, 2026
S3 promises 99.999999999% durability, yet a lost object earns you nothing. Why every cloud SLA measures availability and error rates instead of durability, what EBS and Azure actually commit to, and the numbers that separate marketing from contract.
Read moreSeptember 9, 2026
Route 53, Cloud DNS, and Azure DNS all promise 100 percent uptime, and the credit tiers jump straight from 10 percent to 100 percent of the bill. What the fine print actually requires, how to measure a breach, and how to file before the window closes.
Read moreSeptember 8, 2026
Provider SLA credits flow to whoever holds the account, and your customers hold contracts with you, not with AWS. How the pass-through works, what it costs you, and how to set service commitments you can actually keep.
Read moreSeptember 7, 2026
DigitalOcean rewrote its SLA into per-product pages, and the differences are material: 100% of a breached Droplet charge, 100% of lost time on Volumes, tiered credits on DOKS and load balancers, and a 24-hour notice rule on Spaces.
Read moreSeptember 6, 2026
Google computes financial credits per product, per region, per project, and the 60-day window (30 on several services) runs from when you become eligible. How to batch a multi-service outage into one correct claim and what disqualifies it.
Read moreSeptember 5, 2026
The provider SLA is their promise; your SLO is yours. How to set an error budget that reflects what your users experience, measure it from your own telemetry, and use provider credits as one input in the availability math rather than the whole answer.
Read moreSeptember 4, 2026
Every cloud SLA carries exclusions: factors outside reasonable control, your own equipment and software, suspensions. What the exclusion clauses actually say at AWS, Azure, Google, and DigitalOcean, and how to keep your claim on the right side of them.
Read moreSeptember 3, 2026
The Personal Health Dashboard is free for every account, but the AWS Health API that turns health events into automated evidence needs Business, Enterprise On-Ramp, or Enterprise support. Where the wall sits and what to do on each side of it.
Read moreSeptember 2, 2026
Providers rewrite their SLAs on their own schedules, and the edition in force on your incident date decides the threshold, the credit, and the deadline. How to pin the exact version that governs your claim.
Read moreSeptember 1, 2026
Status pages communicate what a provider decided to say, minutes or hours late. The measurements that matter run from your side: probes, error rates, and the SLA math that turns failures into credits.
Read moreAugust 31, 2026
Regional cloud outages do not respect office hours. How to stage alerts so a 3 AM outage wakes the right person, preserves claim evidence automatically, and leaves everyone else asleep.
Read moreAugust 30, 2026
Filing an SLA credit claim takes about an hour. Worked example: what the credit pays at different cloud bill sizes, what your hour costs, and the break-even point where claims stop making sense.
Read moreAugust 29, 2026
One dead Availability Zone does not entitle you to a credit if your deployment never spanned zones. How AWS, Azure, and Google define the deployment shapes that decide whether an outage is a claimable SLA breach or just a bad day.
Read moreAugust 28, 2026
Automated SLA credit tools watch provider health feeds, detect breaches, and draft claims. Where they stop: filing the request, bespoke SLAs, appeal letters, and the judgment calls that still need a human.
Read moreAugust 27, 2026
AWS rejects late credit claims with no appeal, Azure cuts you off after two months, Google after 30 days, and DigitalOcean after two billing cycles. The dollar math of a missed deadline, and how to recover a claim that is still inside the window.
Read moreAugust 26, 2026
Cloud providers rewrite their SLAs on their own schedules, and the version you signed is often not the one that decides your next credit claim. What AWS, Azure, GCP, and DigitalOcean actually publish, what changed recently, and what to lock in at renewal.
Read moreAugust 25, 2026
Scheduled maintenance can take your service down without counting as an outage. Which clouds exclude it from the SLA, which require notice first, and when you can still claim.
Read moreAugust 24, 2026
Where AWS, Azure, GCP and DigitalOcean keep years of incident history, what their public logs omit, and the numbers worth computing before you commit to a region.
Read moreAugust 23, 2026
Cloud SLAs exclude whole categories of downtime before the uptime math starts: beta services, your own configuration errors, single-datacenter deployments, latency without unavailability. What AWS, Azure, Google Cloud, and DigitalOcean actually exclude.
Read moreAugust 22, 2026
SLA credits repay a fraction of one month of one service. A second region bills you every month. The breakeven math, the single-AZ eligibility trap, and when to engineer around outages instead of claiming.
Read moreAugust 18, 2026
Your SLA credit is a percentage of what you actually paid, not list price. AWS excludes one-time Reserved Instance payments, Azure credits fees actually paid, GCP uses the monthly bill for the affected service. The credit-base rules, with money math.
Read moreAugust 17, 2026
Managed Kubernetes SLAs cover only the control plane, and credits are a percentage of the control plane fee, not your node bill. Exact thresholds, credit tiers, deadlines, and filing steps for EKS, AKS, GKE, and DOKS.
Read moreAugust 16, 2026
When AWS, Azure, GCP, or DigitalOcean goes down, your customers blame you. This runbook covers the 15-minute first update, the 30-minute heartbeat, four templates that work, who owns what, and how to preserve your SLA credit claim while you talk.
Read moreAugust 15, 2026
A denied SLA credit claim is usually fixable. The five reasons AWS, Azure, Google Cloud, and DigitalOcean reject claims, the deadline math that governs your resubmission, and the per-provider path back to a paid credit.
Read moreAugust 14, 2026
99.9% allows 8.77 hours of downtime a year. 99.99% allows 52.6 minutes. Which nines your provider actually owes you depends on how you deploy, and the credit tiers when they breach.
Read moreAugust 13, 2026
A breached SLA does not put cash in your account. How AWS, Azure, Google Cloud, and DigitalOcean pay service credits, when you actually see them, and the small print that changes the amount.
Read moreAugust 12, 2026
The three acronyms on every provider SLA page mean different things. How AWS, Azure, GCP, and DigitalOcean define and measure downtime, and why the SLI is the clause that decides whether you get a credit.
Read moreAugust 11, 2026
Cloud SLA credits die on deadlines: 30 days for Google Cloud and DigitalOcean, about 60 for AWS and Azure, counted from the end of the month, not the incident. Exact windows from the current SLA pages plus a calendar routine that takes ten minutes a month.
Read moreAugust 10, 2026
Providers owe service credits after most major outages, yet almost nobody files. The four real blockers, who actually collects, and the deadlines that decide it.
Read moreAugust 10, 2026
When a client's cloud account has an outage, the provider owes a credit and the client will never file. IT providers can claim in under an hour: who can file on each provider, the exact windows, the evidence you already hold, and how the credit becomes a visible win.
Read moreAugust 9, 2026
How MSPs and IT providers can turn provider SLA credits into a client-facing service: who is allowed to file, the real deadlines, where the money lands, and pricing models that pay for the work.
Read moreAugust 7, 2026
Provider status pages lag real outages by 20 minutes or more and miss whole categories of incidents. What AWS, Azure, GCP and DigitalOcean status pages cover, what they hide, and why SLA credits never depend on them.
Read moreAugust 6, 2026
How to claim Azure SLA credits: reading Service Health correctly, the thresholds that trigger a credit, filing a refund request in the portal, and the two month deadline you cannot miss.
Read moreAugust 6, 2026
Published estimates put downtime at $15,000 per minute for large enterprises and $100,000 to $5 million per hour. How to calculate your own hourly number with revenue, payroll, and churn math, plus a worked example.
Read moreAugust 6, 2026
DigitalOcean's SLA is per product, not one document: Droplets at 99.99% per instance, Managed Databases from 99.5%, Spaces at 99.9%. Exact credit tiers, filing deadlines from 30 days to three months, and the 24-hour notice rule for Spaces.
Read moreAugust 6, 2026
Google Cloud owes you a financial credit when it breaches an SLA, but only if you file inside the window: 30 days for most services, 60 for Compute Engine. Exact windows, credit tiers, evidence, and the filing route.
Read moreAugust 5, 2026
The exact evidence AWS, Azure, Google Cloud, and DigitalOcean require in an SLA credit claim: request logs, timestamps, incident IDs, resource IDs, and the traps that get claims rejected.
Read moreAugust 4, 2026
Measure cloud outages with your own probes: check intervals, NTP time sync, and the monthly uptime math that decides whether an SLA credit claim exists. Exact threshold numbers included.
Read moreJuly 20, 2026
Cloud providers owe you service credits when they breach their uptime SLAs, but you have to ask. Step-by-step claim instructions, thresholds, and deadlines for AWS, Azure, Google Cloud, and DigitalOcean.
Read moreJuly 20, 2026
The exact AWS SLA credit tiers for EC2, S3, RDS, and Lambda, the 60-day claim deadline, what evidence AWS Support actually asks for, and a ready-to-submit claim template.
Read moreJuly 20, 2026
A 60-minute runbook for cloud outages: confirm it is really the provider, protect your systems, capture the evidence your SLA claim will need, and note the deadline before the postmortem fades.
Read more