Telecom · 4 months

Erlang Backend Rewrite for Bulk SMS Provider

Rewrote the messaging backend in Erlang, optimizing performance and cutting cloud costs by 30%.

Client not named under confidentiality terms; metrics from the engagement closeout report.

3×

Throughput headroom

30%

Monthly compute cost reduction

99.95%

Delivery SLO achieved

Before

A monolithic Node.js service was throttling high-volume SMS bursts, forcing the team to over-provision Kubernetes clusters in three regions.

Intervention

We redesigned the dispatcher as OTP applications with back-pressure aware queues, plus observability hooks that expose delivery lag in near real time.

Observed result

3× throughput headroom during carrier promotions.
30% reduction in monthly compute spend thanks to right-sized nodes.
99.95% delivery SLO with automated failover drills.

Limits

These measures describe this engagement, not a guarantee for another system.

Selected records include work delivered before Anystack Engineering was incorporated in May 2024 and work delivered through partner consultancies. Each record states its provenance; these results are not evidence of the current three-person pod model.

See how an engagement is set up.

The proposal records the assigned engineers, working arrangement, review and handover.

How engagements are set up →