The technology underneath

This page is for technical colleagues who want to know what a LobLogix application is built on. Nobody needs any of it to describe, agree, build or try an application. LobLogix is a backend-producing platform as a service, or BaPaaS: it produces a backend that you run and can move.

In short

• Language and runtime: Elixir, on the Erlang virtual machine.
• Records: PostgreSQL.
• Connections: JSON:API, described in an OpenAPI document, and GraphQL, with its own schema.
• Observability: an engine of your own for logs, measurements and traces.
• Where it runs: Preview and Stage on the LobLogix platform. Production, on the business tiers, on Google Cloud, on any other provider’s virtual machines, or on your own bare-metal servers.
• High availability and backups: a three-way database cluster, a hot standby cluster and point-in-time backups, on the business tiers.
• Sovereignty: in production your records stay on machines you control. By default LobLogix cannot access them.

The language and its runtime

Your application is written in Elixir and runs on the Erlang virtual machine, which was built for telephone networks that must not stop and has been in service for more than three decades.

What that gives a business application:

• Many things at once. Each request runs as its own lightweight process, so a slow one does not hold up the others.
• Failures stay small. A part that fails is restarted on its own, without taking the rest down with it.
• Made to run as a cluster. Several copies of an application can work as one, which is the basis of high availability.

You do not need to know Elixir to use LobLogix.

Your records: PostgreSQL

Records are kept in PostgreSQL, a widely used open-source database. In production the database runs on your machines. It is standard PostgreSQL, so the usual PostgreSQL tools can read it, back it up and move it.

How screens and other tools connect

Through JSON:API, which an OpenAPI document describes, and GraphQL, which describes itself through its own schema. See What a developer receives from LobLogix.

An observability engine of your own

Logs, measurements and traces are collected by an engine that runs beside your application. In Preview and Stage, the LobLogix platform collects them for you. In production the engine runs on your machines, so what it collects stays with you. No outside monitoring service is needed.

The Deployment page reads from it. See Health and logs on the Deployment page.

Where it runs

• Preview and Stage run on the LobLogix platform. There is nothing to provide.
• Production, on the business tiers, runs on machines your company controls: Google Cloud, any other provider’s virtual machines, or your own bare-metal servers.

LobLogix installs its own Kubernetes distribution on those machines. A cloud provider’s managed Kubernetes service is not needed and is not used. From a provider you need only machines, a fixed network address and storage for backups.

Production comes after the alpha. See Going live on your own infrastructure: how it is designed.

High availability, on the business tiers

• A three-way database cluster. Three copies of your database kept in step, so one can fail without losing a confirmed change.
• A hot standby cluster. A second running copy of the installation, kept up to date continuously. It can sit with a second provider.
• Several copies of the application, so one can stop without stopping the service.

Backups, on the business tiers

Continuous backups to S3-compatible storage that you control. They can be restored to a chosen moment, which is called point-in-time recovery.

Both are part of production. Preview and Stage run as single copies with no backup. They are capabilities of the design, not an uptime figure.

Your data stays yours

In production the database, the logs and the backups sit on your machines. By default LobLogix cannot access your production database and holds no password or key to it. The AI builds from your specification, not from your records.

Portability

• Between providers. Nothing depends on one provider’s own services, so an installation can be created again with another provider and its records restored from backup.
• Your records. Standard PostgreSQL.
• Your frontend. It connects through open standards, so it is not tied to LobLogix.
• Your code. After the alpha, on the business tiers, you can download a copy of the code LobLogix generated. See If you pause your subscription.
• The AI. Your specification and your checks do not depend on one AI model.

What this page does not promise

No uptime figure and no certification. Production, backups and high availability come after the alpha, on the business tiers. See Security and compliance, plainly.