Data centers go to the clouds

At a certain stage of development, all large companies are faced with the task of modernizing their own IT infrastructure. When the performance of the existing infrastructure is not enough, many of the managers of large enterprises are thinking about building their own new data center. The search for a building suitable for a data center, personnel selection, design, installation of basic and additional elements of the data center - all this requires a lot of time and material resources. But there is an alternative!

Thanks to the active development of cloud technologies and virtualization, providers have already appeared in Russia that provide the service “Cloud Data Center”, which allows you to completely transfer IT infrastructure to outsourcing, while maintaining a high level of scalability and reliability. Such a service is not cheap, but with a balanced approach, it makes it possible to use the company's resources more efficiently.

Is cloud data center really good, as advertised? Is it true that billing helps to save, and server virtualization is enough to build high-quality cloud infrastructure? We will try to give answers to these and a number of other questions under Habrakat.

')

Who is suitable?

Cloud data center as a service will interest primarily large companies whose activities are not related to software development. For small and medium-sized enterprises, as well as for individuals wishing to "move" to the cloud to solve less complex tasks, it will be useful to get acquainted with Amazon Web Services , but for large players, the capabilities of Amazon at a certain stage of development will not be enough.

Possibilities of Amazon Web Services already described on Habré. Let us mention several advantages of Russian cloud providers focused on large business (using the example of CROC’s Virtual Data Center as an example):
• Amazon technical support in English only.
• Amazon does not have the ability to connect user equipment to the cloud, while this option is available with a Russian provider.
• Amazon does not use network virtualization in its services, while CROC uses an SDN solution based on the OpenFlow protocol.
• In the standard Amazon Elastic Compute Cloud (EC2) configuration, information is stored on Instance storage disks that are located directly on the virtualization servers, which affects security (if the server is on, all useful information flies). However, the Amazon Elastic Block Storage (EBS) service makes it possible to store information on hard drives installed on isolated Storage servers, which significantly increases data security. However, users of Amazon services do not have the opportunity to use high reliability storage networks (SAN) (unlike serious cloud providers, where this option is provided).

What is the profit?

So, what do we get by choosing a cloud data center instead of building our own?

To begin, let's talk about the costs of building a data center for its own needs. CROC employees described in some detail the construction of a data center on Entuziastov Street in Moscow ( “How we built a data center” , “Photo Excursion” , “Data Center Infrastructure” ), this data center can be considered as an example.

Choosing and buying a building, designing, modifying a building and finishing, supplying electricity, installing a cooling system, a pumping station, air conditioners, a security system, a fire extinguishing system, and a server room arrangement is not a complete list of what needs to be done only during the construction phase. In addition to the cost of construction itself, you also need to take into account the fact that you will need to pay for the work of system administrators, network salespeople, a virtualization team, an accountant, procurement staff, and many other employees serving the data center. In addition, you will need to purchase racks and servers, pay for the license of products for virtualization, and these are only those expenses that immediately come to mind.

As you can see the capital required costs. Another factor is that, starting the arrangement of a data center, you have only a rough idea of ​​how fast your business will continue to grow. Laying a large supply of equipment, you can miss a lot by spending a huge amount of money on infrastructure, which later will not be used by a third. Excessive costs for racks and servers will entail extra costs for service support and maintenance, the cooling system, generators and much more equipment, which in the end can be very expensive.

Having stopped on cloud-based data center as a service, you will instantly refuse the capital costs associated with the design and construction of a data center. The expenses of the provider will naturally be included in the tariff for the “cloud data center” service, but if you choose a cloud, your costs will not be one-time, but “smeared” for several years. By selecting a trusted cloud provider, you automatically get an experienced team of IT specialists ready for a wide range of complex tasks.

The cost of an error in choosing the initial configuration for a virtual data center will be much lower than in the case of a physical data center. Thanks to per-second or hourly billing, you can synchronize the cloud with the pace of business development and IT needs, adjusting the configuration of the resources used.
Automation of services and real-time management of a virtual data center is another feature that distinguishes the cloud version from the physical data center. By choosing a good cloud provider, you can create servers, increase or decrease their capacity, add disks, make backups, set up a firewall, and all this remotely using the self-service portal. Moving the corresponding slider of the control panel with the mouse, you can add or reduce the amount of RAM used, bring the processor power to the required value and all this without leaving the office and just a few clicks. Billing the cloud data center immediately responds to changes, and within a second or an hour (depending on the provider’s billing) you will pay for the service according to the changes made to the virtual system configuration.
For more information on what criteria to choose a cloud provider, if the cloud data center service interests you, you can read in two words: “Who can be called a cloud provider or how not to get caught on hoster’s tricks?” And “Practical advice on choosing a cloud provider . ”

One more thing that is worth paying attention to is the level of reliability of the data center. If you plan to build a data center yourself, you will have to take care of the reliability and security of the data center at the design stage. In the case of the choice of the “cloud data center” service, the security and resiliency of the system falls entirely on the shoulders of the provider. On Habré there were already detailed articles on the topic of the TIA 942 standard and level classification from the Uptime Institute, we quote the main thing:
Speaking very roughly, TIER I means that in the event of a breakdown, the data center stops working and the systems are being repaired. TIER II says that the data center runs continuously, but if necessary, stops to service critical nodes or repair breakdowns. In case of single failures, TIER III is repaired without stopping, therefore a lot of things are duplicated (it stops only in really extreme cases when several knots break simultaneously). TIER IV means complete duplication: in fact, two TIER-second data centers are built around the racks in the same building. The cost per square meter, respectively, 15K, 20K, 25K and 40K dollars, but this is for America. We, unfortunately, because of customs and Russian specifics, are 2-3 times more expensive. And then only for really large data centers, where there is economies of scale.

When designing the data center, which we considered above as an example, CROC initially set itself the task of conforming to the Tier III standard according to the Uptime Institute classification . To do this, it was necessary to provide redundancy of critical nodes, ways of obtaining electricity and routes of delivery of the coolant. In addition, according to the requirements of the standard, if any node is decommissioned for maintenance, the full functionality of the facility as a whole must be maintained, the data center must ensure uninterrupted operation for 99.982% of the time (maximum idle time is 1 hour 35 minutes per year). When building your own data center, it is very difficult to ensure such a level of reliability; how KROK can cope with this task can be found in this article . Also, in order to better understand these standards and the standardization procedure itself, it will be helpful to read another topic on this topic .

Something about virtualization

We have already mentioned above that, in addition to server virtualization, solutions for virtualizing data storage systems are already available at this stage of technology development.

Using storage virtualization allows you to combine a large number of physical storage systems into a single logical pool. In other words, all storage systems available in the data center can be combined into a single unit, visible from the outside, for example, as 800 TB of disk capacity. In this case, disk fragments of one virtual server can easily be located on different storage systems of a provider.

When choosing a cloud provider, in addition to software solutions for virtualization, which are used as part of the “Cloud Data Center” service, you will also certainly be interested in the hardware available in the provider’s arsenal. In addition to universal Dell PowerEdge servers , as well as reliable server racks , an argument in favor of choosing one provider or another would also be using Dell EqualLogic disk arrays, which we already wrote on our blog, and new Dell Compellent virtual storage systems as storage systems. Dell EqualLogic and Compellent model lines support all modern storage virtualization solutions and provide data security at a very high level.

And in the bottom line ...

• A virtual data center service provided by several large cloud-based providers may be suitable for large companies whose income is not directly related to the development of their own software.
• The use of this solution allows to significantly reduce capital costs, minimize the risks associated with the cost of IT infrastructure and its maintenance.
• When choosing a major player in the IT market as a cloud provider, you can also connect non-standard equipment together with the use of integrated virtualization capabilities and solve specific problems relevant for your company.
• Another useful bonus that can be obtained by working with a large cloud provider is the operational assistance of the provider’s employees when migrating to the cloud, as well as support for the so-called “hybrid mode of operation”. In this mode, you have the opportunity to work simultaneously with the cloud and the physical IT infrastructure, combining them into one network.

In the next articles of our blog, we plan to return to the topic of virtual data centers and talk about the technical side of things in more detail. In the meantime, we will put an end. Thank you for your attention, stay with us!