Highly Available AWS Architecture
More than one of everything that matters, and a failover somebody has actually performed.
From $849 5–10 days
AWS Auto Scaling & Rightsizing
Capacity that follows demand, and instances sized for the load you have rather than the one you feared.
Price and scope
From $399
Typically $399–$849, fixed in writing before anything starts.
What moves it up
- Spot adoption, which needs interruption handling before it is safe to rely on
- A stateful tier, where removing an instance is not simply a capacity decision
- Traffic with a shape — a daily peak or a weekly one — that the policy has to anticipate
Some of this you can check yourself, right now, for free: AWS Cost Estimator →
What actually goes wrong
Two mistakes, and most accounts have both. The first is an instance chosen for a peak that happens twice a year and then left running at that size for the other three hundred and sixty three days. The second is subtler: autoscaling configured against CPU for a workload that is not bound by CPU at all. Requests queue behind a database or an external API, CPU stays flat at thirty per cent while response times climb, and the group never scales — then scales out later for a reason unrelated to the problem, which teaches everyone that autoscaling does not work here.
How I find it
CloudWatch CPU and memory at p95 over 30 days, against the instance size actually provisionedaws autoscaling describe-policies — what metric the group is scaling on, and whether it is the binding oneaws ce get-rightsizing-recommendation, read as a starting point rather than as an answerRequest latency against CPU on the same time axis, which is what shows the two are unrelated
What you get
- Instances resized against measured p95 usage, with headroom stated rather than guessed
- A scaling policy on the metric that actually gates throughput for this workload
- Scale-out fast and scale-in slow, so a policy cannot oscillate against its own effect
- Scheduled capacity where load has a known shape, instead of reacting to a peak already underway
- Spot capacity where interruption is genuinely survivable, and nowhere else
Questions
How do I know what size my instances should be?
From p95 over a month, not from an average and not from a peak. An average hides the spikes that matter and a peak sizes you for a moment. Where memory is the constraint, that needs the CloudWatch agent installed first, because AWS does not report memory on its own.
Should I use Spot instances?
For anything that can be interrupted with two minutes' notice and restarted elsewhere — batch work, stateless web tiers behind a load balancer, CI runners — yes, and the saving is large. For a database or anything holding a session in memory, no, and no discount makes that a good trade.
Will scaling down risk an outage?
Only if it is done by guesswork. Sizes come down in steps against measured usage with headroom left deliberately, and anything ambiguous gets changed in a window rather than on a Friday afternoon. If a workload has no safe answer, I will say so and leave it.
What about Savings Plans or Reserved Instances?
Rightsize first, commit second — always in that order. A one-year commitment on an oversized fleet locks in the mistake and pays for it in advance, which is the most expensive possible sequence and a genuinely common one.
If this is not quite it
Amazon ECS / Fargate Deployment
Containers that deploy without downtime, and that stop deploying when the new version is broken.
From $649 3–7 days
Want this done?
Tell me what you are running and I will come back with a fixed price and a date. If it turns out you do not need this, I will say that instead.
Prefer to talk? Book a free call ↗ · Or hire me on Upwork ↗ · Typical reply within one business day.
Before you enquire
When I answer
Sunday to Thursday, 09:00–18:00 EET. Outside that I will still look, but I will not promise a time.
No 24/7 desk, and I will not pretend otherwise
One person, one time zone. If round-the-clock cover is what you need, you need a team, and I will say so rather than sell you a plan that cannot deliver it.
Your AWS bill stays yours
You pay Amazon directly and you keep control of the account. Nothing here resells your infrastructure or sits between you and your own billing.
A price that starts with "from" is a starting price
Every service page lists exactly what pushes a quote above it, before you ask. You get a fixed number in writing before any work begins.