The pains of startups: how to properly develop IT infrastructure

According to statistics , only 1% of startups survive. We will not talk about the reasons for this level of mortality, this is not our business. Better to tell you how to increase the likelihood of survival through competent IT infrastructure management.







In the article:







It is worth clarifying that by startups we do not mean a coffee shop or an insectarium in a shopping center. We are about technology startups - about those who are haunted by the success of GitHub, Uber, Slack, Miro, etc.



Startups always have a lot of problems that prevent them from shooting: from insufficient investments to an unworked business model. Oddly enough, the problem with the first successes is in the same row.



Early successes are evil for startups who overestimate their capabilities, especially financial and personnel. After the closure of the first successful cases, such optimists have a desire to immediately expand: rent another office, recruit new sellers and developers to the team, and at the same time scale the backend (yes, so that with a margin). Here problem # 1 immediately manifests itself.



Startup people do what they can't



And they don't do what it takes to grow a startup. Let us explain.



Each startup must have at least three roles closed:



  • IT specialist (or technologist);
  • seller (or marketer);
  • visionary (or an entrepreneur who is also often an investor).


Often these roles are confused. For example, a startup is an IT specialist who, in addition, has to sell. He never sold and does it as best he can. A startup like this is a kind of malignant cross-functional team.



But let's say a startup is lucky: there is someone to sell to, and the IT specialist does his own thing. However, a rare IT specialist combines different qualifications: a developer, a tester, an administrator, an engineer-architect. And if it does, it is hardly equally good. He can understand middleware, but not very well with cloud services and virtualization software.







As the backend expands, the burden on the IT professional increases. Something starts to "sink". Worst of all, if it's a critical direction for a startup, such as product development. And now a person has to work overtime, and sometimes around the clock.



Overloading from a lack of people and skills is a characteristic feature of most startups, a consequence of the fact that people are doing the wrong thing.



All services are deployed on one virtual machine



Startups often, based on their own ideas about savings, place development environments, databases, a web server, monitoring, and so on on one VM. At first, this whole economy works more or less tolerably. Problems start when you need to scale.



Startups are usually scaled vertically. That is, they simply increase the number of CPUs, the amount of RAM, disks, etc. - this is a classic monolithic approach, the negative effect of which at some point becomes irreversible. If a young company grows, at a certain stage the price tag for the increased resources jumps to an unaffordable level. In this case, there is only one way to optimize the infrastructure: to reassemble it.



How managed IT helps



For this type of project, we have a managed services class - managed DevOps .



The customer receives "out of the box":



  • preparation of the necessary environments for work: dev, test, prod;
  • configured CI / CD processes;
  • prepared tools for teamwork: task trackers, version control systems, deployment, testing, etc.


At the level of infrastructure and tools, all startups need about the same thing. Comparing the venture capital market to gold mining, the Managed Services Provider (MSP) provides new, quality tools: picks and carts that don't break, cards that don't lie. The prospector can only choose the place where to dig.



Pros of managed IT



Managed IT is a comprehensive service that covers a number of essential needs.



  • At the start, we provide the necessary and customized resources for work, growth and hypothesis testing.
  • We can say for sure how the cost will increase when scaling, because we know that the key metric is the convergence of the startup economy.
  • We advise to save a startup a significant number of man-hours. We can also help with calculating the unit economics of the project.
  • We share the best market practices. The people at ITGLOBAL.COM have worked with a large number of startups. Many of these startups have monthly maintenance. This allows us to piece together the best (and worst) examples and share our experiences with clients.


Two cases from practice



We cannot name specific companies according to the NDA, but the sphere and the product - yes.



Industry: fintech / retail



Product: marketplace



Problems:



  • There was no testing in the CI / CD chain. Adding remote testers only made the build process more complicated.
  • Developers worked simultaneously on one dev server without dedicated environments in containers.
  • 70% of the developers' time was spent on the same actions from release to release. Development speed was very slow.
  • The infrastructure was deployed on low-cost hosting in Germany (i.e., neither speed nor reliability).


This, by the way, is observed in every first project.



Solution - managed DevOps: implemented CI / CD processes, set up correct testing and monitoring, intervened in development at the level of business processes, transferred the infrastructure to productive servers in a Tier III data center.


Result:



  • development efficiency has increased: new features and updates are released faster with less labor;
  • as a result, the cost of the development process as a whole has decreased;
  • the infrastructure has become flexible: the client can quickly scale both up and down;
  • the cost of managed DevOps, according to the client, paid off in six months.


Field: web advertising



Product: AI platform for automating advertising campaigns



Problems:



  • backend on the old hardware, in a data center with a low level of fault tolerance;
  • lack of regular backups;
  • monolithic infrastructure.


Solution - managed IT: we transferred the infrastructure to top-end hardware, set up a Galera cluster for horizontal scaling, showed how the load on VMs would be distributed, set up backups and monitoring. Now, in addition to servicing, we actively consult, including DevOps.


Result:



  • the infrastructure has become microservice: the cost of expansion has decreased significantly, and the possibilities for scaling, at the same costs, have increased;
  • increased reliability and safety of infrastructure;
  • Developers switched from a waterfall build model to CI / CD, which helped to reduce costs;
  • the financial benefit from managed IT, according to the client, became immediately apparent.


Conclusion



The survival of startups depends a lot on luck. One startup can spend money on expensive equipment and get nothing from it. Another will succeed even with a lousy IT infrastructure - just like a gold digger finds a gold mine with an old pickaxe.



However, the modern tools, practices, and professional staff that a Managed IT provider provides greatly reduce the likelihood of failure.



All Articles