Our meetings at the DevOpsConf conference focused not only on the tools of engineers, but also on the processes for which these tools are needed. It seems that this is not enough for the business to see how to get the maximum benefit from the product from DevOps.
Therefore, we added a focus on how people in different positions determine what maturity DevOps practices have been achieved and how to consciously choose the goals of technical development in their organization and team. In other words, I want to see where on the DevOps field I am right now, and what is my next step.
The main measured characteristics of DevOps are the stability of the applications and the performance of IT teams, from the idea to the calculation of features on production. Therefore, we talk a lot about time to market and monitoring and continue the technical track.
And IT teams are made up of living people who can not only give out good KPIs, but also do obviously useful work. After all, if the DevOps approach has gained popularity in the world, then, probably, someone needs it. For you, we met with Product Owners and businessmen who do not always know what DevOps are (as if we know: D) and asked them what it is important for them to get from techies. What is this very good.
The first thing I had to change my vocabulary and way of communication. We did not speak in our usual terms, but tried to use the language of the products. Even as a way to communicate, we chose CustDev-interview, as is customary in the grocery world. These interviews confirm or refute our hypotheses about what is important in business work on a product.
Here are the hypotheses we tested at the meetings:
- TTM is important to Product Owner’s interests.
- The stability of the application affects business performance.
- Developers often disagree with PO’s views on what and how to do.
- TTM helps navigate CustDev. It’s not about the technique of the interview, but about finding and confirming the needs of customers and building a company.
I talk through Zoom with my old friend, who is convinced that a person who has never sold in his life has nothing to do in the Product Owner profession. She often appears in broadcasts on radio and TV, conducts seminars in her subject area. As soon as the self-isolation regime was facilitated, she and her husband and child rented a house on the shores of a magnificent lake and moved to live and work there all summer. Her company has been in the online services market for almost 20 years. In first place in ratings in their subject area.
– Please tell me, are you doing any special work to shorten the development cycle and launch features in production on your team?
“You know, no longer.” If in 2014 the bottleneck was the productivity of IT specialists, both development with testing and the operating team, then for several years I have been solving another problem. How to prepare experiments so that the team has something to do (laughs). Seriously speaking, now I’m sure that the feature will be launched in a period of six hours to a week …
– Wait, 6 hours ?! It…
– From the first conversation with the team to the user. Those. already on prod.
– Why do you need such a speed?
– I did not set such a goal. Rather, I am inclined to think that this is a natural consequence of those changes in the development and operation tools and settings of the interaction between the teams and me, as a representative of the business.
We continued the conversation, and over the next half hour, Natalya said that major changes to the product are now much more confident and calmer. The main factor in this change, she calls her team confidence and team trust To her.
My next call is to Phuket. Igor moved there a few years ago and for this he had not the easiest negotiations with his employer. Toda remote work was a curiosity, and all employees worked in an office in Moscow. Now squeals and fuss of his large family are heard behind the scenes. His company also leads the Russian market in its area.
– Suddenly, I became an expert on a remote site this spring! (laughs)
– Please tell me what conscious steps you took in order to increase the stability of the application and why?
– When we started working with the technical director, terms like LTV, customer retention, or unit economy were not yet in use. We just presented that users are not very happy with the fall of the application 20% of the time …
– Wow, NPS!
– Andrey, there was no NPS. This is now a technologically advanced indicator of how satisfied users are. He will not show me the money. He will not show what is the ratio of the costs of a project and the return on money from its implementation.
– Is this attitude important to you?
– The growth of the audience is important to me. Those. her loyalty and commitment to my product.
– Something else? How is marketing related to uptime?
– Look, “pour traffic” costs money. Whether it’s direct advertising or SEO optimization, organic conversions. When resources are invested in traffic, and on the application side it falls into a black hole, then this is a bit of a loss.
– I.e. It’s important for you now that the application is available.
– Now I am calm, because our uptime is 4.5 nines. Within a month, the application regularly responds to 99.995% of requests.
– DevOps did the trick, DevOps can …
– DevOps also can not. At some point, we felt that our way of development, through changes in the “laboratory” and testing on focus groups, is more expensive than testing on small parts of the audience. Instead of developing a prototype, gathering a focus group and getting approval, and then see their disappointment on live users, I can quickly and quickly check three to five options for 0.1% of users and choose the best one.
– I.e. the developers nagovnovodyat something quickly on crutches, and then you choose the best option.
– Yes, from sticks and blue electrical tape. (Igor frowns, because I’m on the speakerphone, and there are children nearby)
“I see two potential problems here.” First, developers are unhappy that part of their work will be deliberately thrown away. Secondly, those features that “shot” remain crutches.
– Therefore, it is important for us to manage the technical debt. We can say that there is a contract between business and IT: during the year at least 30% of the development time is spent on raking debt. Moreover, we consider these percentages cumulatively for the year. For example, in the spring of 2020, we blocked working with technical debt completely, because the rate of change of the product was important to us. We had to look for new models for using our services.
– Does this mean that for the remaining six months, the guys will already be engaged in refactoring, changing architecture and sawing up crutches, as well as working with application performance?
Then Igor said that he had the opportunity to work ahead of the curve. A significant part of the tasks is aimed at the development of new technologies and interfaces. The first results of the experiments are already available to users, for example, communication in a natural language. At the same time, today we can rather talk about the Research part of R&D. The company is mastering the technology in advance to take advantage of the maturity of artificial intelligence technologies to gain a competitive advantage.
If we talk about experiments, the application settings were divided into three main groups:
- Grocery. Product owner can enable or disable the feature, as well as roll it to a given percentage of users or even a specific list.
- Service interaction settings are the responsibility of developers.
- Admins, they are turned by a team that deals with infrastructure.
It is curious that the first request for separation of settings came out of operation when releases were still launched manually. 20 releases per day to change the settings – this is not what pleased the admins. Already when the settings service was ready, it created a technical readiness to “twist the knobs” to the product owners.
We also phoned the guys who lead the product in companies where the potential for the development of DevOps is much greater. They launch startups or work for a large customer. In other words, and more through pain, they talked about the same values for the Product Owner as the previous interlocutors.
We found that the findings of our study support the findings of the Google report. The 2019 Accelerate State of DevOps: Elite performance, productivity, and scaling.
We highlighted four core values for Product Owners using DevOps:
- Predictability of the feature implementation time and confidence in the quality of software are the necessary basis for active experiments.
- Reliability of work is selling = money. When traffic gets to a running application, this not only allows you to rationally use the budget for promotion, but also increases user loyalty, and hence market share.
- The speed of experiments determines success for both a startup and a business with a mature product. If it is important for a startup to quickly discover user preferences and successful answers to them, then a mature product needs user retention, stability of mass revenue and research – work for the future of technology.
- Trust in developers and mutual trust of developers in the product. Partnership between the IT department and business units, when a “contract” on the methods of interaction is concluded and this contract is executed by both parties. In terms of DevOps, this is technical debt management, maintaining the code in order and involving the team in the interest of its users.
Therefore, at the conference we will talk about both important technical topics and new ones. For example, Alexey Pikulev, who is preparing a workshop on the diagnosis and development of trust, will talk about trust. We will continue to research engineering practices and tools, as well as the organization of processes. We will discuss business communication with the team and between teams at the already traditional process track of the conference.
Guys, do not worry about whether there will still be a lot of technical reports at the conference. Because we, program committee members, love system administrators and DevOps engineers who are not. And even we ourselves are.
We will make most of the activities at the conference interactive, because there are enough talking heads on the Internet. After all, every conscious engineer can not only find video reports on the topics he needs, but also read articles (for example, articles following all DevOpsConf in Alexander Titov’s profile).
Our goal this fall is to give each company the opportunity to land a virtual landing from Product Owners, technical experts and engineers of all stripes. So that the IT special forces navigate the terrain and return to work with a developed plan to capture the universe!
In the following articles we will tell about CTO, developers and security – why do they need a conference?
While there is little time before the conference, subscribe to the newsletter: we will send letters with news about the program, announcements of reports (and transcripts from past conferences), as well as various surprises. AND submit applications on reports or master classes if you want to influence the development of DevOps in Russia. And if you are a product that all of this DevOps only interferes with, write in the comments how you manage to stay afloat.