If you’ve ever experienced downtime during sudden traffic spikes, then have no fear: Auto-scaling is here.
As of today, all Dedicated plans on Platform.sh that have subscribed to the Observability Suite package now benefit from auto-scaling out-of-the-box.
How does auto-scaling work?
Let's start by talking about how this new feature works.
Our Dedicated production environments run on a triple-redundant cluster intended for reliability and stability. This architecture also lets us scale the underlying machines vertically without any downtime—in minutes!
Historically, our Support and Operations teams have been responsible for monitoring site uptime on Dedicated projects and manually triggering upsizes in response to slowdowns.
Now, with auto-scaling—alongside our usual Support/Operations team monitoring—our orchestration system automatically detects site timeouts and immediately triggers an upsize. Thanks to this automation, we can now scale Dedicated clusters by doubling resources as soon as an issue is detected, without waiting for a person to intervene.
What about Horizontal Scaling? Can Platform.sh do that?
Yes, but there is a prerequisite. Your Dedicated cluster needs to be configured in a split architecture:
- Three machines will host all stateful services, including databases and indexes. This layer may only scale vertically.
- Three or more machines will host the stateless applications, including your workers. This layer may scale horizontally.
With that setup in place, auto-scaling will scale the stateful layer vertically and the stateless layer horizontally, adding additional instances to increase capacity faster.
Reach out to your account manager to learn more about split architecture, and how it can help your application.
What are the benefits of auto-scaling?
There are several benefits, including:
- We can now scale your production environment faster when an unplanned traffic surge occurs. You can rely on auto-scaling to handle any situation where a resource increase is needed.
- With a well-configured Fastly service actively caching non-personalized content, this mechanism is transparent to the end-users.
- Using fewer resources, and only scaling when necessary, means keeping your costs under control and reducing the carbon emissions generated by your applications.
Why is this included in the Observability Suite?
The Observability suite from Platform.sh is the all-included package for performance management. With it you can:
- Monitor, profile, and test your application code even before it’s released in production.
- Get actionable insights to improve your code rather than spend time figuring out what's wrong.
- Ensure optimal performance and user experience for your web applications.
Plus, the bundled APM (Application Performance Monitoring) always monitors all your application's requests and responses, and will surface any abnormalities or performance issues either at the code or infrastructure level.
Gathering these insights will allow us to improve auto-scaling even more in the future.
For more technical information, please refer to the documentation, which has been updated to reflect the addition of the auto-scaling feature.
And if you want to enable auto-scaling on your Dedicated cluster, or migrate a project on Platform.sh, please reach out to our team.
How do I get the Observability Suite?
Please reach out to our sales team to see how you can upgrade to the Observability Suite.