Центр обработки данных

Physical Infrastructure

A data center is a purpose-built facility housing compute, storage, and networking equipment in a controlled environment. Power redundancy is achieved through uninterruptible power supplies (UPS), backup generators, and dual utility feeds — Tier IV facilities tolerate any single component failure without downtime. Cooling systems maintain 18-27 degrees Celsius and low humidity to prevent condensation and thermal throttling. Physical security involves badge access, biometric locks, CCTV, and man-trap entry vestibules.

Network Architecture Inside

Data center networks are typically arranged in a spine-and-leaf topology: leaf switches connect servers, and all leaf switches uplink to every spine switch, providing uniform LatencyThe time delay for a data packet to travel from source to destination, typically measured in milliseconds (ms). Lower latency is critical for real-time applications like video calls, gaming, and financial trading. and eliminating spanning tree complexity. East-west traffic (server-to-server within the facility) vastly exceeds north-south traffic to and from the internet in modern microservice architectures. BandwidthThe maximum data transfer rate of a network link, typically measured in bits per second (Mbps, Gbps). Bandwidth represents capacity, not actual speed; real-world transfer rates depend on latency, congestion, and protocol overhead. within a data center runs at 10, 25, 100, or 400 Gbps per port, while uplinks to the internet pass through Internet Exchange Point (IXP)A physical facility where multiple ISPs and networks interconnect to exchange traffic directly rather than through upstream transit providers. IXPs reduce latency, lower costs, and improve redundancy. points or TransitA paid arrangement where one network (the customer) pays another (the transit provider) for access to the rest of the internet. Unlike peering, transit provides full routing table access and is the primary way smaller networks connect to the global internet. providers connected via the facility meet-me room.

Connectivity and Redundancy

Enterprise data centers connect to multiple ISPInternet Service Provider. A company that provides internet access to consumers and businesses, assigning public IP addresses and routing traffic to the wider internet. Examples include Comcast, AT&T, and SK Broadband. providers for redundancy, advertising their IP prefixes via BGPBorder Gateway Protocol. The routing protocol that exchanges reachability information between autonomous systems, effectively determining how data traverses the internet. Often called the "postal service of the internet." over each link. Co-location facilities attract many carriers specifically so tenants can choose connectivity providers and establish PeeringA mutual arrangement between two networks to exchange traffic directly and freely (settlement-free peering) at an interconnection point, bypassing third-party transit providers. Peering reduces costs and latency for both parties. arrangements. Cloud regions are built from clusters of data centers with independent power and cooling zones — availability zones — connected by private, high-ThroughputThe actual rate of successful data transfer over a network, measured in bits per second. Unlike bandwidth (theoretical maximum), throughput reflects real-world performance after accounting for latency, packet loss, and protocol overhead. backbone networks that bypass the public internet.