E-commerce Cloud Infrastructure
Cloud infrastructure for the world's largest retailer of diamond jewellery, running storefronts for brands like Kay, Zales, Jared and Blue Nile on one elastic platform that scales up for holiday and Valentine's peaks.
- Client
- Signet Jewelers
- Category
- E-commerce
- Platform
- AWS cloud + shared commerce platform
- Duration
- Ongoing engagement
- Team
- 6 engineers
0.0%
uptime through peak season
0x
traffic absorbed on peak days
0
banner storefronts on one platform
Instances scale with holiday traffic while every storefront shares one platform. Switch between banners to compare.
Holiday peak · requests/sec (k) vs app instances
Live sessions
48,210
Conversion
3.1%
Avg. order
$612
One shared commerce platform and catalogue behind every banner storefront.
Service health · p95 latency
- storefront-web182 ms
- catalog-search96 ms
- inventory-availability124 ms
- cart-checkout268 ms
- payments-gateway341 ms
- order-routing77 ms
ORD-7741-20931
1 ct. Round Solitaire Engagement Ring, 14K White Gold
$3,499.00
Pick up in store · Mall of America
- Order placed online
- 2Store stock reserved
- 3Payment authorised
- 4Ring sized & polished
- 5Ready for pick-up
Illustrative view with representative data.
Signet Jewelers runs nine banners, including Kay, Zales, Jared, Banter by Piercing Pagoda, Diamonds Direct, Blue Nile, Peoples, H.Samuel and Ernest Jones, across the US, UK and Canada. Its Connected Commerce strategy treats the website, the app and the store as one journey: browse online, book an appointment, try on in store, or buy online and pick up the same day.
Older custom-built e-commerce infrastructure made that hard. Each banner had grown its own stack, holiday traffic had to be planned for months ahead with fixed capacity, and a release to one storefront meant regression-testing several. Peak days such as Black Friday and Valentine's Day can bring several times normal traffic, and those are exactly the days the business cannot afford downtime.
Fixed capacity for very uneven traffic
Servers were sized for the busiest week of the year and sat mostly idle the rest of the time, yet flash sales could still overwhelm them.
We moved the storefront and API tier to containerised services on AWS with horizontal autoscaling on request rate and latency, plus scheduled pre-warming before planned campaigns.
One codebase, many banners
Brand storefronts had drifted apart, so a fix or feature had to be rebuilt per site.
Banners now run on a shared commerce platform and catalogue with per-brand theming and configuration, so a change ships once and every storefront picks it up.
Slow, risky releases
Manual deployments meant changes were bundled into large, infrequent releases, and freezes covered most of Q4.
Infrastructure as code, CI/CD pipelines and blue/green deployments with automatic rollback let teams ship small changes safely, even during peak season.
Blind spots during incidents
Logs and metrics were spread across tools, so finding the cause of a checkout slowdown took too long.
Centralised logging, tracing and p95 latency dashboards per service, with alerts tied to customer journeys such as search, cart and checkout.
- Containerised storefront and API services with autoscaling on AWS
- Shared catalogue, pricing and inventory services for every banner
- Buy online, pick up in store backed by real-time store stock
- Infrastructure as code with repeatable staging and production environments
- CI/CD with blue/green deployments and automatic rollback
- Observability: tracing, centralised logs and journey-level alerts
How the system flows
- 1Customer browses a banner site
- 2Shared catalogue & stock lookup
- 3Cart & checkout
- 4Payment authorised
- 5Store or warehouse fulfilment
- 6Pick-up or delivery
Key operating metrics before the project and after go-live.
Release frequency
+900%Checkout p95 latency
-68%Peak capacity lead time
-92%
API IntegrationsSaksooka
J&T Delivery Service Integration
Native integration between Odoo and J&T Express: shipments booked from delivery orders, labels printed in one click, live tracking pushed back to Odoo and cash-on-delivery reconciled automatically.
0%
less manual booking
0%
fewer failed deliveries
<0 day
COD reconciliation
Point Of SaleMultiple clients
Commission on POS
A configurable sales-commission engine for Odoo Point of Sale, deployed for multiple retail clients, that calculates commissions per cashier and salesperson in real time and pays them through payroll.
0%
automated commission calculation
0
month-end disputes
0%
uplift in attach sales
RetailRetailGiant Inc
AI Inventory Prediction
Machine-learning demand forecasting built into Odoo replenishment: SKU-level forecasts per store, seasonality and promotion effects, and AI-suggested purchase orders that buyers approve in one click.
0%
fewer stock-outs
0%
less excess inventory
0%
forecast accuracy (weekly)
Tell us about your challenge. A solutions architect will reply within one business day with ideas, not a sales pitch.