How to provide disaster tolerance of IT systems using DPC and clouds? TADetails

Ensuring disaster tolerance: top trends

"The fault tolerance of infrastructure is a guarantee of a constant dostupnost customer services and as result – success of business. It is relevant at the same time and for the large companies, and for SMB. Digitalization, development of online services, communication non-feet with consumers and actually flash-like responses to their requests are those factors which will promote popularity of services of DR (disaster recovery) and HA (high availability). At the same time we see that the bigger amount of requests for services in ensuring disaster tolerance comes from averages on the scale of the companies. In our opinion, it speaks rather simply: large players in "corpulent years" laid the foundation in the form of the reliable reserved computing systems that cannot be told about young people, but fast-growing market participants" — the head of outsourcing of computing infrastructures of Croc company Renat Sayfutdinov told.

According to the expert, the choice of these or those technical means put in architecture of the disaster-proof solution can differ depending on the scale of the company, its budget on IT, requirements to the level of fault tolerance and operability of services in critical situations. Replication of data at the level of DWH, a clustering at the level of databases and applications (if it is necessary to provide complete continuity) and also implementation of a specialized software for replication of data belongs to the main options of project implementation (when safety of systems given more important than instant recovery). The last can be used including based on cloud infrastructure.

Large business is able to afford replication at the level of DWH. Medium and small business, on the contrary, often pays attention to cloud solutions because it is cheaper.

"The continuity of business processes, reputational risks, a comprehensive protection of data, SLA, RTO and RPO — even more often the companies of different levels ask a question of ensuring fault tolerance and continuity of IT services. With arrival of understanding of all aspects heads of the companies and Chief information officers even more often are interested in services of DR and HA, providing us with interesting projects and tasks" — Dmitry Ivanyuk, the leading technical expert of Mont group notes.

Own DPC or outsource?

This question especially material: not all companies are able to afford to construct DPC with the due level of fault tolerance. The cost of small data center of racks will be on 10-12 60-80 million rubles, construction costs of larger DPCs are measured by billions. It not only large one-time investments into infrastructure which could be redirected on business development, but also the need for search and deduction of the skilled staff occupied in infrastructure maintenance. It is also a headache of the Chief information officer who should keep in mind that each 3-5 years it is necessary to carry out upgrade of technical solutions. In terms of payback lease of DPC in five-year perspective is more profitable, than creation of own data center. And along with it is is incomparably more convenient in the resource management plan.

But, selecting outsourcing of DPC, there is other question: on what characteristics of commercial data center builds to pay attention that was not excruciatingly painful during the times and resources spent for migration.

"Current trends of business development say that the speed of decision-making and availability worldwide of IT infrastructure of the company are one of the key tasks facing the management of the enterprise at any moment. However service of difficult IT systems a highly skilled IT command means existence of heavy expenses for business that often for medium and small business is a key factor when choosing outsourcing services. At the same time modern DPCs, having all necessary resources conforming to international standards, reliable "iron" and software solutions will provide the companies to all necessary for creation of IT infrastructure. And thanks to convenient tools it is quite possible to cope with such system by forces of one qualified employee" — Dmitry Ivanyuk says.

"First of all it is necessary to pay attention to level of the rented DPC, its certification according to international standards and following to the best practices in the field of operation of data center. In general, all these parameters are considered by certificates of Uptime Institute Tier III — Constructed Facility, the DPC confirming compliance to the Tier III standard in terms of an engineering component, and Gold Certification of Operational Sustainability confirming the high level of operation of an object. Taking place in the data center having these certificates, clients of outsourcing services can be sure of quality of the rendered service and save themselves from need of excess checks" — Renat Sayfutdinov confirms.

Experts also advise to avoid work with released DPCs when the data processing center is actually handed to the customer through the third hands. It can cause additional administrative risks especially as precedents in the market already happened.

As technically there is a reservation of infrastructure in DPC

The first option — to use outsourcing DPC as reserve for a corporate system or infrastructure. The second option — to give on outsourcing both the main platform, and reserve.

"Reservation begins with communication channels. But if the client accommodates in our network of data centers, then between them the fiber optic ring is already laid. Further clusters at the level of a server part, DWH and so on are created — Renat Sayfutdinov lists. — We implement projects on ensuring "turnkey" disaster tolerance, i.e. we not only provide with firmness - places, but also we project all infrastructure, we provide its support and process of switching in case of failures, we unroll the equipment and software for backup. In particular, we use solutions of CommVault as well proved tools for data protection from losses. Products from CommVault are applied not only to reservation of big infrastructures, they are suitable also for small installations: user devices, physical and virtual servers and even mailboxes".

The first stage of any project Croc becomes audit of infrastructure of the company at the level of information systems. After that business determines requirements to fault tolerance by two basic indicators: RTO (recovery time objective is the maximum allowed time of unavailability of systems) and RPO (recovery point objective is recovery point of data after failure).

On the next stage reserve infrastructure is projected and the specific technologies meeting the requirements of business are selected. There is an installation further, and time in half a year-year test switchings between platforms and verification of backup copies are made. Specific implementation time and duration of each of stages depend on the volume of infrastructure and the company.

"A classical example of use of commercial DPC as reserve — our project for Inteza bank. For it we designed reliable infrastructure in addition to the available data center of the customer. In Avon company there was other situation: the customer placed with us and the main, and reserve platforms. We provide them infrastructure on model of the managed service and we provide its permanent availability. Including — we are responsible for support and switching from one platform to another in case of failures" — Renat Sayfutdinov told.

By words Dmitry Ivanyuka, now based on the solution CommVault several scenarios for ensuring the maximum availability of services in infrastructure of customers are implemented. IaaS provides availability of leased virtual infrastructure and applications, and RBaaS allows customers to have an opportunity to rent software solutions for use at the local capacities.

Disaster tolerance using a cloud: in what difference?

Quite often customers select use of cloud or hybrid infrastructure when ensuring disaster tolerance. In the first case it is about placement of services and systems based on the cloud platform distributed between territorially remote data centers. The Croc company implemented such project, in particular, for network of Castorama hypermarkets. In the second case the company has own and/or outsourcing DPC, and the reserve platform is placed in a cloud. Experts call this option flexible in terms of pricing: business will pay only for those resources which it actually consumes in case of activation of the reserve platform.

"How the second scenario looks technically? We reserve the customer's infrastructure — it can be as the virtualized environment, and physical servers. In installation process there is at first a replication of all data from the main DPC of the customer in a cloud, and then, in process of change of data in an initial system, they change also in a cloud. The customer pays only for licenses of solutions of replication, a communication channel and for the volume of the stored data. In case of some problem with the main platform, we activate reserve on a call or the request. The reserve platform, ready to test switchings, at such scenario prepares in two weeks" — Renat Sayfutdinov summarizes.

"Thanks to the solution CommVault we can provide migration of infrastructure of the customer from one environment of virtualization in another. Make converting of local virtual machines in a format of public clouds (Amazon and Azure) and back. Provide the most effective protection of applications and safety of the employees of customers given on notebooks where they would not be" — Dmitry Ivanyuk adds.

According to the expert Croc, today in Russia services on ensuring disaster tolerance it is most demanded by banks, retail chain stores and the industrial companies. In the financial sector and production the requirements for business continuity imposed by regulators push to high demand. In other segments, including retail, creation of failsafe infrastructures or their lease from DPC and a cloud is explained by the growing competition. It gives the grounds for forecasting of growth for the corresponding services (first of all, outsourcing), approximately for 10-15% annually.