Alexey Malafeev, TGC-1: The disaster-proof DPC based on two data centers should work as a unit
In an interview of TAdviser Alexey Malafeev, the director of the Enterprise of means of supervisory and technology control and information technologies of Nevsky branch of JSC TGC-1 told about implementation of the disaster-proof solution for corporate data center.
What solution for DPC can be considered ideally disaster-proof taking into account specifics of productive activity of TGC-1 company?
Alexey Malafeev: Any disaster-proof solution, irrespective of specifics of activity of the company, assumes, at least two geographically spaced data centers. And these DPCs should work as a unit, i.e. balance automatically loading. Respectively, if one of data centers suddenly ceases to work, the second should have sufficient power to provide operability of a system in general. In respect of distance between DPC for providing such operation mode there is a restriction – no more than 100 km.
Tell, please, shortly about background of the project – when the management set the task about need of creation of disaster-proof DPC? In this connection there was such need?
Alexey Malafeev: Such task directly was not set by the management, need arose naturally. Originally server was in head office, in the historic center of the city. At some point problems with supply with the electric power, sufficiency of the place began to arise, and then moving to new office was planned. Also the decision was made that it is better to make two less protected DPCs, than to construct one even if also superreliable.
What was made within its implementation?
Alexey Malafeev: In connection with specifics of activity of TGC-1, we have own optical cables which are laid between all our objects. For project implementation we selected two stations, extreme on a geographical location, – in the North and in the south of the city. The optics was switched so that between these two objects forward channels of communication appeared. This the first that was made. And the second moment which allowed to construct disaster-proof DPC is system implementation of EMC VPLEX.
What the solution is? Why the choice was stopped on EMC VPLEX? What product did replace?
Alexey Malafeev: EMC VPLEX is the virtualizator of DWH allowing to read out and write at the same time data in geographically spaced data centers. Before a similar system from HP worked for us, but, first, it ceased to arrange on performance and reliability, and secondly, the vendor decided to contract support and development of this product. A choice for benefit of the solution EMC was made owing to uniqueness of a system. In my opinion, he in the market has no worthy competitor. In respect of functionality this solution allows to work with a large number enough cheap disk arrays, allowing to provide identical computing powers based on both data centers. It also allows to implement automatic balancing of loading.
What place is taken by this hardware solution in an IT landscape of the disaster-proof solution? With what equipment it is integrated?
Alexey Malafeev: All disk capacity of both DPCs is available through EMC VPLEX. Respectively, all servers see disks only through this system. In both data centers maintenance in current status of similar disk arrays is performed permanent, in real time. The solution EMC VPLEX is also engaged in it.
Whether are selected in a system the main data center and reserve? Power of each of them?
Alexey Malafeev: We went rather long way of evolution, as a result having refused the concept "the main DPC are reserve DPC". Both data centers are equal, equipped with almost identical set of the equipment which is equally configured and even located indoors approximately in the same places. Respectively, DPCs have approximately identical power characteristics.
On what distance the main and reserve centers are carried?
Alexey Malafeev: Purely geographically, as I already spoke, objects are carried on the opposite ends of the city, definitely I cannot specify distance in kilometers. Other parameter - length of optical fiber between data centers is more important. Today it is about 50 km (originally there were about 70).
Whether the EMC company the examination / specialists at a project implementation stage helped, or she was only an equipment provider?
Alexey Malafeev: When a system was brought into commercial operation, it was the Russia's first project of such scale. Therefore, naturally, specialists of EMC actively participated in implementation of this project. Problems with integration did not arise. Moreover, EMC VPLEX as it appeared, was perfectly integrated even with arrays which officially this solution does not support. Our specialists studied therein, directly in implementation process of the project and at the moment support of a system is performed by forces of own IT department of TGC-1. EMC VPLEX seriously develops at the programming layer, updates constantly come. This year we expect release of new, more powerful hardware solution which we are going to deliver at ourselves this year.
How replacement of the hardware will be performed? A system is in commercial operation …
Alexey Malafeev: Data centers are configured in such a way that replacement of any equipment can be executed without prejudice to operational work at any time.
What do you think of certification of DPC on international standards of Tier?
Alexey Malafeev: All this story with certification of data center on compliance to international standards is rather conditional, in my opinion, and in our case certification is hardly required. On the one hand, we locate the small data centers located not in certain buildings and we do not have special diesel generators. But on the other hand, they are also not necessary to us as they owing to specifics of activity of DPC company are directly on an object for generation of energy. Moreover – objects of the company are united in a power ring, and in case of shutdown of one of objects, energy will come from another. Thus, on formal grounds we would hardly undergo certification, but actually reliability level very high. And from a human factor nobody is insured.
Do you have as at the specialist the view of the Russian market of commercial DPCs?
Alexey Malafeev: Ourselves already actually begin to work at this market. At the same time we do not hand over in lease of space under placement of racks, but on a commercial basis we render to the adjacent organizations services in providing a pool of virtual computing resources. Thus, ourselves gradually become commercial data center. In general on the market, in my opinion, lease of computing powers at commercial data centers is interesting to the small companies which for work need about several tens servers.
30