This glossary defines the essential data center and networking terms every IT and business leader should know — from colocation and latency to peering, redundancy, and SLAs. Each definition is written in plain English so you can quickly understand the vocabulary used across infrastructure, connectivity, and cloud decisions.
Use it as a quick reference, or bookmark it alongside the rest of our knowledge base. Terms are organized alphabetically.
Anycast — A routing method where a single IP address is advertised from multiple locations at once, and traffic is automatically sent to the nearest one. It’s widely used by DNS and content delivery networks to cut latency and spread load.
Autonomous System (AS) — A network or group of networks under a single administrative control with its own routing policy, identified by a unique Autonomous System Number (ASN). ASes are the building blocks that exchange traffic across the internet, and operating your own AS is typically a prerequisite for peering at an internet exchange.
Availability — The percentage of time a service or facility is operational and accessible, often expressed as “nines” (e.g., 99.999% uptime, or roughly five minutes of downtime per year). Higher availability means less downtime and is usually backed by a formal SLA.
Backhaul — The transport of data from a local or edge location back to a central network or data center. Minimizing unnecessary backhaul is a core goal of edge connectivity, because every extra mile adds latency.
Bandwidth — The maximum rate at which data can be transferred over a connection, usually measured in bits per second (Mbps, Gbps). More bandwidth means more data can move at once — though actual performance is measured by throughput.
BGP (Border Gateway Protocol) — The routing protocol that networks use to exchange reachability information across the internet. BGP is how peering and transit relationships are established between autonomous systems, and it gives network operators control over how their traffic is routed.
Burstable Billing (95th Percentile) — A common way transit bandwidth is billed, where a provider samples your usage throughout the month and bills based on the 95th-percentile peak rather than your absolute maximum. It lets you burst above your average without paying for the single highest spike.
Cabinet / Rack — The standardized enclosure that houses servers and networking equipment, measured in rack units (U). A full cabinet is commonly 42U. Colocation is often sold by the cabinet, partial cabinet, or private cage.
Carrier-Neutral — A data center that is not owned by or tied to a single network operator, allowing customers to connect to many carriers, clouds, and exchanges of their choice. Neutrality means more options, competitive pricing, and the freedom to change providers as needs evolve.
CDN (Content Delivery Network) — A distributed network of servers that caches content close to end users so it loads faster. CDNs rely on dense interconnection and edge locations to reduce latency and offload traffic from origin servers.
Cloud On-Ramp — A dedicated, private connection from a data center directly into a public cloud provider, bypassing the public internet for more consistent performance and lower egress costs.
Colocation — A service where a business places its own servers and equipment in a third-party data center, renting space, power, cooling, and connectivity rather than building its own facility. It offers enterprise-grade infrastructure without the capital cost of a private build.
Cross-Connect — A direct physical cable linking two parties within the same data center — for example, a customer to a carrier or cloud on-ramp. Cross-connects offer fast, private, low-latency interconnection because the traffic never leaves the building.
Dark Fiber — Fiber-optic cable that is leased to a customer without any provider equipment lighting it, so the customer supplies their own optics and controls the capacity end to end. It offers maximum flexibility and scalability for organizations that want to run their own high-bandwidth links.
Data Center Tiers — A classification system (Tiers I–IV, defined by the Uptime Institute) that rates a facility’s redundancy and expected availability. Higher tiers add redundant power and cooling paths — a Tier III facility supports concurrent maintenance without downtime, while Tier IV adds full fault tolerance.
Disaster Recovery (DR) — The strategy and infrastructure used to restore systems and data after an outage, attack, or disaster, minimizing downtime and data loss. DR plans are often measured against RTO and RPO targets.
Downtime — Any period when a system or service is unavailable. Reducing downtime is a primary goal of redundancy and high-availability design.
Edge Computing / Edge Connectivity — Placing compute and network resources close to users and devices — at the “edge” of the network — to reduce latency for real-time applications like gaming, streaming, and IoT.
Egress Fees — The charges a public cloud provider applies to data leaving its network. Because egress can be a major hidden cost, private cloud on-ramps are often used to reduce it and make costs more predictable.
Failover — The automatic switching to a redundant or standby system when a primary component fails, so service continues without manual intervention. Failover is a key ingredient of resilience.
Fiber Optic — Cabling that transmits data as pulses of light through glass strands, offering far greater bandwidth and lower signal loss over distance than copper. Fiber is the backbone of modern data center and long-haul connectivity.
Hot-Aisle / Cold-Aisle Containment — A data center cooling design that separates the hot exhaust air from server fronts and the cool intake air, improving cooling efficiency and allowing higher power densities per cabinet.
Hybrid IT / Hybrid Cloud — An environment that combines on-premises infrastructure, colocation, and public cloud, using each where it makes the most sense and connecting them into a unified platform. Interconnection is what makes the pieces work as one.
Interconnection — The direct linking of networks, systems, and providers within or between data centers to exchange traffic efficiently, often privately and at low latency. It’s the connecting tissue of hybrid and multi-cloud strategies.
Internet Exchange Point (IXP) — A neutral facility where multiple networks connect to exchange traffic directly with one another, reducing latency and transit costs.
IP Transit — A paid service in which a provider carries a customer’s traffic to and from the entire internet. Unlike peering, transit reaches every destination but is metered by volume.
IPv4 / IPv6 — The two versions of the Internet Protocol that assign addresses to devices. IPv4 uses a limited 32-bit address space that is now largely exhausted; IPv6 uses 128-bit addressing to provide a vastly larger pool. Modern networks typically support both (dual-stack).
Jitter — The variation in latency between data packets. Even when average latency is low, high jitter causes uneven delivery that degrades real-time applications like voice, video, and gaming.
Last Mile — The final leg of connectivity that links a network to the end user or premises. It’s often the most challenging and costly segment to deliver at high performance.
Latency — The time it takes for data to travel from source to destination and back, usually measured in milliseconds. Lower latency means a more responsive experience, and it’s driven largely by physical distance and the number of network hops.
Meet-Me Room (MMR) — A secure, centralized space within a carrier-neutral data center where carriers and customers physically interconnect via cross-connects. The MMR is the heart of a facility’s interconnection ecosystem.
MRC / NRC (Monthly Recurring Charge / Non-Recurring Charge) — Standard colocation and connectivity pricing terms. The MRC is the ongoing monthly fee for a service; the NRC is the one-time setup or installation cost.
Multi-Cloud — The use of services from more than one public cloud provider, often to avoid lock-in, optimize cost, or match each workload to the best platform. Multi-cloud strategies depend heavily on strong interconnection and cloud on-ramps.
N+1 Redundancy — A redundancy model providing one extra component (power, cooling, etc.) beyond what’s needed, so the system keeps running if a single component fails.
2N Redundancy — A redundancy model with a fully duplicated, independent system, providing complete backup capacity. More resilient — and more costly — than N+1.
Network Fabric — The interconnected switching infrastructure that links devices and systems within or across data centers, designed to move traffic with high bandwidth and low latency. Software-defined fabrics allow connections to be provisioned and changed on demand.
On-Net / Off-Net — A connection is “on-net” when it runs entirely over a provider’s own network, and “off-net” when part of it is delivered over another carrier’s infrastructure. On-net connectivity is typically faster to provision and more consistent in performance.
Packet Loss — The failure of data packets to reach their destination, forcing retransmission and degrading performance. Even small amounts of packet loss can noticeably harm real-time and high-throughput applications.
Peering — A direct exchange of traffic between two networks for the destinations they each serve, often settlement-free. Peering frequently takes place at an IXP and can be arranged privately (private peering) or through a shared route server (public peering).
Point of Presence (PoP) — A physical access point where a network connects to the rest of the internet or to other networks — for example, a carrier’s presence inside a data center. More PoPs generally mean more connectivity options.
Power Density — The amount of power available to a cabinet or footprint, measured in kilowatts (kW). Higher-density deployments — driven by modern servers, GPUs, and AI workloads — require careful power and cooling design.
PUE (Power Usage Effectiveness) — A measure of data center energy efficiency, calculated as total facility energy divided by IT equipment energy. A value closer to 1.0 indicates greater efficiency.
Redundancy — The duplication of critical components — power, cooling, network paths — so that the failure of one does not cause an outage.
Remote Hands — On-site technical support provided by data center staff on behalf of a colocation customer — for tasks like rebooting equipment, checking cabling, or swapping hardware — so the customer doesn’t need to travel to the facility. Often available 24/7.
Resilience — The broader ability of an environment to withstand and recover from failures, combining redundancy, failover, and recovery practices.
Route Server — A system at an internet exchange that simplifies peering by letting a network exchange routes with many others through a single connection, rather than configuring each peering session individually.
RTO / RPO (Recovery Time Objective / Recovery Point Objective) — Two key disaster-recovery metrics. RTO is how quickly a system must be restored after an outage; RPO is how much data loss (measured in time) is acceptable. Tighter objectives require more robust DR infrastructure.
SDN (Software-Defined Networking) — An approach that separates network control from the underlying hardware, allowing connections and policies to be managed through software. It underpins virtual interconnection, where links to clouds and partners can be provisioned or torn down on demand.
SLA (Service Level Agreement) — A contract that defines the level of service a provider guarantees — such as uptime percentage, response times, and remedies if targets are missed.
SOC 2 — A widely recognized compliance framework that evaluates how an organization manages data based on five trust principles: security, availability, processing integrity, confidentiality, and privacy.
Throughput — The actual amount of data successfully transferred over a connection in a given time. It reflects real-world performance, which may be lower than the available bandwidth due to congestion, packet loss, or overhead.
Tier 1 Carrier — A network large enough to reach every other network on the internet purely through settlement-free peering, without paying for transit. Access to Tier 1 carriers in a facility is a marker of strong, high-quality connectivity.
Uptime — The amount of time a system or facility is operational, typically expressed as a percentage. The counterpart to downtime.
What is the difference between bandwidth and throughput? Bandwidth is the maximum possible data transfer rate of a connection; throughput is the actual rate achieved in practice, which can be lower due to congestion, packet loss, or overhead.
What does N+1 redundancy mean? N+1 means a system has one more component than it needs to operate, so it can keep running if a single component fails. 2N goes further, fully duplicating the system for complete backup capacity.
What is the difference between peering and transit? Peering exchanges traffic directly between two networks for their own destinations, often for free. Transit is a paid service that carries traffic to the entire internet.
What does carrier-neutral mean? A carrier-neutral data center isn’t tied to any single network provider, letting customers choose freely among many carriers, clouds, and exchanges.
What is a meet-me room? It’s the secure, central space in a data center where carriers and customers physically interconnect through cross-connects. The meet-me room is where a facility’s interconnection ecosystem comes together.
What’s the difference between latency and jitter? Latency is how long data takes to make a round trip; jitter is how much that delay varies from packet to packet. Real-time applications need both to be low and stable.
What are remote hands? Remote hands are on-site data center technicians who perform physical tasks — reboots, cabling, hardware swaps — on your behalf, so you don’t have to visit the facility in person.
What is a good PUE for a data center? Lower is better. A PUE near 1.0 reflects high efficiency; many efficient modern facilities operate well below industry averages, though exact figures vary by design and climate.
What do RTO and RPO mean in disaster recovery? RTO (Recovery Time Objective) is how fast you must restore service after an outage; RPO (Recovery Point Objective) is how much data, measured in time, you can afford to lose. Both shape how a DR strategy is built.
Bookmark this page as a quick reference whenever an unfamiliar term comes up — and explore the rest of our knowledge base for deeper explainers on interconnection, edge connectivity, and building resilient infrastructure. If you’re weighing a colocation or connectivity decision and want to talk it through, get in touch for a fast, flexible quote.
© 2025 Downtown Properties LLC. All rights reserved.