For 11 years we have been managing the IT infrastructure of clients of all sizes, from large companies to startups. We also tried different models of work in the time and materials format, when not the result is paid, but the time of the performer. All these models are, let’s say, inferior.

A year ago, we decided to test a service that is actively developing in the West – managed services, or managed IT. In Russian: “integrated management of IT infrastructure”. The company that provides this service is called the Managed Services Provider (MSP).

MSP manages server and network equipment, system and special software, cloud services. It also coordinates all the client’s IT processes, including interaction between contractors.

We are currently using managed IT in three projects with clients from different market segments:

• a large network of gas stations as part of an oil company;
• clothing manufacturer and distributor (4 independent brands);
• technology startup.

Of course, a year is not the longest distance. Although already enough to collect feedback, analyze metrics and evaluate the results.

Key takeaway: managed IT is the only IT service delivery model today that benefits both the customer and the supplier, and even third-party contractors.

Why this is so – we will show on the example of each case. But first, let’s criticize the established models a little.

Cons of outsourcing and outstaffing

Experienced service providers with large portfolios understand that almost any project goes beyond budget, time, or both. Because in IT there are a lot of tasks that have not been solved before. It is difficult to find a tried and tested method for them. An example is startups where the IT strategy can change at least every day.

An accurate assessment can only be given to small typical jobs: installing a database, deploying a new server, adding storage capacity – but these jobs by themselves do not bring value to the business.

Hence, there are three characteristic disadvantages of outsourcing:

• IT companies are often insured: they include, for example, more hours in the estimate, “with a margin”. This delta can be violated, and in both directions.
• It is difficult to track the actual work progress. There are commands that use services like Toggl to count the time. But this is inconvenient for either the contractor or the customer. Because in IT, a lot of time is spent on preparation and on interaction between participants – how to take this into account
• The client, in general, does not need to buy a watch from a specialist, but to solve his particular problem.

Regarding outstaffing: normal IT companies don’t even consider such a model. Because this is a kind of modern slavery. You find people, sell them for rent – and you get something like a French legion, which is involved in incomprehensible operations, it is not clear where and with dubious effect for the outstaffer company.

The term itself, by the way, has long been taken over by developers and municipal services.
Outstaffing is more about low-skilled workers in orange bibs.

Managed IT: when business and IT people begin to understand each other

Managed IT, or managed services, is a popular service delivery model in mature IT markets. According to the report of the analytical portal Market Watch, the global turnover of managed services will reach $ 250 billion by 2022.

A provider who works according to the managed IT model focuses not on the number of hours, but on the ultimate need: to give the client what he wants. It records the metrics that the business needs – that is, the one who allocates the IT budget. Metrics are specific services and results that these services deliver to meet business goals.

Why is it important? Because we often see that there is little understanding between the IT department and the business. Business in IT is not particularly versed, and this is not their task. In turn, IT specialists are not always able to convey their thoughts to the business – for example, when they want to explain why additional resources or staff are needed.

Managed IT creates a “bridge of understanding” between IT and business, makes everyone feel good.

A year of testing the model revealed some nuances. We noticed, for example, how difficult it is to change the consciousness of people: everyone understands what outsourcing and outstaffing are, but they do not particularly strive to delve into the intricacies of managed IT. And that’s okay.

The main thing for us is that the clients who participate in the projects have already appreciated the advantages of the new model.

Case No. 1: Gas stations network

Problems faced by the client before coming to us:

• Large infrastructure: about 200 servers inside the loop + 120 outside of it. Various software installed, test-, prod-, dev-sites deployed.
• There are many contractors who provide and maintain this infrastructure. Contractors don’t really want to communicate and interact with each other (and they shouldn’t). They fulfill their role, and the technical requirements are addressed to the client, who forwards requests to other responsible companies and teams. So the client turns into a kind of router, which greatly slows down the work.
• Different models of service delivery from contractors: someone uses outsourcing, someone sells software at a fixed price, offering support for this software, etc. This complex process also needs to be conducted.
• There is no qualified coordinator or “one-stop-shop” to advise on every type of IT work. That is, there is no one who makes the final decision and performs the function of a facilitator.

As a result, it often turns out that the client’s IT specialists are not doing what they need – not strategic planning or, for example, development. Most of the time they spend on maintaining what is already just in working order. It turns out a work for the sake of work. All this slows down the business, it develops more slowly.

Decision

• Private Cloud. We have moved the client’s capacity to a custom-built private cloud installation. Now he has a dedicated, secure infrastructure on top-end hardware.
• Consulting. Since we have been providing complex services to a large number of clients for a long time, there is an understanding of what is good and what is bad. We give expert advice on which direction is better to develop IT, and what is better not to do.

Difficulties

Because the client’s infrastructure is large, it has accumulated technical debt. Some long-debugged processes are not quite correctly configured. There were two tasks to solve:

1. Copy quality processes that are already in place.
2. Understand the rest of the processes, propose an improvement – so that they are optimal in terms of time and resource costs.

Even a year later, optimization (under item 2) continues. When you take on such a project, you accumulate information gradually. Because, as a rule, there is no documentation, it takes a lot of time to restore everything.

Plus, those processes that have already taken shape are not always easy to change. People are used to them, and changes are not perceived equally by everyone. But this is normal, this applies to any process changes. It’s great that in this sense the client meets us halfway.

Case # 2: Clothing manufacturer and distributor

The client is a holding company that includes four separate divisions (brands). For each of them, order statistics are collected. For this, a special software is used – the order management system. The client found a contractor who supplied the software, and organized the integration with all divisions himself.

We provided the client with a virtual infrastructure and adjustment of all processes – so that the infrastructure would work correctly with the client software. Plus, we took it upon ourselves to communicate with the contractor who administers this software. The project is now in an active phase.

It is worth talking about one important task that had to be solved at the implementation stage.

Single point of failure

Previously, each of the client’s divisions had separate infrastructure and software. A simple service for one of the brands – for example, a payment acceptance service – led to losses only for that brand. But the management decided to consolidate the infrastructure and software.

After moving to us, the client’s software works only on our infrastructure – and this is a single point of failure. If our servers fail, the whole holding is already incurring losses. Even with an idle time of 5 minutes, the amount will be sensitive.

The client asked a logical, seemingly, question: “If your system is down for an hour and we lose, for example, half a million dollars, can you refund it us?” The answer was: “No.”

Not all SLAs are equal

We cannot trust the business performance of all of our clients. We do not know how these metrics are collected and calculated. It turns out that any client will come to us, say that he lost half a million – and we will have to believe him? But then our compensation may exceed the amount of all money received for the service. This is pretty weird.

We have an SLA of 99.9%, the standard for service providers. But we offer 99.9% “by documents”, although our real SLA is close to 100%.

Because we use the best equipment on the market. We save by reducing the likelihood of incidents and downtime.

Meanwhile, there are many IT providers who are engaged in imaginary savings: they buy cheaper hardware – for example, Supermicro servers. And this is a completely different “three nines”.

SLA compensation is not intended to reduce customer downtime losses, but to ensure that the IT service provider also incurs losses. That is, so that the supplier also hurts.

The client understood and accepted all this.

Case # 3: Technology startup

Let’s start with a generalization. Statistics (same as our experience) show that out of ten startups, about one survives. Startups are in constant search, the same applies to infrastructure: many often change the direction of IT development, and this is not always true.

Typical startup mistakes

• In an effort to save money, startups raise one server and deploy everything on it: a web server, a database, a development environment, applications, logging, monitoring, and so on. And when the moment of growth comes, the company begins to simply increase the server resources – to increase the number of CPU and RAM. And this vertical growth always hits the wallet painfully. If you don’t grow very quickly, the cost curve goes up smoothly. But if growth continues, this curve at some point sharply changes the degree. The economy stops converging.

Solution: building a microservice architecture, when each service lives on a separate server.

• Often in startups, several specializations are accumulated on one person. For example, a developer is both an administrator and a tester. And a business owner is an investor, an account manager, a product, a marketer, and a seller. In such companies, employees are in constant overload, working almost around the clock. Burnout happens more often here than anywhere else.

Solution: transfer of functions for management and strategic development of IT infrastructure to a service provider.

Our client

The startup is developing an intelligent platform for online advertising campaigns. This is a complex and resource-intensive software that is rapidly evolving. To support it, the client previously rented an old “hardware” with outdated software from a data center that does not meet Tier III requirements.

What we did:

• We provided a virtual infrastructure on high-quality hardware, transferred software to a new infrastructure. Since then, the performance indicators have increased markedly.
• Set up the databases, organized the Galera cluster – so that the client could scale both vertically and horizontally.
• Offered to abandon the idea of ​​a monolithic infrastructure and explained why it is important to grow towards microservices. And told how to do it.

Conclusion

Why the managed IT model is good:

• The business is beginning to understand how the metrics they want to get are related to the metrics that IT services provide. And also – how the payment will increase with increasing resource consumption and growth, so that it can be tracked, and the economic model would converge at all stages.
• IT specialists are starting to speak the same language with business. They understand how to justify increasing resources. The IT department takes over its immediate responsibilities. In such conditions, it is much easier to get satisfaction from work.
• The provider who provides the infrastructure also wins, manages it as it should, and makes recommendations based on best practices. The supplier gets a cool experience and can warn customers against mistakes that others have already made.
  For customers, the model is also beneficial in that a normal service provider provides resources in all possible clouds – both in its own and in the clouds of hyperscalers.
• Of course, managed IT has its limitations. The service is actively evolving as it receives feedback from the market, somewhere it needs to be improved, including on new projects. However, it is already obvious to us that it is promising. Managed IT is exactly a cut above outsourcing, outstafing and other derivatives.

Our favorite analogy for managed IT is car sharing. The whole world is changing towards consumption by subscription. Managed IT – also providing a subscription service, only in B2B. As in car sharing, the client receives a working service that he uses without thinking about its support, configuration and optimization. The provider takes care of this.