As we entered the cloud providers market and fell in love with B2B. Case MCS

Mail.ru Group is a set of divisions and projects, from Mail.ru Lady and online games to social networks and IT solutions for businesses. Today we want to talk about one of them - the Mail.ru Cloud Solutions (MCS) cloud platform, which officially marks a year in January, but its history began even earlier. We talked to the vice-president of Mail.ru Group, head of cloud and business services Yegor Ganin, who told us how this project started, how the B2C approach helps in working on the B2B market and why the cloud provider has its own laboratory for experiments on iron.

How it all began

Please tell us what kind of infrastructure the Mail.ru Group services and products use. And did it form the basis of the cloud?

Mail.ru Group has a really big infrastructure. “Human” - in the form of formed teams that have been gathered for years, and technical - servers and support staff. In addition, many Mail.ru Group projects, due to their own scale, are themselves infrastructure customers, so we also are in the shoes of a customer and understand very well how IT systems are built and what the client needs from them. But in 2016, a new project for Mail.ru Group appeared not only due to “domestic demand”. In December of that year, we launched Hotbox and Icebox object storage on the market. Launched them with a difference of several months. The warehouses were built exclusively on the infrastructure of Mail.ru Clouds.
')

Mail.ru Cloud fulfilled and performs enormous volumes of tasks from the point of view of the market. “Flooding” and storing hundreds of terabytes daily is an impossible task for most providers and IT companies. Just because it is not their competence. Now any company can search for a technical director who will help build the infrastructure to save several terabytes per day. From three to five to eight terabytes, they create a special department, they are looking for a person. And we save hundreds of terabytes daily. As they say, we are able to store, love, practice.

We have historically cooked inside our company, not realizing how unique we have the knowledge and skills. At some point, "removed the blinders" and realized that we can help colleagues in the shop, opening the market their expertise. We took the “Mail.ru Cloud”, which then already stored the files of hundreds of millions of users. I note that it was at the beginning. Now we are gradually shifting the focus to the specifics of cloud service for B2B customers.

Next came the question of what access to the repository we should provide. Make your own access protocol that exactly answers all the properties of the repository, come up with new APIs so that customers adapt to them - we thought it was a long time and unpredictable. The first thing we looked at was the Amazon Web Services platform and their S3 protocol, which was practically the standard. We supported him in our cloud. As a result, we got “cold” and “hot” storage with support for S3, one of the most popular protocols.

Did you immediately start targeting an external client? Internal customers were not? Or were they at the testing stage?

Mail.ru Cloud was both an internal customer and a performer. It was launched in 2013, and by 2016 it had gained an impressive audience and experienced a lot. The launch at the facilities of such a project allowed us not to look for someone else inside the company.

In this context, for any service is characterized by the phrase: "children's sores." Whatever service you create, it will “fall” under load, even if you check it in advance. Therefore, the market is new to skeptics.

Few people can say about themselves that even a year before the official launch, he provided data storage for hundreds of millions of users. And we can. Therefore, our S3 storage is already an adult and completed service, and was already on launch. We were harassed only by repeating the methods of the Amazon Web Services API - the list of methods turned out to be limitless. But we dealt with them.

You said that one of your first customers was 1C. Tell me, please, with what tasks they came? What did you want to do with the cloud?

At first, we did not know that 1C came to us. They participated in the beta testing, where we launched everyone and tried to do this for a couple of hours or days, depending on the load (ours and services). By the way, high response speed is an important moment for us. Critical to business and proper working interaction with customers. We initially try to make the client not have to wait. This concerned the service already at the stage of beta testing applications.

The guys from 1C are open to the world and to everything new, they actively support any undertakings and test novelties with undisguised curiosity. I think that curiosity led them to us. They silently tested, downloaded, stored, and then opened the cards and came to talk about the production capacity and the contract. Usually, when a large customer arrives, companies try to offer him the best of everything: a manager, if you wish, a separate data center, where no one is allowed in, a private farm, a hypervisor. At the same time, our head was occupied with other things. Therefore, I do not know what exactly they did then, and what they are doing now - I will not say.

We are B2B from B2C

We come from B2C, and we have the DNA of a B2C company. It was difficult for us to adapt to the B2B market. And I will lie if I say that we did it without pain.

For me, the B2B marketplace is a solution for a specific customer. There are so many conditions and people in the process that the product behind them is lost, it simply does not end up with it. There is a realization of the needs, requirements, plans and ambitions.

The customer comes, says what he needs. And the performer begins to somehow implement it. The result does not look like a product. It looks like integration, custom development, whatever, but not a product that the performer would use himself.

Naturally, there are no other designers, productologists of this level, as in B2C companies. For many years, I and all my colleagues “sharpened” under the product - so that each particular button was in its place. "Their place" is determined by a bunch of A / B tests, experiments, conversions and other things. If the letters go away, then they leave not just like that, but as messages needed by the user, and their appearance is determined by a large number of studies. This allows us to enter the market with a higher level of knowledge and expertise regarding products, their vision, UX, design.

If before the administrator had to create a virtual infrastructure, he went to the service. This service was once created by the same administrator who was hired by an integrator or provider. The administrator somehow made the interface, somewhere there is a button, tasks are solved. But few people consider the number of actions and their complexity. For example, we calculated that in Horizon from OpenStack, in order to create a virtual infrastructure, you need to make 20 clicks. This is nonsense, because the virtual infrastructure must be configured in 2-3 clicks maximum. For this reason, we started rewriting Horizon functionality.

Initially, we had an OpenStack infrastructure, because we had to run quickly. But we transform it into what is correct in our understanding.

Therefore, I think that a B2C company with a “DNA” set up to develop products takes a more confident position regarding those who have never done it.

Can you trace the development dynamics of the team and the people who are currently working in the project? Given the fact that initially you had and partly maintained B2C-attitude, do you take those who specialized in B2B?

Of course. For us, “roots” in B2C are both an advantage and a disadvantage. We were aware of this from the very beginning. I understood that we are strong in B2C: we will perfectly make the product, rebuild it perfectly, we will monitor it. And the team will aim for that.

But in the classic B2C product is done for free. This has a plus: you need to make a quality service so that the user does not leave, does not switch to the same free solutions of competitors. You earn on advertising and do not sell anything to the user. You treat the user as if you “owe him”, and you do the product for yourself - otherwise it will not work.

But in the B2C approach there is a minus for the users themselves: when creating a product, you get used to decide in favor of what the “majority” need right now. However, the "majority" will not bring really strong solutions and does not know what the product should be. To overcome this inertia, you need to have a supply of confidence and willpower. Create something that will be necessary tomorrow. And we take this from B2B, in which vendors do not sell the product, but solve the problem.

But in B2B there are some imbalances. Instead of the user, here is the client. Unfortunately, many vendors here refer to customers as if customers already "should" them. Sign a contract, pay money, take work. What happens next is not particularly important. The main thing is to do the work, close the contract and start doing something new.

Our attitude to the client as a user is good, we quite naturally feel that “we owe him”, as a result, we differ favorably from our competitors. But this is a weak position in the market with "sharks", where we had to work. We needed guys with “teeth”, seeing in the client the person who owes us too.

We periodically give this “injection” to the team. We have a very strong staff in terms of business development specialists. The product team also has people from B2B. All of them are not about a B2C user, but about a B2B client. It turns out a complete merger: “altruists” from B2C learn how to monetize their attitude, and “egoists” from B2B sincerely begin to see the user in the client.

We switched from the format, when only the product knows what to do, to the format when the customer orders, and the product sees how to do it better, then to turn it into a product for others. This is a complex scheme, which, in my opinion, has no flaws.

About clients: we aim at digital and enterprise

How has the client's portrait changed from what it was at the start? How will it change, according to your estimates? With changes in regulation, new rules, laws and trends for digital transformation, which is now gaining momentum again.

The market can be divided into three segments. The first segment uses low-cost solutions. Such solutions are offered by low cost airlines. These companies operate with very low margins. They provide the service "as is", and you use it at your own peril and risk. This is a normal approach. It's just that this is the case when you bought a service, and you have no substantial guarantee or support.

The second segment we call digital. These are companies that were created on the Internet and earn money on it: they provide various services, from content creation to selling cars. In addition to those already mentioned, this is FL.ru, CarPrice and others.

And the third segment is the enterprise. This includes large holdings, usually far from the Internet, with a large number of needs. In Russia, these are, most often, commodity companies.

Initially, we aimed at digital and enterprise.

Tell me more in detail. Obviously, government agencies will not switch to foreign infrastructure.

State structures - a separate segment. They are burdened by laws and internal policies and recommendations. Generations have to change before the “gos” go to the clouds.

Every year, the Enterprise is more and more willing to talk about clouds, sometimes it even takes the initiative.

Digital - these are people who have come to the stage when they either need to start using cloud solutions, or they already use them. They realized that it was time to switch to the cloud infrastructure, because in the data center it became crowded and expensive, there were difficulties in scaling and launching new products.

It turns out, Digital does not need to be trained. He is well acquainted with the topic.

Yes, they are our colleagues in this market. These people have heard about the clouds, often go to conferences, some make presentations, with stories about the successful implementation of cloud infrastructure. Digital is well trained inside specialists who go on either “down” - to make their own start-ups, or to go up to the enterprise, where they start using clouds.

In the case of an enterprise, an understanding of the need to move to cloud models often comes from the digital segment — from those colleagues who either worked in digital or were close to it. But the use of clouds in such companies is hampered by internal regulations and recommendations of the security services, many of which have a simple meaning: “inside — possible, outside — impossible”. If you come to sell them a cloud, they “wrap themselves in a greatcoat” and leave. Just because the cloud is incomprehensible, unfamiliar, and somewhere even forbidden.

Then the wording “private Cloud” is born. This is such a way to convince yourself that you use the cloud when you can not. And, on the one hand, it allows to accustom big business to this concept, which means, sooner or later, it will remove the fear of it. On the other hand, it inhibits the development of the market. No matter what is deployed in a private cloud, it is still deployed on physical servers that are serviced internally. From an economic and infrastructure point of view, this is the same as on-premise. In fact, it has the same limitations in scaling and requires the same purchases of iron, which means it does not reduce costs. And taking into account the need for internal development - multiplies them.

But so, of course, not everywhere. And in the classic enterprise there are people who now understand that there is no need to continue to purchase servers, spend money on internal infrastructure and development. They understand that it is right to switch to a new consumption model, a sharing economy, where resources are allocated on demand, and expenses become event-driven. Gradually, these people are transferring the development environment and test environment to the cloud — not production and not critical data.

The cloud can be either public or hybrid: when you have one part of the infrastructure located in your own data center, and the other is in the service of another team. Moreover, the second part can be easily scaled and rebuilt. Only then you can be sure that business is starting to save.

Own "cloud laboratory"

Can you, from the client side of a B2B cloud, explain how your choice of equipment and location of data centers affects the quality of services a client receives? Either in value, or in speed, or in something else.

Mail.ru Group is one of the largest customers of servers and other equipment for creating Internet services in the country. Because of the scale we have favorable conditions for the purchase of server equipment. Few can compare with us.

Does it save both you and the customer?

Yes. In addition, we have our own lab equipment. Here comes some card or a new processor. Further, most of the two scenarios. The first one is to wait until the new product is “on the market”. See how the community responds to it. It will take a year and a half. The second option is to buy and see what happens.

We go the third way - use the laboratory. Regularly we look at new items from vendors. For example, NVIDIA or Intel give us the latest equipment for testing one of the first. In the laboratory, we assemble test benches to meet our needs: one solves storage problems, the other is assembled for virtual machines.

It turns out a potential "combat" stand, which can then become the basis of the real.

Yes, the future of combat. In the process of testing the stand changes its configuration. Ultimately, we get a proven solution.

So we collect equipment for ourselves on exclusive conditions, with exclusive parameters and prices. This allows you to compete with many players in the market.

How we work with clients

How do you work with feedback? How much is this accepted in this service sector? Do your niche often take a detailed interest in customer needs and collect feedback, which then may affect the development of services?

We try to be as close as possible to the customer. It does not always work right away, but we clearly understand that it is necessary to maintain constant communication with the customer. Moreover, the contact should not be in the form of only marketing newsletters and Telegram-channels. This is not our story. We want to build a relationship of trust with our partners. This is achieved by a large number of tools, methods, methodologies and our active actions.

For each major client, we will organize convenient ways of communication. From the client’s side, all interested people are connected to the dialogue, from ours - an account manager who leads the client, technical specialists and product managers. The latter give the client the opportunity to participate in internal processes, discussing plans for the development of a service or team.

In this case, the client is not communicated by the conditional assistant of the call-center, but by the person who influences the product.

Yes, that is the main point. We want a person to create a feeling of warm and competent communication at any stage of work and interaction. Some customers have phone numbers not even account managers, but a product team.

Are there any “sharp corners” that customers are talking about in the market? How do you smooth them? How do you explain risks, remove concerns, especially when they are unnecessary? Or is it all now more connected with the formation of an understanding that the cloud can and should be used?

Our customers have already made this decision. But there are those who are not quite ready for the cloud. They have to cook for a long time.

Here you expect that the external environment will push them?

Not. Do not hope for the external environment. Rather, we are this external environment - only for the market. The only people who can influence people's decisions and the development of the industry are large companies like us.

How IT giants form the cloud market

We came to the market with an understanding of its growth rate - 40–50% per year. This is a good pace, but not the maximum possible. In order for the indicators to increase, we definitely need to stand behind this "trolley with clouds" and push it along with everyone.

I strongly welcome any launches of competitors, any changes, the struggle for customers. If someone launches new services or platforms, he automatically popularizes some kind of PaaS or IaaS. Once again creates an information hub that will help convince another skeptic that clouds are the norm. Therefore, we are investing in the PR of new services, organization of events. This is all very important. We have no illusions that someone will do it for us or it will happen by itself.

Cloud infrastructure makes it possible to launch several products instead of several products a year a month, this rhythm “swings” the Russian IT market as a whole. At the same time, one should not forget about scaling services. If the service began to grow, at some point — almost certainly — it would “fall.” Falls divert power from the core business. When a service uses the cloud, this risk decreases because it scales fairly quickly. Clouds increase the speed of development and launch of businesses, and, as a result, catalyze the growth of our economy.