Infrastructure

[smashwilson]

Before we set up a continous delivery pipeline, we should make some decisions about what we want our long-term production infrastructure to look like. We have a lot of flexibility here.

The general trade-off that we face is between services that will handle more of the day-to-day system administration tasks for us, at the cost of more money and less control, and hosting platforms that

Based on my current understanding of the system architecture, we will need:

• A domain name.
• A TLS certificate.
• A place to host the Express server (compute resources).
• A place to host the database.
• Backup and restoration.
• Some answer for monitoring and log management. (We'll want to know quickly if the system crashes.)

Considerations

Some questions to keep in mind as we decide:

• How much traffic do we plan to support? What is our anticipated user count? Do we anticipate needing to scale horizontally?
• How much data do we need to store? (Rough estimate: on the order of millions of rows a day.) Do we prefer RDMS or something looser like MongoDB?
• What time and expertise do we have access to in terms of system and database administration? Do we have someone who's comfortable perfoming system maintenance, upgrading and applying security patches, installing system software and managing user accounts?
• What kind of uptime do we want to achieve?
• Do any of these services offer special deals for educational or charity projects?
• Do have a need to prefer Microsoft products, given their sponsorship position?

Options

Platforms-as-a-Service

These are services that do everything for you. Once you do the initial account creation and setup, everything you push to master is automatically live. You can usually manage scaling and capacity with sliders on some kind of web dashboard. These also tend to be expensive, especially as you need to scale and for storage capacity.

• Heroku; [:dollar: pricing]
• Zeit's now; [:dollar: pricing]

Host it ourselves

Click a button, get a VM. This is what we're running on now as a testing environment.

Compute:

• Linode
• AWS EC2
• Microsoft Azure Virtual Machines
• Google Cloud Compute Engine

Databases:

• Amazon RDS: PostgreSQL, MySQL, MariaDB, Oracle, SQL Server
• Google Cloud SQL: PostgreSQL, MySQL, SQL Server
• Azure Databases: PostgreSQL, MySQL, MariaDB

[smashwilson]

Oh - and I'm personally available to set up, or help set up, pretty much any of these. I have the least experience on the DBA side, I think (I can get a database running but not tuned and performant...). I also don't think I can reliably commit maintenance or administration time because of $DAYJOB and family obligations 😄 .

[bluek1te]

@jesusinadidas (Grayson) has expressed interest in setting up and possibly maintaining the infrastructure, so he's going to do a bit of the research involved. Hopefully we will be able to get the majority of these services running in this cycle. I'll reach out to Cedrik for the funds.

[bluek1te]

@jesusinadidas Emailed Cedrik, we will most likely try to use as much of the Microsoft stack as we can due to their sponsorship

[jesusinadidas]

@bluek1te Okay. Are there any services we still need to make a decision on, or are we pretty much going Microsoft across the board?

[bluek1te]

Most likely Microsoft across the board; I think they have a free trial for new accounts so you might be able to try deploying some of the apps we currently have onto their cloud

[bluek1te]

Made a separate email to handle all of the accounts for the webservers; if anyone would like the credentials to work on this, please DM me on slack