Part of the Cloud Computing Puzzle

Platform as a Service

Subscribe to Platform as a Service: eMailAlertsEmail Alerts newslettersWeekly Newsletters
Get Platform as a Service: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn


paas Authors: Liz McMillan, Stackify Blog, Pat Romanski, Mehdi Daoudi, Elizabeth White

Related Topics: ActiveState Journal, Platform as a Service, DevOps Journal

Blog Post

PaaS and #Microservices By @bcferrycoder | @DevOpsSummit #DevOps #PaaS

Part 5 - Advice for individuals, teams, and organizations considering a move to microservices

In the first four parts of this series I presented an introduction to microservices along with a handful of emerging microservices patterns, and a discussion of some of the downsides and challenges to using microservices. The most recent installment of this series looked at ten ways that PaaS facilitates microservices development and adoption.

In this post I'll cover some words of wisdom, advice intended for individuals, teams, and organizations considering a move to microservices. I've gleaned this advice from the various articles and meetups mentioned above, from dozens of discussions with architects and developers at a handful of organizations who are moving forward with microservices, as well as from my experience building a couple microservices-based systems before microservices was a thing.

Words to the Wise / Words to the Weary
Start small, move fast
Don't expect to "pull a Netflix." What I mean by this is that you throw millions of dollars and buildings - full of smart engineers and architects, with buy-in from the top of the organization, at a rapid corporate-wide metamorphoses from monolithic apps to microservices apps. That's just not realistic for most organizations today. Dramatic changes to the org-chart are almost impossible. Change needs to be small, with minimal investment, and rapid results. With enough measurable success, the slow transition to a microservices-based approach is more within reach.

Choose the right project
Ideally, for your first foray into microservices, choose a project with some of these characteristics:

  • Short-lived
  • Internal
  • Non-production
  • Non-mission critical
  • Small
  • Greenfield, if possible

By focusing on these types of projects you minimize risk while you get the tooling, the process (more accurately, lack of process), and knowledge in place as a foundation for further microservices efforts.

Consider hackathons and lab projects
Taking the lead from Mozilla, consider lab projects as the first steps into microservices. Hackathons and developer contests are particularly well suited. These are short-lived, with participation from passionate and energetic early-adopter-type developers who like to experiment.

Eliminate IT barriers
I can't emphasize this enough: Go to any lengths to eliminate any IT barriers currently impeding app development. Do not, I repeat, DO NOT force a developer to submit a service ticket to provision a resource (VM, data service...). Doing so defeats many of the advantages of microservices and maintains the destructive hierarchy that commonly exists between developers and IT.

However, many organizations might find that disrupting their current ticket-based process to obtain resources is too much of a stretch. If so, find another way. Use an external cloud, such as AWS or Rackspace or HP Helion Public Cloud, for early microservices development. Or, if that's not permissible, build up three Linux PCs and put them under someone's desk. I'm serious here - with today's modern tooling (read PaaS) it's trivial to build a fully functional cloud out of a handful of PCs and with the same tooling; it's equally trivial to move a complex microservices-based application from an under-the-desk cluster to an enterprise data center once the app is ready for prime-time. The point is, go to almost any lengths to make sure there's no ticketing system or other process to interfere.

Allow devs to push to production
A developer who can push to prod is often empowered, and incented to produce higher quality code. This has countless advantages, probably the most significant of which is improved employee retention.

Be ready to throw it out
Do not overthink the early microservices you build. You can easily get mired in the weeds, designing naming schemes, service discovery mechanisms, and inter-service communication protocols. Instead of all that, just roll up your sleeves and dive right in. Build a handful of small microservices and be prepared to rewrite them. One major advantage of building isolated independent microservices is that it can be rewritten rapidly. Take advantage of this.

Identify internal proponents early
Not everybody will be on board with microservices from the start. But some will, and it's helpful to identify these folks as they'll be invaluable for a wider drive to microservices in the future.

Quantify success
The success of a technology change is meaningless and wasted if the benefits are not measurable or measured. Identify a handful of metrics that you can quantify, metrics that microservices will improve. Here are some examples:

  • How long does it take to push out a new feature?
  • How many service desk tickets are required to build a new app?
  • How many days/weeks does it take to onboard a new developer?

Come up with your own, and track them, with the goal of demonstrating the improvements that microservices bring to the table.

Scale from day one
Scaling is an important motivator for moving to cloud computing, and you should be able to scale microservices effortlessly. Make sure the tooling is in place to instantly, automatically, and elastically scale your microservices in response to system load.

Monitor from day one
Similarly, it's important to monitor the overall operation of the system and each microservice that's part of the system. Wire in monitoring tools to each microservice you build before you even write a single line of code for the microservice. Monitoring should not incur additional design and implementation overhead - it must be built into the platform.

Automate everywhere
As discussed previously, eliminate, as much as possible, any need for human intervention for a microservices operation. Almost everything should be automated. This includes monitoring, scaling, logging, rollback, failure detection, testing, and deployment. Again, the right tooling is essential here.

Explore toolsets, understand patterns
The use of microservices is evolving fast, but some established patterns are emerging now, and will continue to emerge. Keep your eye on these, and experiment with them. Learn about ways to support graceful degradation and fault tolerance. Get a handle on the circuit breaker pattern, and explore the Netflix tools that they have generously made freely available. Tools like Chaos Monkey and Hystrix are worth looking at, and possibly adopting. Additionally, it's helpful to understand containerization, and especially recognize the limitations of containers for cloud application development.

And definitely, get a solid grasp of Platform-as-a-Service, the essential foundation for serious and successful microservices deployment.

Expect to fail
Failure is okay. At least, with a small targeted non-mission-critical microservice, failure is okay (and that's kind of the whole point). Learn from the failures, keep calm, and carry on.

Step away from the infrastructure
Do not build anything that's tied to a particular infrastructure. For example, if you do decide to deploy a microservice to AWS, don't hard-wire the microservice to use AWS APIs and facilities. Instead write to an abstraction layer (read PaaS) so that the microservice can be moved to any cloud at any time in the development and delivery cycle.

Just as important, don't get mired in the plumbing. Get out of the IaaS weeds. Allow the PaaS to take care of the underlying details so you can focus on what's important: the functionality, the interface, and the service level agreements.

Determine where you're using Microservices today
You might already be using microservices, at least as a consumer. Do you use Google's geocoding API to convert addresses to locations, or Twilio to send text messages, or Weather Underground to determine weather conditions at your customer location? These external services can be considered microservices, in the sense that they're isolated and loosely coupled from your application codebase, and they can evolve independently.

By looking at places you're using external microservices, you will be able to identify functionality of your existing internal applications that can be split off into microservices, leveraging all the power of microservices in a small, contained service. Again, take small steps.

Use the right PaaS for the job
Look carefully at the license agreements for the tools and services you adopt. Some PaaS offerings, for example, base their licensing costs on number of applications (microservices) deployed to the PaaS. This model is an impediment to a successful microservices adoption since a penalty is incurred for each additional service that's deployed. Others, like Stackato, use a memory model where license fees (if any) are charged based on the overall amount of memory making up a PaaS cluster. The memory pricing model makes much more sense for a microservices deployment: you can construct your application from as many microservices as you like, without incurring additional cost for each individual microservice, other than the memory it consumes.

Get started now
I recommend you get your feet wet now, by downloading the latest Stackato Micro Cloud from ActiveState, deploying a simple service, and discovering the many features of PaaS that ease the journey from a monolithic to microservices approach.

The post Microservices and PaaS - Part V appeared first on ActiveState.

More Stories By John Wetherill

Originally from Canada, John has spent much of his career designing and building software at a handful of startups, at Sun Microsystems, NeXT Inc., and more recently in the smart grid and energy space. His biggest passion is for developer tools, or more generally any tool, language, process, or system that improves developer productivity and quality of life. Without question, Stackato is one such tool and the reason why he is here. No stranger to technology evangelism, John spent several years in the late 1990's on Sun's Technology Evangelism Team spreading the Java Gospel across the globe and focusing on the prolific number of Java technologies. Now John is now returning to his roots, as a technology evangelist working for a Canadian company, albeit remotely from Santa Cruz.

Comments (0)

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.