The Microservices and Cloud Hangover
Everyone wants feature velocity but no one wants to increase investments in Devops tooling, SRE and other associated needs. Headed for a rude awakening. Therein lies the opportunity!
Microservices, as we know, are a great way to speed up time to value (feature to production), in the way that (in theory) to get a new feature out, only one or a few microservices need to be updated, vs a whole monolithic codebase. Each micro service can basically be updated, deployed, and scaled independently. It further (in theory) helps code be more efficient, as many new applications or services can reuse micro services already built - less code duplication, less code to maintain. But what is the aftermath of a microservices architecture approach?
Same with the cloud trend - be it public, private, hybrid or multi. This container-based, on demand architecture helps companies (in theory) to save resources. The services don’t always have to be up, workloads can take turns using the infrastructure, etc. We get it. But what are the hidden costs of that flexibility?
Well, one thing is that both these trends come with more surface areas:
- More APIs
- More dependencies
- More components involved
- More combinations needed to be tested, secured, and monitored
- More moving pieces that needs to be coordinated across release projects
Which leads to increased quality, security, and resilience risks, which in turn translates into increased SLA violation risks. And it also clouds the visibility across projects and releases.
If you exponentially invest in API/microservices security, Devops, Cloud Specialists (Cloudops?), and SRE, sure, you may truly gain the benefits and cost efficiencies of microservices and cloud trends. But do organizations have QA and SRE teams built out as a top line item on their budget sheets today? Probably not. And where is API security on the CISO’s list?
It is surprising that in spite of the digitalization trend, that makes businesses more dependent on online customer experience for continued growth and loyalty, investments in SRE and Devops have stayed flat. Almost opposite direction: companies have strived for further engineering efficiency and made overworked engineering, support, SRE, and devops teams even thinner.
As always, where there is pain, there is opportunity:
- I predict a rise in SRE and Devops tooling and as-a-service, efficiency/visibility products for cloud spend and cross-environment project management
- I predict a rise in API security
- I predict a rise in QA automation - using ML to create tests
- I predict a rise in cloud deployment and CI/CD services replacing the costly growth of cloud specialists in Devops teams
- I predict a rise in API and Integration quality as a service
It is just a natural evolution of the hangover from the Cloud and Microservices hype. If you are working on something exciting to address these pains, feel free to ping me.
Attended QCon and this talk was interesting and in the same area https://www.infoq.com/presentations/microservices-monolith-antipatterns/
Sounds like a conference talk :-)
Like many other trends preceding microservices, it is often seen as a panacea by those outside of the inner circles of IT. Everyone wants the pain to go away, but few peel it back to find the root causes. It’s not uncommon to hear questions of puzzlement such as “we invested in microservices, so why are we not moving faster?” Frankly, it boils down to a change in philosophy and mindset, not necessarily an embrace of microservices. Invest in the systems and processes and thinking that we know accelerate development and protect important architectural properties. Modularity, reduction in coupling, high cohesion, the Unix philosophy. If we do that, microservices tend to emerge naturally. We have to connect those dots for CxOs in order to understand the importance of making the proper investments and behavioral changes.
I predict intelligent infrastructure #smartcomputing #noOps
Hear hear!