START Streaming Service: How It Started, and How It Is After Five Years
My name is Kirill Evseenko. I’ve been CTO at START (start.ru) for almost 4 years now, after hopping on almost at the very, well, start. My total experience in engineering is over fifteen years. I started out as a system administrator, and at 30 I was promoted to a team lead. For most of my career, I’d been team lead and CTO, but mostly at small companies with modest engineering teams. Once START, including my team, skyrocketed, I began running into all those managerial, technical, and process-related problems that start-ups always experience in the growth stage. At this point, the speed at which you launch new features is more valuable than building a proper project management system, so start-ups usually shape their workflows without a good thought about scaling their product in the future. Personally, I had no such experience — I built it up alongside START and my team. I first tried to dig up some information on how to manage it all without giving myself any excuses and sacrificing this and that; I failed. No books, no talks at professional technical conferences — all we could do was test our own hypotheses and assumptions. I had no trusted sources, nor could I find anything to help me avoid mistakes. At length, I decided to write an article to share the knowledge I’d amassed during this growth stage. I know my experience isn’t going to be a silver bullet for someone who’d hit the same wall, but at least I hope it’ll give you some perspective.
Company History
START was founded by the Yellow, Black, and White Group in 2017. Since launching, we’ve developed an international streaming service with such world-famous projects as Gold Diggers, The Kitchen, Addicted, and others. This subscription-based service has a Russian-based production studio and a partnership with Disney Russia, as well as other major distributors. The Yellow, Black, and White studio was founded in 2007. In 2017, we developed around 20 titles, and now we’re up almost to 60 for 2022. START is growing, and not only in numbers; we’re also evolving in terms of quality and project scale.
Netflix first established its video streaming service and then began producing content. At START, we were producing quality video content first and foremost. We now have a film production studio and wish to promote our own movies via the D2C (Direct to Consumer) model. Our primary growth driver is our production; all our series and movies are released on START, or as part of our limited subscription inside other streaming services. We settled on this business model from the get-go.
Technical Aspects at Launch
At first, START did not have much to show in terms of in-house development; we deployed a very bare-bones online platform for Yellow, Black, and White and turned to third parties to develop the missing features and modules. Originally, the platform consisted of external blocks, including the recommendation system and billing; they were then integrated with each other. In a nutshell, we rolled out a minimum viable product to launch the business; for instance, we didn’t even have our own CDN (Content Delivery Network) for video streaming. The IT team has grown since, and START doesn’t outsource any software development anymore, but back then the team had barely a dozen of programmers — a far cry from the 80 we employ today. We plan on further expanding our team this year, doubling its size.
Platform Development
START was launched with limited functionality and then developed into a multi-featured streaming platform. As the project was growing, we were running into complex challenges and high outsourcing costs, especially when dealing with integrating external products. In 2018, our focus fully shifted towards in-house development; we started with the billing system, it being one of the key platform infrastructure elements. We also developed an encoder and started expanding our own CDN network.
We kept developing our product and adding new features. Over the last four years, we’ve implemented authorization with social media accounts and integration with payment systems; we’ve also translated our service into six languages and developed a partner integration network with other services. At the same time, we kept improving and maintaining the partner API. We still conduct experiments all the time to improve the overall customer experience and support streaming on new devices as they come to market. We constantly improve visual and streaming elements, and the bulk of product transformation happens around app development. That includes Smart TV, mobile, and web services; each of these has progressed quite a bit. I remember how, back when we started, the team couldn’t even launch across all platforms simultaneously — for example, there are quite a few Smart TV vendors, which required more effort as far as development goes. We want to make sure START keeps meeting the customer’s evolving demands, so we keep adding new features both to the platform and the apps. We roll out all these new tools and tech to secure future business growth and client retention.
In terms of infrastructure, we opted for building and optimizing our own cloud-based Kubernetes cluster. An independent microservice is responsible for balancing CDN (Content Delivery Network) sites. We’ve got five sites as of now, three of them in Russia, and the remaining two in Kazakhstan and North America. We are connected to backbone Internet providers and have direct links to communication providers, which gives us excellent connectivity and allows us to offer high-quality streaming to the end-users without compromising on quality.
Our Team
It’s worth mentioning that we managed to the keep genuine start-up spirit alive at START.
Many guys who joined the team back in 2017 are still with us, but obviously have grown a lot professionally since. Some of them came to START having little experience in the industry, but our corporate training and support helped them make progress and become fit for their current positions. At the moment, we have a number of specific development areas, each with a separate technical team assigned. For the platform, we have experts responsible for the Backend, Web, iOS, Android, and Smart TV; testing teams are in charge of the Backend, Web, Mobile, TV, and Automation; then, we’ve got crews handling project management, operations, DWH and RnD.
Our own team allows us to quickly implement new functionality, be flexible and quickly respond to any unforeseen situations. And I’m very proud to work with such.
Apart from the functionality transformation, START underwent a substantial change in development-related processes. Obviously, some techniques that worked for a 20-strong start-up are going to be useless for a company with 200-odd employees, yet we try to stay flexible and use hybrid practices, plucking the best from an array of frameworks to make our work as effective as possible. Consequently, we developed a systemic approach to meet all our business goals. These days, we’re working in two-week sprints; each team chooses the best schedule for their particular case. For instance, the product and project teams have daily synchronization.
How the Platform Works Now
Under the hood, streaming platforms are not very different from each other. START uses up-to-date tech such as Kubernetes and Docker, and we’re actively developing a microservice architecture. Today, we run about twenty different microservices and a client API gateway. The primary services include:
● User management system;
● Product management system;
● Billing system;
● Analytics system;
● Video upload and management services;
● CDN.
What the Future Holds
We expect growth opportunities to arise from personalized and user-friendly experience in the next few years. Consumers already stay connected at all times and want to watch their favorite movies and shows across different devices. We keep adding multi-application features so that users can seamlessly switch between apps. We strive to make our applications easy enough to use and offer equally available Smart TV, Mobile, and Web versions with the same set of features. Recently, we’ve redesigned the browser functionality, and we plan on adding further features shortly. As of now, START would like to add TV channels to the subscription plans and improve the recommendation system.
Conclusion
Well, it looks like I didn’t give you a recipe for preparing the team and engineering tools to withstand your product’s (service, app, whatever) growth, did I? Couple of years ago, I visited a major local team leads’ conference, expecting to find answers to my questions; I couldn’t. Maybe the way each IT company grows is unique, or maybe the managers act like me and just work on continuous delivery, neglecting to make any notes or update their company’s wiki on how each step was made. See, I honestly wanted to share some brilliant ideas and pro tips, but the thing is, I just don’t remember all those steps anymore. We simply ran as fast as we could, always trying to ramp the speed up some more. Anyway, here’s the only advice I can give (and I guarantee it’s 100% effective): write documentation. The real pro move here would be to hire a technical writer or two to make sure that every stage, procedure, implementation, and specification is described and carefully stored on Confluence, or anywhere else where it’s going to stay safe.
#START, #VOD, #Videoservice, #Engineering
