Practical recommendations of transition to "clouds" of IT of infrastructure of Rolf Group company

Very interesting performance of Andrey Yakimov, the head of department of infrastructure of Rolf Group about transition of IT of infrastructure of the company to "clouds" to conferences of DPC 2011. Today we will talk about the practical recommendations of transition to the external operator of cloud services. The main objective of the presentation — to tell about how it entered the project of ROLF Group, and generally – about parts: from preparation for the introduction in the project to assessment of possible risks. You watch video.

Birth of the idea of transition to clouds

The idea arose in the company last summer and now continues to develop actively. Briefly about the project. Probably, all know about our production area: it is first of all automobile business – car sales. Our centers cover Moscow and St. Petersburg, but there are they and in regions. Generally it is own buildings where about 200-300 people work. The main infrastructure is concentrated in our main center where there are about two hundred servers, in our plans – to bring this quantity to 350-380 servers.

What was decided to make with infrastructure

What did we decide to make? We decided to integrate infrastructure of the cloud of NR company which is in Germany. The first project phase which proceeds now — integration of 150 servers, both physical, and virtual, the IaaS model and also transfer of a number of services, first of all corporate e-mail, on the SaaS model. So our main direction – IaaS, and SaaS is e-mail, backup, AV, AntiSpam which enter service. The second phase so far passing the analysis — transfer of the remained 150 servers which are distributed on the different centers, and implementation of VDI. So we will use VDI as service too that will help us to get additional benefits and to achieve a bigger plasticity. Now we work with two Russian providers, plus we have two channels in Germany.

Why we came up with the idea of clouds? Actually we in something were lucky because the company approached that moment when we saved a large amount of iron (350 servers and 3500 PCs), and rather old – age of 3-5 years. All further plans were frozen, and we seriously thought that in the next two years everything should be changed. But it had to develop into serious technical and human costs. The crowded server rooms, worn-out communications — from conditioning to the electric networks – here what was a situation. Changes were required also by nonflexible infrastructure: there was a classical situation at which business during crisis at first was reduced, then began to extend, and infrastructure remained at the level of pre-crisis times.

The cloud cannot be an establishing order method: work of structure should be arranged at a preparation stage. Before passing to new rails, any company should ask itself several serious questions: How many at us servers? what on them is set? How many does exist applications and business systems and who is responsible for it? Whether there are development plans for business systems? Whether the process approach is performed? Whether the directory of services is created? Whether our costs and the budget for business are transparent? As far as data transmission trusted network? So all this should be adjusted at you even before transition. Fortunately, even before work on the project we already had mature ideas, business-related, and we did not experience difficulties.

And can be to construct a private cloud?

In the beginning, selecting cloud services, we wanted to construct a private cloud. But it would not solve our problem as it should work with the same old iron. Both specialists, and capital costs of maintenance of all system in working order would still be required for us. Besides the private cloud has no sufficient elasticity. It grows up, but does not solve a problem of worn-out iron as iron in this case all the same remains. So you purchase some pluses in virtualization, but in many respects and lag behind.

About a public cloud

The public cloud did not suit us in general because ROLF – very dynamic and impulsive company, at us is many businesses. Complete solutions including on technical characteristics, and the public cloud could not give it are necessary for us. Therefore NR company offered us a hybrid cloud. A part of servers is placed here, a part is located in the territory of our supplier. As a result the optimal combination combining advantages of private and public clouds turns out: a flexible system with the smallest cost. As I already spoke, we transfer the majority of servers to the IaaS platform, the SaaS servers belong to our German partners, and PaaS we at the first stage did not select as on it it was difficult to count costs and to define perspectives of development of the company. In comparison with PaaS of the IaaS and SaaS platform are rather transparent and easily allow to make it.

And How much is it?

Concerning the price. To the introduction in the project at you a certain budget because it will be very heavy to start from scratch should be already put. Involve third-party business to your project, but do it cautiously because your investors hardly "in a subject" and, most likely, do not know the actual cost of services. So it is necessary to discuss with business very accurately conditions that expectations of investors were justified.

The choice of provider – a responsible step

Selecting provider, we cut at once all insignificant service providers because ROLF – the company experienced, and we could not risk scales. Thus, at once paid attention to large suppliers: the size really matters and is a guarantee that the supplier will not bring. If to select the large company – such as NR — it is possible to be sure that big capacities, big staff of professionals will be involved. The reputation of the company is much more expensive, than any financial performance. Especially if you look for provider with whom you will work the next five years.

Carrying out the tender was the most difficult moment

Carrying out the tender was the most difficult moment of preparation for the program for us. To impracticablly compare services – it is necessary to rely only on descriptions of suppliers and data on technical characteristics. In addition to technical characteristics there are many things entering service: different windows of service, the different number of the included services and so on. So it is necessary to compare one apple to another and to select what is better. And this with the fact that they look almost equally. That is why on studying of services which to us were offered by suppliers three months left. We tried to lead these services to some common denominator to compare their cost, to define, what is cheaper that it is better and that it is worse. But these "tortures" — the important and useful thing further helping to be guided by the necessary supplier. What it is better – Russia or abroad? Here everything is simple: the main capacities are abroad and consequently, the priority was given it.

Audit for transition to "clouds"

For creation of a cloud we booked audit on a paid basis. We consider this procedure obligatory as only audit will allow to determine the actual cost after the analysis of infrastructure and carrying out all measurements. It was impossible make it at the initial stage: we needed to understand accurately actual working costs for five years. Pay for professional audit several tens of thousands of dollars not a problem besides if know that these costs will pay off in the future. As a result the provider will be responsible not only for quality of the solutions, but also for the accuracy of calculations. As for our network, it was constructed long before the project: already there was a certain circle of providers. We used the same providers for carrying out channels to Germany. But for this work it is necessary to determine rather big terms: at us for it two months plus three – on working off of channels left.

About a business case. We, of course, made it, and everything turned out rather simply: we already had a budget model and we represented as how many will cost, including the cost of clouds. For us a cloud and IT in general – not a method to save costs, and to give a method to our business certain advantages, to make it as much as possible transparent and at the same time to earn money. Migration in cloud infrastructure

On migration. We began work in this direction with complete shift of all systems, i.e., almost from scratch. We advise to make it and that who only wants to be connected to the project. Correctly select the sequence of migration, you do not make any changes on the project course if it is already started. Remember that not only your systems should work in a new way, but also the personnel should think in a new way too. Apply service approach to infrastructure.

What risks exist upon transition of IT of infrastructure to clouds?

Possible risks of the introduction in the project:

• technical complexity of migration. It is the main project risk. That to avoid it, prepare detailed plans of migration, involve in process of software makers, test each step, avoid non-return points in plans. Having transferred channels to Germany, we did not face this risk as Germany works completely under the European convention, all procedure is certified and does not go to a section with our legislation;
• data transmission channels. If there is no channel – there is no cloud also. Demand from operators of detailed plans with the weekly statuses;
• dependence on provider. The detailed analysis of top players of the market and also the tender and audit of infrastructure will help to secure themselves against this risk. Study experience of similar projects, whenever possible pay visits to DPC of other companies;
• FZ-152. Take all measures known today;
• Sizing. Constrain desire of engineers to add power "just in case". Require justification.

The closest plans of the company on use of cloud resources

Our next plans – transition to PaaS and VDI. By the end of the year we are going to finish the first project phase and from the following if the business case passes successfully, to start the second. Thus, on we will have very few servers: a part will leave on VDI, and generally the equipment will be provided by the SaaS and IaaS models. It will open new opportunities for our business.

• Source