Retail and ecommerceCloud and software

Surviving a 3.4x peak on a re-architected storefront, then spending less off-peak

Vega's storefront buckled every peak season and cost too much the rest of the year. We re-architected the hot paths and matched spend to demand.

Client
Vega Commerce
Timeline
18 weeks
Platforms
AWS, multi-region with Cloudflare edge
Team
5 Synabix engineers across two phases
3.4x
peak traffic handled with no incidents
29%
lower annual infrastructure spend
from weekly to daily
release cadence
The challenge

Where they started.

Vega over-provisioned all year to survive a few peak weekends, and still saw slowdowns when it mattered most. Releases were risky and the team had stopped trusting deploys.

The stack
AWSCloudflareKubernetesArgo CDNode.jsRedis

What we did

  • We profiled the monolith and moved checkout and catalog, the paths that drive load, onto autoscaling services.
  • Cloudflare edge caching and a tuned CDN took pressure off the origin and sped up every page.
  • GitOps delivery with automated checks took releases from a weekly event to a daily routine.
  • Rightsizing and reserved capacity matched spend to the demand curve across the year.
We had our biggest sales weekend ever and barely watched the dashboards. The yearly bill still came down. I did not think both could be true at once.
Marcus BellamyCTO, Vega Commerce
Practices involved
Lower cost, lower risk, one partner

Let's write yours.

Tell us where the business hurts. We will map a pragmatic next step and a plan you own.