Designing the physical infrastructure of the data center - from science to art
Among the stages of the life cycle of the data center operation stage occupies a special place. Because the mistakes that were made at the design and construction stage of the facility, the operational team will have to “clear up” over the years. If all the other stages of creating a data center take several months, the operation of data centers lasts for years, and this period is extended: the generation of data centers built in the mid-2000s has become morally obsolete in five years, and the current generation of data centers is being built for 10–15 years.
According to statistics, 60 percent of errors that occur at the stage of operation are laid at the design and construction stage. Therefore, to reduce the number of design and construction errors with which the operation service has to deal, you need to think about the initial stage of creating the data center.
Quality control of design and construction directly affects the data center, which enters the phase of operation. Accordingly, the design and construction phase should be independently examined at all reference points.
In foreign practice, an independent examination was seriously reinforced by the development of an insurance institution. An independent organization provides quality control and insures the project itself in terms of financial risks. In Russia, this possibility is not yet talkable. Our risks lie with the customer, who is ultimately responsible for everything. And without independent control, design solutions that need to be executed qualitatively, by careless movement of the designer’s thoughts or the builder’s hands can turn a TIER III level data center into a level II or even I object. Therefore, quality control is one of the cornerstones not only of the design stage subsequent stages.
')
The customer, although he is often not a specialist in the field of data centers, as the most interested person is simply obliged to understand what he will end up with from the project. And an independent expert should be a link, and one that the customer could trust 100 percent. That is, the customer must be technically competent enough to understand what the designer is laying into this project.
One of the most acute problems, the solution of which could significantly reduce the number of errors, is due to the fact that the design and operation departments usually do not work together: as a rule, the operation department joins only at the last stage of the creation of the object, which means that it cannot affect the progress of its construction.
During the operation in the data center, more than one generation of equipment can change, so it is very difficult to plan exactly the equipment that will be in the data center in ten years. Many designers can correctly draw, but few people think about what will happen in 15 years and how this object will be operated day after day.
Another stumbling block is a compromise between complexity, quality of the project and price. The real experience and qualifications of the designer are constantly faced with this. It is obvious that, without being fully responsible for the project itself, the controlling organization will not be able to fully discharge the responsibilities assigned to it.
In addition, another problem is that designers today do not think about the cost of ownership. They have no incentive to think that Capex will significantly exceed Opex over time and how much it will cost for 15 years. The financial analysis is most often carried out by a consulting organization, or again this is the customer’s own expertise.
In addition, the project must have a reserve. A back-to-back project, without a stock, may be ideal on paper and might appeal to an investor. But then the service of operation will commemorate the dashing designer. One of the arts that a designer or project manager must demonstrate is to build a stock and convince the investor that this stock is needed.
No even the most prudent designer can take all these factors into account. Therefore, in order for the project to be acceptable, participation in the creation of a data center's own service for operating the enterprise is necessary. And if the designer, in addition to the technical solution, also developed an operation program, he would have understood this process better and could have foreseen the risks.
In practice, in data centers, many defects are eliminated during operation. But defects are not always terrible, if they are eliminated cheaply. This is the art of the designer - not to make large, costly mistakes. And minor mistakes will always be forgiven for him and will continue to work with him.
Of course, the designer needs to know the standards, it is necessary to follow the best practices, you need to have experience. However, there is still such a thing as the art of design. To avoid many mistakes can the one who owns this art.
According to statistics, 60 percent of errors that occur at the stage of operation are laid at the design and construction stage. Therefore, to reduce the number of design and construction errors with which the operation service has to deal, you need to think about the initial stage of creating the data center.
Quality control of design and construction directly affects the data center, which enters the phase of operation. Accordingly, the design and construction phase should be independently examined at all reference points.
In foreign practice, an independent examination was seriously reinforced by the development of an insurance institution. An independent organization provides quality control and insures the project itself in terms of financial risks. In Russia, this possibility is not yet talkable. Our risks lie with the customer, who is ultimately responsible for everything. And without independent control, design solutions that need to be executed qualitatively, by careless movement of the designer’s thoughts or the builder’s hands can turn a TIER III level data center into a level II or even I object. Therefore, quality control is one of the cornerstones not only of the design stage subsequent stages.
')
The customer, although he is often not a specialist in the field of data centers, as the most interested person is simply obliged to understand what he will end up with from the project. And an independent expert should be a link, and one that the customer could trust 100 percent. That is, the customer must be technically competent enough to understand what the designer is laying into this project.
One of the most acute problems, the solution of which could significantly reduce the number of errors, is due to the fact that the design and operation departments usually do not work together: as a rule, the operation department joins only at the last stage of the creation of the object, which means that it cannot affect the progress of its construction.
During the operation in the data center, more than one generation of equipment can change, so it is very difficult to plan exactly the equipment that will be in the data center in ten years. Many designers can correctly draw, but few people think about what will happen in 15 years and how this object will be operated day after day.
Another stumbling block is a compromise between complexity, quality of the project and price. The real experience and qualifications of the designer are constantly faced with this. It is obvious that, without being fully responsible for the project itself, the controlling organization will not be able to fully discharge the responsibilities assigned to it.
In addition, another problem is that designers today do not think about the cost of ownership. They have no incentive to think that Capex will significantly exceed Opex over time and how much it will cost for 15 years. The financial analysis is most often carried out by a consulting organization, or again this is the customer’s own expertise.
In addition, the project must have a reserve. A back-to-back project, without a stock, may be ideal on paper and might appeal to an investor. But then the service of operation will commemorate the dashing designer. One of the arts that a designer or project manager must demonstrate is to build a stock and convince the investor that this stock is needed.
No even the most prudent designer can take all these factors into account. Therefore, in order for the project to be acceptable, participation in the creation of a data center's own service for operating the enterprise is necessary. And if the designer, in addition to the technical solution, also developed an operation program, he would have understood this process better and could have foreseen the risks.
In practice, in data centers, many defects are eliminated during operation. But defects are not always terrible, if they are eliminated cheaply. This is the art of the designer - not to make large, costly mistakes. And minor mistakes will always be forgiven for him and will continue to work with him.
Of course, the designer needs to know the standards, it is necessary to follow the best practices, you need to have experience. However, there is still such a thing as the art of design. To avoid many mistakes can the one who owns this art.
Source: https://habr.com/ru/post/225173/
All Articles