Skip to content
Atlas Hoster
Client portalRequest a quote
High availability and high traffic

Hold Black Friday, the launch, the front page

One server is one point of failure. We design, deploy and manage redundant architectures: load balancer, several web servers, cache, replicated database, CDN. And we load-test them before your customers arrive.

Request a quote Three starting points
  • On-call engineers 24/7
  • On quote
High availability and high traffic

The reference architecture

01

CDN and web application firewall

Cloudflare in front of everything: pages and images cached close to the visitors, attacks and bots filtered.

02

Load balancer

HAProxy or nginx spreads traffic over several web servers, drops the one that stops answering, and terminates TLS.

03

Redundant web servers

Two identical servers or more, deployed by automation: one failing is invisible, and one is added for a peak.

04

Application cache

Redis or Varnish absorb repeated reads: sessions, carts, pages, query results.

05

Replicated database

A primary and a replica, failover tested, point-in-time backups: your data survives a server.

06

Storage and backups

Shared files or object storage, verified backups and a documented restore.

How we work

  1. Audit

    We read your application, your traffic and your past peaks, and we measure what limits you today.

  2. Design and quote

    A diagram, a sizing, a monthly price per component, and what one more tier costs.

  3. Deployment and migration

    Infrastructure deployed by automation, migration rehearsed, cut-over scheduled outside your peak hours.

  4. Load tests

    We simulate your peak (k6, Locust) and fix what bends: cache, queries, limits, sizing.

  5. Operations

    Monitoring, 24/7 on-call, monthly report, and a written post-mortem after every incident.

Three starting points

Every architecture is sized for you; these packs give the order of magnitude. Prices on quote.

Shop

PrestaShop, WooCommerce, Magento

  • 2 web servers + load balancer
  • Primary database + replica
  • Redis, CDN and web application firewall
  • Load test before every sales operation
On quoteRequest a quote

Media

News sites, high-traffic portals

  • Aggressive page cache, purged on publication
  • Images served by the CDN
  • Web servers added on demand
  • Protection against viral peaks
On quoteRequest a quote

Application

SaaS, APIs, business applications

  • Orchestrated Docker containers, zero-downtime deployments
  • Replicated database, queues
  • Centralised logs and metrics
  • Staging environments identical to production
On quoteRequest a quote

Frequently asked questions

How long to set up an architecture?

Two to six weeks depending on the application: audit, design, deployment, rehearsed migration, load tests, cut-over. We give a schedule with the quote.

Does my application have to change?

Sometimes a little: sessions in Redis rather than on disk, uploads on shared storage, HTTP cache enabled. We list these points at the audit and help your developers.

What happens during an incident?

On-call acts within the contractual delay, you are kept informed in the portal and on the channels you chose, and you get a written post-mortem: cause, duration, measures.

A peak ahead?

Black Friday, campaign, launch, back-to-school: tell us the date, we prepare the architecture and test it beforehand.

Request a quote Managed services