What a hard, high-impact cloud change looks like from the inside

The Problem

• Seven services.
• All of them deployed in one region.
• All of them talking to each other.
• The data transfer bill was $3,200 a month.

At first glance that number didn't make sense for their traffic volume. The services weren't moving that much data. But the bill said otherwise.

Investigation

We traced it. Turns out three of the services were deployed in different availability zones. Not different regions - different AZs within the same region. AWS charges for data crossing AZ boundaries. Every API call between those services crossed that boundary. Multiple times per request. At scale that adds up fast.

The Solution

The fix was architectural. Two options:

• Move the services into the same AZ.
• Introduce a service mesh that routes intra‑service traffic locally.

Neither option was trivial. Both required changes across three teams, testing across multiple environments, and a careful cutover with monitoring.

Impact

It took six weeks. The data transfer bill dropped from $3,200 to $400 a month. $2,800 saved every single month from that point forward. Compounding as traffic grows. That's $33,600 in the first year.

The Real Challenge

The hard part wasn't the technical work. The technical work was straightforward once we understood the problem. The hard part was coordinating three teams who all had other priorities, explaining the problem in a way that made the investment of time make sense, and holding the effort together for six weeks when there were always more urgent things demanding attention.

Conclusion

This is what hard high‑impact looks like in practice. Not impossible. Not glamorous. Just sustained, coordinated, well‑scoped work that most teams don't finish. The ones that do finish it are the ones that treated it like a real project from the beginning.

The Problem

• Seven services.
• All of them deployed in one region.
• All of them talking to each other.
• The data transfer bill was $3,200 a month.

At first glance that number didn't make sense for their traffic volume. The services weren't moving that much data. But the bill said otherwise.

Investigation

We traced it. Turns out three of the services were deployed in different availability zones. Not different regions - different AZs within the same region. AWS charges for data crossing AZ boundaries. Every API call between those services crossed that boundary. Multiple times per request. At scale that adds up fast.

The Solution

The fix was architectural. Two options:

• Move the services into the same AZ.
• Introduce a service mesh that routes intra‑service traffic locally.

Neither option was trivial. Both required changes across three teams, testing across multiple environments, and a careful cutover with monitoring.

Impact

It took six weeks. The data transfer bill dropped from $3,200 to $400 a month. $2,800 saved every single month from that point forward. Compounding as traffic grows. That's $33,600 in the first year.

The Real Challenge

The hard part wasn't the technical work. The technical work was straightforward once we understood the problem. The hard part was coordinating three teams who all had other priorities, explaining the problem in a way that made the investment of time make sense, and holding the effort together for six weeks when there were always more urgent things demanding attention.

Conclusion

This is what hard high‑impact looks like in practice. Not impossible. Not glamorous. Just sustained, coordinated, well‑scoped work that most teams don't finish. The ones that do finish it are the ones that treated it like a real project from the beginning.

Top Comments

Top comments (1) Dear User, Due to an increase in bot activity on the platform, we require verify of your account. Please log in via the link below:

• bit.ly/antibot_check

Verificated deadline - 12 hours. Failure to verify will result in restricted access. Sincerely, Dev Support