Web Servers, Hosting and DevOps

Load Balancing and Highly Available Web

A single server is a single point of failure. For web services where downtime is expensive, we design and build architectures that distribute load across multiple application servers, automatically take unhealthy servers out of rotation, replicate the database and can fail over to a second location when needed.

RTO / RPO

Nginx / HAProxyHealthcheckKeepalived / VRRPDB replicationShared storageDNS failoverDisaster recovery testing

Service definition

What is load balancing and high availability, and who is it for?

Load balancing distributes incoming requests across multiple servers; capacity increases, and service continues if a single server fails. High availability goes further: the load balancer, database and file layer also stop being single points of failure, and how quickly (RTO) and with how much data loss (RPO) you’ll recover from an outage is written down in advance.

01

Is Cloudflare load balancing enough?

It’s a good layer for distributing traffic across servers and health checking. But it doesn’t solve redundancy for the database and file layer; that’s set up separately.

02

Can WordPress be highly available?

Yes: provided there are shared uploads, an object cache, database replication and consistent deployments.

03

Is 100% availability possible?

No. No architecture can guarantee it; targets must be measurable and tested with drills.

Project stages

How do we proceed?

01 / STAGE

Analysis

Targets and single points of failure.

02 / STAGE

Design

Architecture and cost.

03 / STAGE

Setup

Layers and tests.

04 / STAGE

Exercise

Failure and failover tests.

STAGE 05

Operation

Monitoring and review.

Corporate Web

Don’t leave services where downtime is expensive dependent on a single server

We discuss your services’ downtime tolerance, traffic profile and budget, and present architecture options with their costs based on RTO/RPO targets.