#30: ChatGPT for platform engineering

#30: ChatGPT for platform engineering

Hey there,

Can AI be the bacon ?? to your platform BLT ???

Let’s get bakin’


ChatGPT for platform engineering

ChatGPT and other AI tools will be a game changer for debugging and troubleshooting infrastructure. The days of slogging through logs, Googling error messages, or perusing manuals for fixes will soon be behind us. ChatGPT will be able to do all of it for us.?

Take k8sgpt, for example. It advertises itself as a tool that diagnoses and triages your K8s clusters in simple English. While I doubt this tool will be able to replace real, human SRE expertise, it can still help teams save time and focus their efforts on more complex problems.?

I can already see a future where AI can fix problems in real time, reducing the need for engineers to be on-call. And who would complain about that? Or, perhaps engineers can talk to their platforms like a sci-fi captain talks to his mothership: “Run me diagnostics.” And then it’s done.?

Let me be very clear: AI is not human, and it won’t be anywhere near sentient for a long time. In the meantime, we can leverage machines to lift repetitive work and toil off our shoulders. ChatGPT isn’t creative, it just formulated solutions based on existing information it can find. It’s impressive (and incredibly useful), but it isn’t human.?

Regardless, the future of platform engineering is bright with AI. Imagine using human language to get AI to do the following:


  • Templates based on your requirements. No more looking for YAML online.
  • Templates based on another service.
  • Diagnostics and statistics:?

“How many requests did we get from Canada in our Load Balancer?”?

“Where is this DDoS attack likely coming from?”

“Give me a full post-mortem of the issue that happened yesterday.”


  • Platform Engineers configuring resources:

“Give me the configuration for a resource definition for an AWS bucket.”

Or “Create it for me”

  • Recommendations to upsize or downsize infrastructure based on traffic.
  • Cost recommendations for your cloud providers based on your usage.


I’m really excited about the potential for AI to boost platform engineering. But what are your thoughts? Let me know on Twitter @luca_cloud.?

Accelerate your cloud adoption with landing zones

Landing zones can increase the productivity of your cloud adoption strategy. They help teams hit the ground running, collaborate, stay compliant with security best practices, and scale operations.


No alt text provided for this image


When building a Google Cloud landing zone for example, there are a few areas you should focus on:

1?? Resource management. This includes more than just billing and budgets; it also involves your organizational structure. Project liens are an important resource management tool. A project lien blocks projects from being deleted without approval.

2?? Access management. Choose components that allow you to create and manage identity groups, use IAM service accounts, and securely work with external identity providers.

3?? Networking. Your developers should be able to benefit from your network without having to interact with it directly or frequently. Build your landing zone to establish clear firewall rules, enforce logging standards, and harmonize common cloud service configurations.

4?? Observability and post-mortem. Use your landing zone to standardize logging practices, implement asset inventorying, and set up helpful tools like the Google Cloud Security Command Center. Then, your developers won’t have to configure audit logging or other observability functions for every project.

5?? Security and compliance. Implement security practices worth following. Also, consider how your landing zone will comply with frameworks like SOC2, ISO27001, HIPAA, etc.

6?? Shared resources. Your landing zone can streamline interactions with shared resources like K8s clusters or artefact registries. Abstracting away this complexity can help improve the developer experience.

Want to learn more about landing zones? Watch the full webinar.


No alt text provided for this image

?? Does GPT-4 really understand what we’re saying? ?? According to David Krakauer, the answer is yes and no.

?? Speaking of GPT-4, has anyone used it to review and refactor your code?

?? Enabling true developer self-service requires more than just shifting left.?

?? “You build it, you run it” doesn’t scale. ? Thanks to the folks at PurePerformance for having me on their podcast.

?? Another take on the DevOps vs. SRE vs. platform engineering debate ??


No alt text provided for this image


Last but not least, have you joined the Platform Engineering Slack channel? If not, you're missing out. Here are the highlights:?


And that’s a wrap for this week! As always, this newsletter is a community project. So if you have anything awesome to share from the cloud-native world, send it our way.

Stay crunchy ??

Luca

要查看或添加评论,请登录

社区洞察

其他会员也浏览了