At Shoreline.io, I've seen firsthand the critical nature of maintaining robust and resilient cloud services. Our platform, vital for incident automation, is a tier zero service. This means we must remain operational despite the chaos of other systems failing. That begs the question: How does Shoreline do it for itself?
Fault Tolerance and Dependency Minimization
In this video I talk about how, at Shoreline, our philosophy is straightforward: be prepared for any single point of failure. This could range from an availability zone to a node, a process, or even an entire region needing a restart.
This is a challenging approach but the foundation of our resiliency. We also strive to minimize our dependencies. Every external service we rely on is a potential failure point. Limiting these dependencies enhances our resilience, ensuring we can degrade gracefully rather than collapse in the face of adversity.
Why Shoreline Uses SQLite
A prime example of our unconventional yet effective strategies is our use of SQLite for database management. Most of us probably think of SQLite as a database for our cell phones.
But we use that because it allows us to:
- partition the data
- limit the blast radius of issues
- ensure that data is copied in at least two locations before being deleted
- time partition our queries across a multitude of SQLite files, rather than expecting a single database to be up, (because, like everything else, they fail).
This approach might seem unconventional, but these decisions fortify our infrastructure against large-scale outages.
Our Commitment to Reliability
I've witnessed the impact of significant outages in my time at AWS and other platforms. Through our diligent practices and innovative approaches, Shoreline will continue to provide reliable service without facing such disruptions.
Learn more about why I started Shoreline.