TNS
VOXPOP
Where are you using WebAssembly?
Wasm promises to let developers build once and run anywhere. Are you using it yet?
At work, for production apps
0%
At work, but not for production apps
0%
I don’t use WebAssembly but expect to when the technology matures
0%
I have no plans to use WebAssembly
0%
No plans and I get mad whenever I see the buzzword
0%
Service Mesh

Istio Advisor Plus GPT: Expert System Meets AI for Service Mesh

This combination of AI with Istio's documents is very handy for working out what's what with this popular open source service mesh.
Dec 14th, 2023 12:15pm by
Featued image for: Istio Advisor Plus GPT: Expert System Meets AI for Service Mesh
Clip art via Pixabay, CNCF.

Remember expert systems? Computer scientist Edward Feigenbaum is credited with developing the concept in the 1960s. The idea was by putting a rules-based system, typically written in the first Artificial Intelligence (AI) language, Lisp, on top of a knowledge base, you could create a system that could answer questions only an expert could answer.

While some systems, such as Mycin, an early attempt to match drugs to a diagnosis, worked well enough, they proved too slow and expensive to be successful.

That was then. This is now.

Today, generative AI programs such as OpenAI’s ChatGPT are fast and cheap. This gave Tetrate, the service mesh and networking company, the idea to create an expert system based on ChatGPT: Istio Advisor Plus for the Istio service mesh.

This open-source service mesh helps you to run distributed, microservices-based apps. Istio is frequently used with Kubernetes to establish a programmable, application-aware network in conjunction with the Envoy service proxy.

Istio Advisor Plus uses a Python frontend, the OpenAI APIs, and data from Tetrate and other cloud native documentation to present an Istio-savvy expert system. That last part is vital because, left to its own devices, generative AI systems have a bad habit of hallucinating. Or, as I prefer to think of it, making things up.

Armed with real information, though, Istio Advisor Plus can, drumroll please,

  1. Explain Istio Concepts and Features: It provides detailed explanations of Istio’s core features, including traffic management, security, observability, and how these concepts can be applied within your service mesh.
  2. GuideIstio Configuration: Need advice on configuring Istio components such as gateways, virtual services, and destination rules? Istio Advisor GPT offers tailored guidance to suit your specific use cases.
  3. Help troubleshoot Istio Issues: When encountering challenges with Istio, this GPT model can help diagnose problems and provide solutions. This includes interpreting error messages, reviewing configuration files, and recommending best practices.
  4. Performance Optimization Tips: Get insights and strategies for optimizing the performance of your Istio service mesh, covering aspects like scaling, tuning, and efficient resource utilization.
  5. Security Best Practices: Ensure your service mesh is secure with guidance on implementing zero trust security principles within Istio. This includes mutual TLS, authorization policies, and securing ingress/egress configurations.
  6. Visual Representation of Processes: Istio Advisor GPT can create visual diagrams using Mermaid code to represent complex networking or service mesh processes, making them easier to grasp.
  7. Assistance with Istio Upgrades: Planning an Istio upgrade? This GPT model provides valuable guidance, including addressing compatibility issues and leveraging new features effectively.
  8. Bug Reporting Guidance: If you encounter a potential bug in Istio, Istio Advisor GPT can assist you in compiling a detailed bug report, ensuring you provide all the necessary details for effective troubleshooting. This can be a Big help.
  9. Referring to Relevant Documentation: For complex queries, this GPT model can point you to specific sections of the Istio documentation or other relevant resources for deeper understanding.
  10. Advice on Istio Ecosystem Tools: Gain insights into the tools and integrations within the Istio ecosystem, such as Prometheus, Grafana, and Jaeger, and discover how to use them effectively.

Of course, you can find all this and more by going through the documentation. But, instead of searching through documents, FAQs, and forum messages, which may or may not be up to date, you can have the program do it for you.

It can also answer more sophisticated questions.

For example, when I asked, “How do you use Istio and Envoy together to pull data from an Nginx database?” It replied that I “should configure Envoy as the sidecar proxy for a service in the Istio service mesh that communicates with the Nginx database.”

It also gave me an example of a Kubernetes service configuration and an Istio Virtual Service for my hypothetical Nginx database.

It then went into more detail as to how, in an Istio-enabled Kubernetes cluster, every pod of my service that needs to communicate with the Nginx database must have an Envoy sidecar proxy automatically injected. And, that, I’d need to define an Istio Virtual Service and a Destination Rule for the Nginx service, secure my communications lines with mutual TLS (mTLS), and be certain to use Istio and Envoy’s telemetry features to monitor and log the traffic.

Now, that was a simple question, but the answer was accurate, to the point, and I didn’t need to dig through the documents to find it. No, it’s not writing code for me, but it’s helping me understand what’s going on to give me an intelligent grip on the situation.

Unlike generic GPT-4, it also gave me references for its answers. Albeit they didn’t take me to specific spots in the documentation, but to the sources it used to provide me with an answer. Still, I really appreciated that.

Sounds interesting? Give it a try. You’ll need a ChatGPT Plus subscription, which gives you access to the latest version of the OpenAI Large Language Model (LLM), GPT-4.

Frankly, I’m very impressed. I can seriously see many other projects creating similar expert systems for their programs. Let’s face it, cloud native computing is complex and hard to understand. There are so many moving parts. Anything that can help us quickly and more easily come to terms with the technology is a good thing in my book.

Group Created with Sketch.
THE NEW STACK UPDATE A newsletter digest of the week’s most important stories & analyses.