Because of the criticism and controversy surrounding DevOps, I wrote this article. Let’s look at the confusion in the industry, touch on the interesting question “When will DevOps cease to exist” and discuss when there is no DevOps on the project.
The DevOps approach is shaped by experienced market engineers just like Agile for example. Due to the increasing complexity of projects and competition, it became clear in a practical way that the values of the Development and Operations teams needed to be combined. This was one of the causes of the movement.
The DevOps definition below is best learned if you are reading it for the first time. The phrase “no one can formulate what it is” will not work in interviews at world-class companies.
DevOps is a set of tools and practices that help automate and integrate processes between the development team and the infrastructure team so they can build, test, and release faster and more reliably.
(Amazon AWS: What is DevOps? https://aws.amazon.com/devops/what-is-devops/Atlassian: What Is DevOps? https://www.atlassian.com/devopsGoogle DevOps: https://cloud.google.com/devops, https://ru.wikipedia.org/wiki/DevOps etc..)
Indeed, the fact that DevOps is a set of practices aimed at “something” is difficult to perceive. And at first there was no exact definition. The Agile approach (methodology) had the same problem of perception. This was interestingly discussed on Podlodka Podcast https://podlodka.io/104 . In my opinion, the most notable confusing issues in DevOps are:
1. Indeed, at first, in the market, it was customary to confuse and frighten people with the wording.
2. Differences of IT market regions.
3. Scale of projects.
Different regions of the IT industry
The market brings different requirements and expectations to engineering positions. SRE found itself in the same situation. Google engineers in the first book (https://sre.google/sre-book/introduction/) wrote that SRE is primarily a developer who is well able to monitor and maintain his code in PROD. Now, often, SRE is a separate position not related to development.
In the US market, developers are often responsible for DevOps practices, while in the CIS market, DevOps is often handled by Ops engineers (SysAdmins). Thus the question “What is DevOps for you?” had the right to exist.
Different project scales
Sometimes the question “what does SysAdmin/Operations do” comes up. The question also has a right to exist and is often asked by engineers of small services.
For example, there is a small online store selling one Ferrari per month. For such a store, the simplest but fantastically beautiful Frontend is enough. After handing over such an online store to the end customer and going live, one or two programmers are enough to release new features.
In comparison, for a media platform like Youtube for only one player, you need a network of servers with different roles. Then servers / services responsible for high load, fault tolerance, etc. will be added. Preparing and deploying code, automating and maintaining infrastructure on such projects takes days. It’s not that it’s a problem or uninteresting. It’s just that on a regular basis, the developer is busy writing features or the main code of the service and not maintaining automation for CI / CD and infrastructure. But the question “what do SysAdmin/Operations do” on such projects disappears.
When we don’t have DevOps
To better understand the definition, you can look at the question from the other side: when do we not have DevOps?
if there is no cross-functional team (For example, you are writing the Linux kernel. Then, simply, you are a system programmer who needs CI to be configured for teamwork and reduce integration costs when merging)
if there is no Time To Market (No need to deliver new features to users)
if there is no culture (Language trap, culture cannot “not exist”. When “everything is bad” is also a culture, it’s just like that. I’ll give a simple example: the team does not share responsibility for values. For example, for Dev it can be frequent feature releases and for Ops – sales stability)
When will DevOps cease to exist?
Just like Agile, Scrum or ITIL, DevOps has formed a maturity model and is moving into a default approach (PMI Defining DevOps, ITIL® 4 and DevOps White Paper ). When Scrum appeared, everyone “implemented” it, underwent lengthy trainings, bought books, shared their experience. Now joining the Scrum team, it is enough to read how the Scrum framework is implemented on one sheet.
You can look at the path of DevOps through the technology maturity cycle. The market regulates the process and DevOps moves from the top of the hype to the plateau of real demand as the default standard. You saw this process relatively recently with BigData. Now artificial intelligence is moving along a similar curve.
We can say that DevOps is becoming less noticeable, but it does not disappear anywhere, gaining a foothold in its market segment. The speed of learning practices and approach becomes faster and more accessible. I wonder what the next round of development of the industry?
“Accelerate! The Science of DevOps How to Build and Scale High-Performance Digital Organizations”: https://www.litres.ru/dzhez-hambl/uskoryaysya-nauka-devops/
An interesting SRE course from Google (you can watch it for free if you select the “Audit” button at the start of the course): https://www.coursera.org/learn/developing-a-google-sre-culture
Books of the “founders” of SRE (in Russian): https://www.piter.com/collection/all/product/site-reliability-workbook-prakticheskoe-primenenie And https://www.piter.com/product_by_id/110769373 + free originals: https://sre.google/books/
DevOps Vs. SRE: Competing Standards or Friends? (Cloud Next ’19): https://www.youtube.com/watch?v=0UyrVqBoCAU&t=855s
Love DevOps? Wait until you meet SRE | Atlassian https://www.atlassian.com/incident-management/devops/sre
Book “founders” DevOps (in Russian): https://www.mann-ivanov-ferber.ru/books/rukovodstvo-po-devops/