Service Mesh Recommendations

Determine if you are Ready for a Service Mesh, Need One, and Best Practices for Tackling One

Service Mesh Recommendations

Determine if you are ready for a service mesh, need one, and best practices for tackling one

Pre-requisites include: container ecosystem, skill development and exploration, a business case.
Technology best practices: determine architecture and support model, evaluate tool usability.
People best practices: PlatformOps team and other role and responsibility adjustments.

Consider a Service Mesh

For IT leaders who are considering a service mesh

Organizations shouldn’t jump into a service mesh project if they are just getting started with containers and microservices.

There is no hard and fast rule on the number of containers you need to have in production or the number of microservices you should have before a service mesh makes sense. Reliability, scalability, and observability challenges will emerge as your application and service portfolio grows.

Here are some signs that your roadmap should include a service mesh.

Ready

You have an established container ecosystem. A well-managed container system and container-as-a-service environment are needed before you tackle a service mesh. Immature DevOps practices, monolithic applications, and technical debt can get in the way of realizing the full value of a service mesh.

You are investing in upskilling. Service mesh has evolved a lot in the past few years, so your staff should be up to speed on the latest technology. If they don’t have hands-on experience with it, plan for that time investment.

You are comfortable with open-source emerging technology.

You can build a business case. Do not buy into the hype: write your specific story about value, cost, and risk.

Emerging needs

You are attempting to implement custom-built distributed tracing using a patchwork of vendor-specific solutions. A service mesh will help you implement vendor-agnostic distributed tracing.

You are spending a lot of time implementing security policies. A service mesh will bake “hooks” into the services layer to make it easier to incorporate security policies.

Best Practices for Service Mesh

For IT leaders who are getting started with a service mesh

These are some of the best practices for including a service mesh in your IT infrastructure.

People

Invest in your platform team. You may need to beef up your platform team skills or staffing. A Platform Ops team usually manages the service mesh and other “developer self-service” aspects so developers can focus on delivering business functionality.

Revisit roles and responsibilities. Since a service mesh is part of a new model with clearer separation between developers who create applications and operators who run the infrastructure, some DevOps roles may need to be tweaked to give developers access to some service mesh features, while shielding them from other service mesh concerns.

Plan for operational changes. The value of being able to observe performance and troubleshoot problems may set higher expectations around continuous access to data. Both IT staff and product managers may come to expect well-designed dashboards and instant updates on system activity.

Technology

Determine architecture. Determine if you want to start using service mesh through/with container management offerings vs. deploying it separately. A service mesh is often incorporated into container orchestration efforts using Kubernetes or other tools.

Prioritize usability of the tools. Evaluate the many products and tools that support service mesh, since they are evolving rapidly. Look beyond feature lists to evaluate the ease of use for your IT administrators.

Determine support model. You may need to plan for some staff involvement in the various community-driven, open-source initiatives that are driving service mesh innovation. Or you may opt for commercial vendor support.

Review your security policies. Since security can be vastly improved with a service mesh, you should understand your current container security objectives and include them in your service mesh roadmap.