SAN (storage area network)

A storage area network (SAN) is a comprehensive hardware and software solution designed for centralized, reliable and high-performance storage, management, protection and access to digital data. In the modern world, information is one of the most valuable assets of an enterprise, and a storage system acts as a fundamental element of the IT infrastructure, ensuring its safety and availability. A storage system combines hard disk arrays or solid state drives, controllers, specialized management software, uninterruptible power supplies (UPS), high-density network interfaces and monitoring tools. Unlike a simple set of disks in a server (JBOD — Just a Bunch of Disks), a real storage system has built-in fault tolerance mechanisms, redundancy (RAID arrays of various levels: 0, 1, 5, 6, 10, 50, 60), data replication between sites, automatic recovery after failures and thin provisioning, which guarantees the continuity of business processes even when individual components fail (disk, controller, power supply, fan).

Main functions and capabilities of a storage system

A modern storage system performs a wide range of tasks, the key ones being: consolidation of disparate data from various servers and applications into a single pool for centralized management from a single console, ensuring high read and write speeds for critical systems (databases, ERP, CRM, document management systems), implementing backup and disaster recovery policies (Disaster Recovery), fine-tuning access rights for different categories of users and groups with differentiation of access levels (administrator, operator, user, auditor). The storage system supports such advanced functions as deduplication (removal of duplicate data blocks) to save disk space by up to 10 times, compression to reduce storage costs, creation of instant snapshots for quick data recovery at any point in time without interrupting application operation, thin provisioning for flexible distribution of capacity between applications as needed, encryption of data at rest on disks to protect against physical access, as well as automatic tiered storage (auto-tiering), in which the most frequently used "hot" data is automatically placed on fast SSD drives, and rarely used "cold" data on cheaper HDDs. The use of the scaling technology allows scaling storage capacity almost without limits — from several terabytes to tens of petabytes in one system, and scaling can be performed without stopping the system (online expansion). For Russian organizations working with personal data, the use of a certified storage system is a mandatory requirement of regulators, including the FSTEC of Russia and the FSB of Russia, especially when processing data in state information systems and CII facilities.

Architecture and types of storage systems

There are three main types of storage systems, each with its own architecture and scope of application. SAN (Storage Area Network) is a high-speed specialized storage network operating over Fibre Channel (FC) protocols at speeds of up to 32 Gbit/s and above or iSCSI over Ethernet 10/25/40/100GbE, which provides block access to data at the level of logical volumes (LUN — Logical Unit Number). SAN provides minimal latency (microseconds) and maximum performance (millions of IOPS), which makes it an ideal choice for databases, Exchange email, high-load applications and virtualization platforms (VMware, Hyper-V). NAS (Network Attached Storage) is a file storage connected to a regular Ethernet network and providing access via NFS (for Unix/Linux clients) or SMB/CIFS (for Windows clients) protocols. NAS is easier to deploy and manage, optimal for file servers, user home directories, corporate portals and document collaboration systems. The third type is unified storage systems (Unified Storage), which combine the capabilities of SAN and NAS in one device, simultaneously supporting block (FC/iSCSI) and file (NFS/SMB) access. Such systems are especially convenient for medium and large enterprises seeking to unify infrastructure, reduce administration costs and simplify staff training. Modern storage systems also support the NVMe over Fabrics (NVMe-oF) protocol, which provides access to fast NVMe drives over the network with latencies minimally different from local connection, which is critically important for high-performance computing (HPC) and real-time systems.

Criteria for choosing a storage system for an enterprise

When choosing a storage system, many factors must be considered. Required capacity and performance (number of input-output operations per second — IOPS, and throughput in MB/s), load profile (read/write, block size). Types of supported disks: HDD (large capacity up to 22 TB, low cost per gigabyte, low performance), SSD (high performance up to 1 million IOPS, average cost), NVMe (maximum performance up to 10 million IOPS, minimum latency less than 100 µs). Available connection protocols: Fibre Channel (8/16/32 Gbit/s), iSCSI (1/10/25/40 Gbit/s), NFS v3/v4, SMB 2.0/3.0, NVMe-oF (RoCE v2, FC-NVMe). The level of redundancy and fault tolerance: supported RAID levels (0, 1, 5, 6, 10, 50, 60, RAID-TEC for systems with protection against triple failures), the number of controllers (active-active with load balancing or active-passive), complete duplication of power supplies (N+N), fans and network ports. The presence of data deduplication and compression functions, snapshot support (snapshots for recovery at any point in time), remote replication (synchronous — for zero RPO over short distances and asynchronous — for geographically distributed data centers), volume cloning and business continuity (MetroCluster, Active-Active). Total cost of ownership (TCO), including the cost of equipment, management software licenses, maintenance, electricity consumption and cooling. For state customers, a mandatory requirement is the presence of the storage system in the Unified Register of Russian Radio-Electronic Products.

Storage systems in Russian practice and import substitution

On the Russian market there is a steady trend of import substitution in the storage system segment. Domestic manufacturers (YADRO, Arenadata, Aerosdisk, KNS Irbis) offer storage system lines included in the Unified Register of Russian Radio-Electronic Products and compatible with Russian software, including the Astra Linux Special Edition, RED OS, ALT Linux, Rosa Linux operating systems, as well as the Postgres Pro, Tarantool and Arenadata DB database management systems. Such systems provide a level of reliability and performance comparable to world brands (Dell EMC, NetApp, HPE), and at the same time fully comply with import substitution requirements for state and municipal customers. When building or modernizing a data center, specialists of the Fintech company recommend an integrated approach that includes not only the supply and configuration of a reliable storage system, but also professional services for its integration with the existing infrastructure. Engineers perform a full cycle of work: from auditing the current IT infrastructure and calculating the required performance to installation, commissioning and subsequent warranty and post-warranty maintenance. The installation service allows the customer to get a fully ready-to-use storage system "turnkey" with documentation and staff training. A properly designed and configured data storage system allows not only ensuring reliable preservation of information, but also significantly increasing the efficiency of corporate applications, reducing downtime and disaster recovery time, as well as reducing the total cost of ownership of the enterprise's IT infrastructure in the long term (Total Cost of Ownership, TCO), which is a key indicator of the economic efficiency of any infrastructure project.