Release Notes
March 18, 2024

Flightcrew believes in Autoscaling

By
Tim Nichols

TL;DR: Flightcrew makes configuring HPAs and VPAs self-serve for infra teams and developers*. We think you should scale things up and scale things down to stay moving! 

What can I do with this release?

With a lightweight, 5-minute installation, Flightcrew will:

  1. Profile your workloads to understand how they scale based on traffic patterns

  2. Maintain a live record or autoscaler recommendation options for the workload to handle its traffic, including:

    • Ensuring you hit your SLOs, posture, and policies

    • The impact of these recommendations on your services and infrastructure

    • Plain-english explanations on what these things do (via LLMs on docs, known issues, etc.)

    • Warnings about upgrades and dependency changes

  3. Tell you when you drift from safety and send you a PR 

In summary, you get the benefits of autoscaling (resiliency and savings) without your own time and headspace.

Why did you build this? 

Short-answer: We had built a workload profiler, and then users started asking us for HPA recommendations to save time and reduce cost. Like everything with K8s, autoscaling is complicated and can take a few cycles to get the handle of it.

VPA was the natural next step for certain workloads, we also think it will get a lot more popular when you can resize in place.

Long-term, microservices and platform engineering is pushing more and more workload management to feature teams who are not used to thinking like SREs. Flightcrew simplifies infrastructure for them.

Do I need Flightcrew for this? 

Honestly… it’s a judgment call. You could get some big wins spending a month on basic scaling heuristics without Flightcrew. 

But consider Flightcrew if:

  1. You’ve got complex architecture with dynamic traffic

  2. Setting up HPAs for the first time seems scary or time-intensive

  3. You want another ~10% off cloud costs after you’ve installed HPA

  4. Your SREs don’t have the time to spend tuning configs

  5. Your feature teams don’t have the time to understand or deploy their own autoscalers

So what’s next?

Let us know if you’ve got questions or want to try it out.
2024 is "The Year of KEDA" so stay tuned for future developments…

*We also do this on some Serverless platforms, reach out if you're interested in learning more.

Release Notes
March 18, 2024

Flightcrew believes in Autoscaling

By
Tim Nichols

TL;DR: Flightcrew makes configuring HPAs and VPAs self-serve for infra teams and developers*. We think you should scale things up and scale things down to stay moving! 

What can I do with this release?

With a lightweight, 5-minute installation, Flightcrew will:

  1. Profile your workloads to understand how they scale based on traffic patterns

  2. Maintain a live record or autoscaler recommendation options for the workload to handle its traffic, including:

    • Ensuring you hit your SLOs, posture, and policies

    • The impact of these recommendations on your services and infrastructure

    • Plain-english explanations on what these things do (via LLMs on docs, known issues, etc.)

    • Warnings about upgrades and dependency changes

  3. Tell you when you drift from safety and send you a PR 

In summary, you get the benefits of autoscaling (resiliency and savings) without your own time and headspace.

Why did you build this? 

Short-answer: We had built a workload profiler, and then users started asking us for HPA recommendations to save time and reduce cost. Like everything with K8s, autoscaling is complicated and can take a few cycles to get the handle of it.

VPA was the natural next step for certain workloads, we also think it will get a lot more popular when you can resize in place.

Long-term, microservices and platform engineering is pushing more and more workload management to feature teams who are not used to thinking like SREs. Flightcrew simplifies infrastructure for them.

Do I need Flightcrew for this? 

Honestly… it’s a judgment call. You could get some big wins spending a month on basic scaling heuristics without Flightcrew. 

But consider Flightcrew if:

  1. You’ve got complex architecture with dynamic traffic

  2. Setting up HPAs for the first time seems scary or time-intensive

  3. You want another ~10% off cloud costs after you’ve installed HPA

  4. Your SREs don’t have the time to spend tuning configs

  5. Your feature teams don’t have the time to understand or deploy their own autoscalers

So what’s next?

Let us know if you’ve got questions or want to try it out.
2024 is "The Year of KEDA" so stay tuned for future developments…

*We also do this on some Serverless platforms, reach out if you're interested in learning more.

Release Notes
March 18, 2024

Flightcrew believes in Autoscaling

By
Tim Nichols

TL;DR: Flightcrew makes configuring HPAs and VPAs self-serve for infra teams and developers*. We think you should scale things up and scale things down to stay moving! 

What can I do with this release?

With a lightweight, 5-minute installation, Flightcrew will:

  1. Profile your workloads to understand how they scale based on traffic patterns

  2. Maintain a live record or autoscaler recommendation options for the workload to handle its traffic, including:

    • Ensuring you hit your SLOs, posture, and policies

    • The impact of these recommendations on your services and infrastructure

    • Plain-english explanations on what these things do (via LLMs on docs, known issues, etc.)

    • Warnings about upgrades and dependency changes

  3. Tell you when you drift from safety and send you a PR 

In summary, you get the benefits of autoscaling (resiliency and savings) without your own time and headspace.

Why did you build this? 

Short-answer: We had built a workload profiler, and then users started asking us for HPA recommendations to save time and reduce cost. Like everything with K8s, autoscaling is complicated and can take a few cycles to get the handle of it.

VPA was the natural next step for certain workloads, we also think it will get a lot more popular when you can resize in place.

Long-term, microservices and platform engineering is pushing more and more workload management to feature teams who are not used to thinking like SREs. Flightcrew simplifies infrastructure for them.

Do I need Flightcrew for this? 

Honestly… it’s a judgment call. You could get some big wins spending a month on basic scaling heuristics without Flightcrew. 

But consider Flightcrew if:

  1. You’ve got complex architecture with dynamic traffic

  2. Setting up HPAs for the first time seems scary or time-intensive

  3. You want another ~10% off cloud costs after you’ve installed HPA

  4. Your SREs don’t have the time to spend tuning configs

  5. Your feature teams don’t have the time to understand or deploy their own autoscalers

So what’s next?

Let us know if you’ve got questions or want to try it out.
2024 is "The Year of KEDA" so stay tuned for future developments…

*We also do this on some Serverless platforms, reach out if you're interested in learning more.

Engineering teams use Flightcrew to make
every pull request more reliable

Engineering teams use Flightcrew to make every pull request more reliable

Engineering teams use Flightcrew to make every pull request more reliable