How has the recent turmoil within the OpenAI offices changed your plans to use GPT in a business process or product in 2024?
Increased uncertainty means we are more likely to evaluate alternative AI chatbots and LLMs.
No change in plans, though we will keep an eye on the situation.
With Sam Altman back in charge, we are more likely to go all-in with GPT and LLMs.
What recent turmoil?

Tip for Installing a Service Mesh: Start with the Pain Points

Jun 10th, 2019 3:00pm by
Featued image for: Tip for Installing a Service Mesh: Start with the Pain Points

Raygun sponsored this podcast.

Tips for Installing a Service Mesh: Start with the Pain

On this episode of The New Stack Makers podcast, we talk with Zack Butcher, who has long been involved in and advocating for cloud engineering. He’s a founding engineer at Tetrate, and was an engineer on the Istio team at Google. We caught up with Butcher at the recent Service Mesh Days, where he led a four-hour training on installing and using Envoy, Istio and service meshes.

The reason for choosing a service mesh is not connected to how many services you run, or how complicated your system is, Butcher said. It’s for when you can’t understand what is going wrong with your distributed system. So pick it up not because it’s cool, but because it will solve an actual pain point in regards to visibility, reliability, or security.

According to Butcher, here are the steps to take:

  1. Figure out where exactly you are going to deploy and why.  
  2. Investigate the implementation itself.  Who is going to be managing it? Who will be using it? 
  3. Start to plan the subset you going to expose the service mesh to your developers. A common mistake is giving all the complexity to their developers at once. “The organizations that most successfully adopt it are the ones that curate the way in which their developers interact with the system,” he said. “They provide small pieces at first.” So for the most successful adoption, limit the scope of the initial launch.
  4. Start to familiarize yourself with the data plane. Most companies are using Envoy for this, and the lowest risk place is at ingress. “Let it run there,” he said, “learn how it runs, how your team processes it, build processes, then when it’s time to move it in to do a sidecar deployment. You’ll have much more confidence because you’ve been operating the control plane, you’ve been operating the data plane so you understand the knobs and failure points.” And start planning how you are to deploy to your developers.
  5. Once you’re confident with the data plane and want to deploy a mesh, pick the highest pain point of the three pillars: security, traffic management, or visibility, and start there. Use the service mesh to solve that, gain confidence on how to use it and operate it, then move on to your next pain point.

Take a listen to this podcast to find out more about if it’s really as complex as it looks, how service mesh helps out with security, and what to do once you get the basics installed, and how canaries combined with service mesh.

In this Edition:

2:51: At what point does a company need service mesh?
7:24: How do you decide that you’re ready to install a service mesh?
14:27: If I’m an engineer, and I’m on board with service mesh, what sorts of things do I need to look at and be aware of?
19:27: There are these three pillars and pain points, you pick the one that is hurting the most and start with that. And then what?
22:12: Can you talk a bit about canary deployment?
25:51: Where do people go for more information?

Image by enriquelopezgarre from Pixabay.

Group Created with Sketch.
TNS owner Insight Partners is an investor in: The New Stack.
THE NEW STACK UPDATE A newsletter digest of the week’s most important stories & analyses.