The storage wasn’t the problem.

A lot of companies assume their AWS bill is driven by infrastructure size.

In reality, it’s often driven by behavior.

In this case:

very small storage footprint

no heavy traffic

nothing “obviously expensive”

Yet the bill kept growing.

After digging in:

• ~85% of the S3 cost was data transfer
• caused by a pipeline re-downloading the same data every few seconds

Not a scaling issue.

Not a capacity issue.

Just inefficient usage.

We changed one parameter (polling frequency), and:

costs dropped ~40% immediately

No migration.

No refactor.

No savings plans.

The post We cut AWS S3 costs by 40% overnight. appeared first on Elad Levy.

Monthly bill peak : ~$5,600

Nobody could explain why.

So I opened Cost Explorer. Also, to remove confusion I always make sure I cancel savings plans or reserved instances. they are just confusing and lower the bill in an artificial way. It’s another way of saying “Instead of fixing my amazon cost I will prepay amazon years in advance to kind-of-artifically-fix my bill” and let them continue milking my credit card. Moo.

What was actually happening

Not traffic. Super funny. This happens all the time. some wise guy decided the company will manage like twitter so everything is oversized. When I look back 1 year they peak at 5 reqs per second. In other words – you can run the whole app in iPhone 7 and still have room for growth

Leftovers. its exactly the same as starting a subscription for something you dont need and find after 2 years your credit card is bleeding.

• Oversized RDS instance running far below capacity
• 200GB database volume for a ~10GB database
• Multiple instances with almost no traffic
• Years of forgotten S3 backups
• Large “EC2-Other” charges nobody ever checked

After cleanup:

• No rewrite
• No downtime
• No feature changes

New bill: ~$700/month

~90% reduction

• No Spot instances
• No Savings Plans
• No long-term commitments

>>>>Just removing waste.

What I keep seeing

Cloud costs rarely explode.

They accumulate.

A bigger instance “temporarily”

A staging server nobody deletes

Backups never expire

Storage never resized

After a few years nobody remembers what anything is for.

Most infra problems are not scaling problems.

They are ownership problems.

If your AWS bill doesn’t match your traffic, it’s usually fixable.

book a call with me here https://lnkd.in/dBZ8xjEa

The post I audited an AWS account last year. appeared first on Elad Levy.

Cloud spend rises.

Delivery slows.

Complexity increases.

And the instinct is:

“Should we rebuild?”

Usually, that’s the wrong first move.

Before rewriting anything, I ask a simple question:

When was the last time someone actually opened AWS Cost Explorer (or any otherr cloud) and reviewed the last 6 months of spend?

When was the last time you checked on all the company subscriptions to endless tools?

Not yesterday.

Six months.

In most companies I see:

• Storage is heavily over-provisioned
• Databases have 200GB allocated for 10GB of real data
• Idle environments sit untouched
• “Temporary” services never get turned off

And here’s the part most teams miss:

Those costs often don’t show under “EC2 instances.”

They hide under EC2-Other.

Which means they don’t get questioned.

I’ve seen meaningful cost reductions without touching core architecture — just by introducing ownership and review cadence.

The problem usually isn’t tooling.

It’s that infrastructure cost has no clear owner.

Rebuilds feel strategic.

Cost discipline feels operational.

Only one improves margin immediately.

Most of the time, the real leverage isn’t building something new.

It’s subtracting what shouldn’t be there.

If your platform is live and things feel heavier, slower, or more expensive than they should, I’m happy to do a short sanity-check call.

https://lnkd.in/dBZ8xjEa

The post I keep seeing the same pattern in growth-stage companies. appeared first on Elad Levy.

That’s the question a CEO asked me one morning.

When I dug in, here’s what I found:

• Cloud costs spiraling out of control.

excess backups. Do you really need to take a snapshot of the application server EVERY day ??

Deployments that relied on luck more than process (“deploy and pray”).

• No planning, no structure—just chaos.

Most leaders would’ve written this off as a lost cause.

I saw it as an opportunity.

In a matter of weeks, we:

• Cut AWS bills down by right-sizing their infrastructure. Saving over 50% in infra costs
• Built a real release process—fast, safe, repeatable.
• Automated monitoring and rollback—no more 2 a.m. firefights.
• Put sprint and version planning in place—suddenly, delivery was predictable.

Here’s the real lesson:

If your tech feels like a black box, it’s probably draining money and adding risk. With the right changes, it doesn’t just stop bleeding—it becomes a growth engine.

I personally believe that the CEO as well as other key executives in the organization should understand their tech a bit more beyond the “it works” and “it’s expensive” replies

The last months I came across a couple of CEOs that really care about their tech and actually wish to understand why it was chose and how it works. I think this is healthy we all understand each other’s work, plus learning new things is always fun :))

Reply BLACKBOX if you feel your tech is a black box and you have no idea what is going on under the hood

The post How can a company with barely any users have AWS bills bigger than their revenue? appeared first on Elad Levy.

There was a time when starting a tech company meant:

Buying physical servers

Running tape backups

Managing your own Windows domain, Exchange, and printers

Fast forward to today:

We’ve traded it all for cloud convenience—at a massive cost.

At one company, we were burning $2.5M/year on AWS. We slashed it to $100K by doing one thing: rethinking what we actually needed.

• Devs push bloated Docker containers
• Infra teams automate chaos with buzzwords
• Companies rack up six figures in monthly cloud bills
• Real-time dashboards process data no one looks at in real time

We’ve normalized waste:

10 engineers doing what 2 could

Dozens of tools to manage our tools

Cloud bills that exceed payroll

It’s not anti-cloud. It’s anti-waste.

Sometimes, dedicated hosting or hybrid models just make more sense.

Amazingly, when declouding YOU GET BETTER PERFORMANCE FOR LESS MONEY !

I’ve helped teams reduce AWS bills by 90%+ without sacrificing performance.

The secret?

• Know your workload
• Optimize for real usage, not theoretical scale
• Embrace architectural simplicity

Reply “DECLOUD” if you want me to post a second article showing a real-world declouding case.

If you are interested in discussing declouding, please book a meeting with me: https://lnkd.in/ddk578gp

#technology #Startups #LeanStartups #DevOps #CloudComputing #AWS #Startup #FinOps #SRE #EngineeringLeadership #CostOptimization #CloudWaste #DeCloud

The post Fire Your Cloud, Not Your Engineers appeared first on Elad Levy.

Cloudflare is down:

https://lnkd.in/dVZ-UdWJ

Supabase is down/slow

Google cloud and Firebase are down and/or responding slow

Calendly is also responding super slow

AI tools like Cursor are also down

Dive is up ! And I am happy to say that last year we “De-Clouded” all our platform to save on costs and give better performance and stability to our clients.

The more control your company has over your software and infrastructure – the more you will shine in these moments when everything collapses !

This is a classic example that shows that stable websites and companies will capture a lot of their competition. Just because they are more stable and trustworthy.

you can book a call with me here (once Calendly is back ) to learn more about Dive and the services we offer as well as contact me for consultation about tech and help your company shine

https://lnkd.in/ddk578gp

The post https://lnkd.in/d54h5Tdb appeared first on Elad Levy.

The post Thanks again for John Koetsier and Singular for hosting me. appeared first on Elad Levy.

#devops #cloud #python #aws

The post Dear network, I am looking for a full time DevOps wiz for an exciting project related to data and games. appeared first on Elad Levy.