What ISOGrid is
A cloud platform built in Algeria, for teams who want to ship without running servers.
ISOGrid runs your software. You bring an application; the platform builds it, places it in a region, gives it an address, keeps it running, and bills you for what it uses. You do not provision machines, patch them, or wake up when one fails.
This page is the shape of the whole system. Everything else in this guide is a detail of one part of it.
The pieces you work with
An organization is who owns things and who pays. Every application, database, store and domain belongs to exactly one. People are invited into an organization with a role that decides what they may do, and one person can belong to several — a consultancy might be a member of a client's organization and of its own.
A region is a place your work runs. Regions are physically separate: a different building, a different power feed, a different route to the internet. When you deploy, you choose one. What a region costs and what sizes it offers are set per region, so a cheaper region really is cheaper rather than the same price with worse hardware.
An application is one running program with an address. It has a size, a number of copies, an environment, and optionally a domain of your own. Most applications are built from a code repository; you can also bring one you have already packaged.
A database, a store, a managed service are the stateful things an application talks to. They are ordered separately from applications because they outlive them: you will redeploy an application a hundred times against the same database.
Credit is what everything is paid from. You top up, consumption draws down, and the platform tells you before you run out rather than after.
What happens when you deploy
- You point ISOGrid at a repository, or upload something already packaged.
- The platform works out how to build it. If it cannot, it asks you — it does not guess and produce something that fails at three in the morning.
- It builds, privately. Your source and the result are visible only to your organization.
- It places the result in the region you chose and starts the number of copies you asked for.
- It gives you an address that works immediately, over an encrypted connection, with a certificate the platform obtains and renews.
- It watches the copies. One that stops is replaced without anyone being paged.
Steps two through six are the product. Step one is the only part that is your job.
What the platform guarantees, and what it does not
It keeps your application running as long as the region is up and the application can actually start. A program that crashes on boot will be restarted, fail again, and be reported to you — the platform cannot fix code.
It keeps your data. Databases and stores are written to more than one machine where you have asked for that. See Databases for what each arrangement survives and what it does not.
It does not back anything up unless you ask it to. Copies of your data on
several machines protect you from a machine dying. They do not protect you from
a mistaken DELETE, which is copied to every machine as faithfully as anything
else. Backups are a separate thing you turn on, and you should.
It does not read your data. Platform staff can see that your database exists, how large it is and whether it is healthy. Reading what is in it requires your credentials, which the platform hands to you once and does not keep in a form anyone can retrieve.
Where to go next
- Getting started — an account to a running application.
- Deploying an application — the deployment model in detail.
- Addresses and domains — how people reach what you deployed.
- When things go wrong — the failures you are most likely to meet, and what to do about each.