IT and data center glossary

DataHouse.net glossary

IT and data center glossary: colocation, cloud servers, dedicated servers and hardware

A practical glossary for infrastructure decisions. The explanations focus on how a term affects sizing, price, SLA, migration, security and future growth in DataHouse services.

Colocation and data center

Terms used when your equipment is placed in a professional facility instead of an office server room.

Data center

A facility designed for production IT: power, cooling, physical security, connectivity, monitoring and operating procedures.

Colocation

A service where the customer places owned hardware in the operator's rack, using data center power, network and access procedures.

Rack unit / U

A height unit for rack equipment. 1U equals 44.45 mm; server depth, rails, power and cabling still need separate verification.

Rack cabinet

A cabinet for multiple servers and network devices. In colocation it is sized by U space, power, cooling, cross-connects and access rules.

Tower server

A non-rack server chassis. It can be colocated, but it needs different space, power and airflow assumptions than a rack-mounted server.

Remote hands

Technical assistance performed by data center staff: checking LEDs, connecting cables, rebooting equipment or replacing agreed components.

SLA

Service level agreement describing availability, response rules and responsibility. For infrastructure it should be linked with monitoring and escalation.

Power feed

A dedicated electrical circuit delivered to a rack or device. In colocation it defines available power, connector type and redundancy options.

UPS

Uninterruptible power supply that keeps equipment running during short grid events and bridges the time before generator power is available.

Generator

Backup power source for longer utility outages. It matters together with fuel policy, load tests, maintenance windows and transfer procedures.

PDU

Power distribution unit inside a rack. Metered or switched PDUs help monitor load and remotely control selected power outlets.

Redundant power

A design with two independent power paths, usually A and B feeds. It works best when servers have dual power supplies.

Hot aisle / cold aisle

Rack layout that separates warm exhaust air from cold intake air, improving cooling stability and energy efficiency.

Cooling capacity

The amount of heat the facility can remove from racks. High-density servers, NVMe and GPU systems need separate cooling checks.

Humidity control

Environmental control that reduces static electricity and condensation risk. It is part of stable long-term hardware operation.

Fire detection and suppression

Systems for early fire detection and extinguishing. In data centers they should protect equipment while limiting damage from the suppression method.

Physical access control

Badges, permissions, logs, procedures and escort rules that decide who can enter the facility, room, cage or rack.

CCTV / monitoring

Camera and operational monitoring used for security evidence, incident review and supervision of critical areas.

Cross-connect

A physical cable connection between customer equipment, operator network, carrier, meet-me room or another tenant service.

Meet-me room / MMR

A dedicated interconnection area where carriers and customers terminate circuits and order cross-connects.

Carrier-neutral data center

A facility where customers can use multiple telecom operators instead of being locked to one network provider.

PUE

Power usage effectiveness, an energy efficiency metric comparing total facility power with IT equipment power.

Tier III

A data center design class associated with concurrent maintenance assumptions and redundant capacity components.

Maintenance window

A planned time for infrastructure or service work. It should be communicated and linked with risk, rollback and customer impact.

Private cage

A separated, access-controlled colocation area for one customer, usually used when physical isolation and compliance matter.

Rack density

Power and heat load per rack. Dense configurations need careful checks of power feeds, cooling, weight and cabling.

Floor loading

Maximum supported weight for equipment in a room or rack area. It matters for dense servers, storage arrays and battery systems.

Colocation models, space and facility types

Additional facility and commercial terms used when comparing rack, cage, suite and provider options.

Server colocation

Server colocation describes a colocation service variant: space, operator support, power, network and responsibility split between customer and data center.

Retail colocation

Retail colocation describes a colocation service variant: space, operator support, power, network and responsibility split between customer and data center.

Wholesale colocation

Wholesale colocation describes a colocation service variant: space, operator support, power, network and responsibility split between customer and data center.

Managed colocation

Managed colocation describes a colocation service variant: space, operator support, power, network and responsibility split between customer and data center.

Colocation facility

Colocation facility describes a colocation service variant: space, operator support, power, network and responsibility split between customer and data center.

Colocation provider

Colocation provider describes a colocation service variant: space, operator support, power, network and responsibility split between customer and data center.

Edge data center

Edge data center is a data center facility type that differs by scale, location, customer profile, connectivity and operational requirements.

Enterprise data center

Enterprise data center is a data center facility type that differs by scale, location, customer profile, connectivity and operational requirements.

Hyperscale data center

Hyperscale data center is a data center facility type that differs by scale, location, customer profile, connectivity and operational requirements.

Carrier hotel

Carrier hotel is a data center facility type that differs by scale, location, customer profile, connectivity and operational requirements.

Internet exchange (IX)

Internet exchange (IX) is an interconnection concept for networks, carriers and customers, important for peering, transit and cross-connects.

Point of Presence (PoP)

Point of Presence (PoP) is an interconnection concept for networks, carriers and customers, important for peering, transit and cross-connects.

Meet-me room (MMR)

Meet-me room (MMR) is an interconnection concept for networks, carriers and customers, important for peering, transit and cross-connects.

Cabinet

Cabinet defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Enclosure

Enclosure defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Open rack

Open rack defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Closed rack

Closed rack defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Half rack

Half rack defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Quarter rack

Quarter rack defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Full rack

Full rack defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Cage

Cage defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Shared cage

Shared cage defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Suite

Suite defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Private suite

Private suite defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

White space

White space defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Raised floor

Raised floor defines how data center space is allocated: from a rack or cabinet to a separated area with access and cabling rules.

Cold aisle

Cold aisle relates to airflow organisation between racks, reducing mixing of cold and hot air and improving cooling efficiency.

Hot aisle

Hot aisle relates to airflow organisation between racks, reducing mixing of cold and hot air and improving cooling efficiency.

Hot aisle containment

Hot aisle containment relates to airflow organisation between racks, reducing mixing of cold and hot air and improving cooling efficiency.

Cold aisle containment

Cold aisle containment relates to airflow organisation between racks, reducing mixing of cold and hot air and improving cooling efficiency.

Rack colocation

Rack colocation is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Cabinet colocation

Cabinet colocation is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Cage colocation

Cage colocation is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Server housing

Server housing is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Server hosting

Server hosting is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Equipment housing

Equipment housing is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Equipment colocation

Equipment colocation is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Cabinet rental

Cabinet rental is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Rack rental

Rack rental is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Cage rental

Cage rental is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Space rental

Space rental is a commercial term for renting space or infrastructure for customer equipment, usually tied to space, power, links and support.

Server roles and compute platforms

Terms that describe what a server does and where it sits in a physical, virtual or clustered environment.

Bare metal server

Bare metal server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

Physical server

Physical server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

Rack server

Rack server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

Blade server

Blade server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

GPU server

GPU server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

AI server

AI server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

Storage server

Storage server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

Database server

Database server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

Application server

Application server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

Compute server

Compute server describes a physical server role or form factor, affecting chassis, power, cooling, disks, network and administration.

Hypervisor host

Hypervisor host is a virtualization or cluster term that defines where a workload runs and how it is scaled or moved.

Virtual server

Virtual server is a virtualization or cluster term that defines where a workload runs and how it is scaled or moved.

Instance

Instance is a virtualization or cluster term that defines where a workload runs and how it is scaled or moved.

Cluster node

Cluster node is a virtualization or cluster term that defines where a workload runs and how it is scaled or moved.

Cloud server, VPS and private cloud

Terms that help compare virtual machines, elastic resources and managed cloud environments.

Cloud server

A virtual server delivered from a cloud platform. It is useful when fast provisioning, flexible resources and easy resizing matter.

VPS

A virtual private server with separated resources. It is a practical start for websites, APIs, panels, test environments and smaller business systems.

vCPU

A virtual processor unit assigned to a VM. The real performance also depends on CPU generation, oversubscription policy and workload profile.

vCore

A commercial or platform name for a virtual CPU core. Always check whether it means reserved compute or shared best-effort capacity.

RAM

Memory available to the operating system and applications. Too little RAM usually causes swapping, slower databases and unstable application behavior.

vMem

Virtual memory assigned to a VM or cloud server. It should be sized with application memory, cache, database buffers and OS overhead.

Swap

Disk space used as emergency memory when RAM is exhausted. It can prevent crashes, but heavy swap usage means poor performance.

Load

A system load indicator showing how much work is waiting for CPU or I/O. High load is not always the same as high CPU usage.

Snapshot

A point-in-time state of a virtual machine or disk. It helps with change rollback, but it is not a full backup and should not replace backup policy.

Backup

A separate copy of data or system state used for recovery. It should define retention, restore tests, RPO, RTO and responsibility.

High availability / HA

An architecture that limits downtime by avoiding single points of failure. It needs more than one component and tested failover procedures.

Private cloud

A cloud environment dedicated to one organisation or workload group, usually chosen for stronger isolation, governance and predictable resources.

Cloud

A model of delivering compute, storage and network resources as a service, usually with faster provisioning than buying and installing hardware.

Public cloud

A cloud platform shared by many customers. It gives broad services and elasticity, but governance, cost control and data location need planning.

Hybrid cloud

A design combining private infrastructure with public cloud or external services. It is common during migration and for backup or burst capacity.

IaaS

Infrastructure as a Service: virtual machines, disks and networks delivered as a service while the customer manages the OS and applications.

PaaS

Platform as a Service: runtime, database or application platform where less system administration is needed than with raw virtual machines.

SaaS

Software as a Service: a complete application delivered over the network, usually paid as a subscription and managed by the provider.

Virtual machine / VM

An isolated operating system instance running on a hypervisor. It behaves like a server but shares physical hardware with other VMs.

Hypervisor

The virtualization layer that runs and isolates virtual machines. KVM, VMware ESXi and Hyper-V are common examples.

KVM

Kernel-based Virtual Machine, a Linux virtualization technology often used for VPS, cloud servers and private cloud platforms.

Hyper-V

Microsoft virtualization platform used for Windows-oriented environments, private clouds and workloads integrated with Microsoft tooling.

VMware ESXi

Enterprise bare-metal hypervisor used for virtualized server platforms. It is often chosen for mature tooling and operational familiarity.

Proxmox VE

Open-source virtualization platform based on KVM and containers, often used for private cloud, lab and managed virtualization clusters.

Virtuozzo

A virtualization and container platform used by some VPS and hosting environments, often associated with service-provider automation.

Parallels

A virtualization and hosting software brand historically connected with service-provider panels and desktop virtualization tools.

Plesk

A hosting control panel for websites, mail, databases, DNS and certificates, often used by agencies and business hosting customers.

cPanel

A popular Linux hosting control panel for managing web hosting accounts, mail, DNS, databases and backups.

DirectAdmin

A lightweight hosting control panel used for WWW, mail, DNS, FTP, databases and reseller hosting environments.

Cluster

A group of servers or nodes working together to increase availability, capacity or manageability of an application or platform.

Node

One member of a cluster, such as a physical server, virtual machine or Kubernetes worker taking part in a larger system.

Container

A lightweight application package isolated at the operating-system level. Containers start quickly but still need orchestration, storage and security.

Docker

A popular toolset for building and running containers. In production it should be paired with registry, updates, secrets and monitoring.

Kubernetes

A container orchestration platform that schedules workloads on nodes, handles service discovery and supports rolling updates and scaling.

Container registry

A repository for container images. Access control, scanning and image retention are important for production use.

Image / template

A prepared operating-system or application image used to create new virtual machines or containers consistently.

Cloud-init

A startup automation mechanism used to configure cloud servers during first boot, such as users, SSH keys, packages and scripts.

Autoscaling

Automatic adjustment of the number of instances or resources. It needs metrics, limits and application readiness for scaling events.

Load balancer

A component that distributes traffic between multiple servers or application instances to improve availability and performance.

Object storage

Storage for files treated as objects, often accessed by API. It works well for backups, archives, media and application assets.

Block storage

Disk-like storage attached to a VM or server. It is usually used for operating systems, databases and application volumes.

Ceph

A distributed storage platform used for object, block or file storage. It needs several nodes and careful capacity planning.

Reserved resources

Compute or memory resources guaranteed for a VM or service. They reduce performance surprises compared with best-effort sharing.

Overcommit / oversubscription

Allocating more virtual resources than physical capacity, relying on the fact that not every workload peaks at the same time.

Noisy neighbor

A neighboring workload consuming shared resources and affecting performance. Isolation and resource guarantees reduce this risk.

API

An application programming interface used to automate provisioning, monitoring, billing or integration with other systems.

Dedicated servers and hardware components

Hardware terms that affect performance, isolation, I/O, failure tolerance and future expansion.

Dedicated server

A physical machine assigned to one customer or workload. It gives stronger isolation and predictable resources than a shared virtual layer.

CPU cores and threads

Processing resources for applications, databases and virtualization. More cores help parallel workloads, but clock speed and CPU generation still matter.

ECC RAM

Server memory with error correction. It reduces the risk of memory-related data corruption and is important for databases and long-running services.

NVMe

A fast storage protocol for SSDs connected through PCIe. It matters for databases, virtualization, search indexes and workloads with high I/O demand.

SSD / SATA

SSD is flash storage; SATA usually means lower throughput and latency than NVMe. It can still be useful for capacity or less demanding data.

HDD

A mechanical hard disk drive. It is useful for capacity and archives, but slower than SSD or NVMe for random I/O.

SSD

A flash-based drive with lower latency than HDD. It is a common choice for application data, databases and virtual machines.

RAID

A disk layout that can improve availability or performance. RAID is not backup; it protects against selected disk failures, not deletion or corruption.

NIC / 10G

Network interface card and link speed. For backup, storage, virtualization and large transfers, 1G, 10G or higher can change the whole design.

IPMI / iDRAC / remote console

Out-of-band management used to power-cycle, see console output or install the OS even when the main operating system is unavailable.

GPU

A graphics/accelerator card used for AI, rendering, video processing or specialised computing. It changes power, cooling and server chassis requirements.

VRAM

Memory on a GPU card. It limits the size of AI models, rendering scenes and data batches that can fit on the accelerator.

NVIDIA Blackwell

A generation of NVIDIA AI accelerators for demanding model training and inference. It raises requirements for power, cooling, chassis and drivers.

AI accelerator

A specialised card or module for machine learning workloads. Selection depends on VRAM, supported frameworks, power and expected model size.

CUDA

NVIDIA's parallel computing platform used by many AI and GPU-accelerated applications. Driver and library versions must match the workload.

OpenCL

An open framework for parallel computing on GPUs, CPUs and accelerators. It matters when workloads should not depend only on CUDA.

SPARC AI appliance

A specialised appliance or server profile for smaller AI models and dedicated inference tasks, sized by accelerator, RAM, storage and network.

Intel Xeon

Server CPU family commonly used in enterprise servers. Choice depends on generation, cores, clock, memory channels and platform features.

AMD EPYC

Server CPU family known for high core counts, memory bandwidth and PCIe lanes, useful for virtualization, databases and dense workloads.

Bare metal

A physical server used without a shared virtualization layer. It is chosen for isolation, predictable performance or custom virtualization.

Motherboard

The main server board connecting CPU, memory, disks, network cards, management controller and expansion slots.

CPU socket

The physical CPU mounting point on the motherboard. Socket count affects maximum CPU, memory channels and expansion options.

CPU generation

The processor family and age. Newer generations can improve performance, power efficiency, security features and virtualization support.

Clock speed

CPU operating frequency. It matters for workloads that do not scale well across many cores.

CPU cache

Fast memory inside or near the CPU. It can improve database, virtualization and compute workloads with repeated data access.

NUMA

A memory architecture in multi-socket servers where CPU locality affects latency. It matters for databases and virtualization tuning.

DDR4

Previous-generation server memory still widely used in production systems. It can be cost-effective for stable workloads.

DDR5

Newer server memory generation with higher bandwidth and density, used with newer CPU platforms.

DDR4 / DDR5

Server memory generations. They differ in speed, density, power use and compatibility with CPU and motherboard platforms.

DIMM slot

A memory slot on the motherboard. Slot count and supported module sizes define maximum RAM capacity.

Hot-swap bay

A disk bay that allows replacing a drive without powering down the server, if the chassis and controller support it.

Backplane

A board connecting disk bays with the controller. It decides how drives are wired and whether SATA, SAS or NVMe is supported.

HBA

Host bus adapter used to connect disks to a server, often preferred when software RAID or distributed storage manages the disks.

LSI RAID

A common family of hardware RAID controllers. Firmware mode, cache and battery or supercapacitor protection matter for production use.

Hardware RAID

A dedicated controller handles RAID logic. It can include cache and battery protection, but adds controller dependency.

Software RAID

The operating system or storage platform handles RAID logic. It is flexible and transparent, but consumes host resources.

SAS

Enterprise storage interface often used for server disks and backplanes. It supports reliable operation and higher queue depth than SATA.

IOPS

Input/output operations per second. This metric matters for databases, mail servers, virtualization and busy application storage.

Storage latency

Delay of a storage operation. Low latency is often more important than peak throughput for databases and transactional systems.

PSU

Power supply unit converting facility power for the server. Power rating and efficiency must match CPU, disks, NICs and GPU load.

Redundant PSU

Two or more power supplies allowing a server to continue running after one PSU or power feed fails.

PCIe

Expansion bus used by NVMe disks, GPUs, NICs and HBAs. Lane count and generation limit maximum throughput.

Boot disk

Disk or volume containing the operating system. It should be separated from data layout when restore and maintenance matter.

Operating systems, panels and databases

Software platform terms that appear when choosing a VPS, cloud server, dedicated server or managed service.

Linux

An open-source operating system family widely used for servers, cloud, containers, databases and network services.

Unix

A family of operating systems and design principles that influenced Linux, BSD and many enterprise server environments.

Windows Server

Microsoft server operating system used for Active Directory, Remote Desktop, Microsoft workloads and Windows-based applications.

Debian

A stable Linux distribution often chosen for servers because of its conservative release model and large package ecosystem.

Ubuntu Server

A popular Linux server distribution with broad documentation, cloud images and frequent use in web, DevOps and container environments.

CentOS

A former RHEL-compatible Linux distribution. Many older production systems still run CentOS and need migration planning.

AlmaLinux

A RHEL-compatible Linux distribution often used as a CentOS replacement for enterprise and hosting workloads.

Rocky Linux

A RHEL-compatible Linux distribution created as another migration path for CentOS-style server environments.

RHEL-compatible

A Linux distribution compatible with Red Hat Enterprise Linux packaging and behavior, useful for enterprise software support.

Package manager

A tool such as apt, dnf or yum that installs, updates and removes software packages.

Kernel

The core of the operating system controlling hardware, memory, processes, filesystems and network stack.

systemd

A Linux service manager used to start, stop and supervise services, timers and system targets.

LAMP

Linux, Apache, MySQL/MariaDB and PHP stack commonly used for websites and business applications.

LEMP

Linux, Nginx, MySQL/MariaDB and PHP stack, often selected for high-traffic web applications.

MySQL

A popular relational database used by websites, applications and business systems.

MariaDB

A MySQL-compatible relational database fork, commonly used in Linux server and hosting environments.

Percona Server

A MySQL-compatible database variant focused on performance, diagnostics and operational features.

PostgreSQL

An advanced open-source relational database often selected for complex queries, consistency and application features.

Database replication

Copying database changes to another server for availability, read scaling or disaster recovery.

Galera Cluster

A synchronous clustering technology used with MySQL-compatible databases such as MariaDB or Percona XtraDB Cluster.

Database dump

A logical database export used for migration, backup or testing, but often slower to restore than physical backup.

Connection pool

A pool of reusable database connections that reduces overhead and protects databases from too many new sessions.

Query cache / buffer pool

Database memory areas that influence read performance. They must be sized together with RAM and workload profile.

Service panel

A management interface for hosting, mail, DNS, databases or server administration, such as Plesk, cPanel or DirectAdmin.

Network, security and continuity

Network and operational terms that matter during production migration and after the service is launched.

Public IP

An address reachable from the Internet. Production services also need firewall rules, DNS, reverse DNS, monitoring and abuse handling.

BGP / ASN

Routing mechanisms used by network operators. They matter for multihoming, routing policy, Looking Glass checks and resilient connectivity.

Firewall

A network or host-level policy deciding what traffic is allowed. It should be tied to service inventory, logging and change control.

Anti-DDoS

Protection against traffic floods and volumetric attacks. It should be planned with routing, monitoring, thresholds and incident procedures.

VLAN

A logical network separation layer. It helps isolate management, production, backup, storage and customer environments.

Throughput / bandwidth

The amount of data that can pass through a link. It affects backup windows, migrations, replication and user experience.

RPO / RTO

Recovery point objective and recovery time objective. They define acceptable data loss and acceptable restoration time.

DNS

The system translating domain names into IP addresses and other records. Production DNS needs redundancy, correct delegation and monitoring.

Reverse DNS / PTR

A DNS record mapping an IP address back to a name. It is especially important for mail servers and operational reputation.

MX record

A DNS record pointing to mail servers for a domain. It must match mail routing, spam filtering and fallback design.

SPF

A DNS policy listing servers allowed to send mail for a domain. It helps reduce forged sender abuse.

DKIM

A cryptographic signature added to email so receivers can verify that a message was authorised by the domain.

DMARC

A domain policy combining SPF and DKIM results with reporting. It helps protect domains against impersonation.

TLS / SSL

Encryption layer used by websites, APIs, mail and panels. Modern configuration needs valid certificates and safe protocol versions.

Certificate

A digital credential binding a domain or service name with a public key. Expiry, chain and algorithm strength must be monitored.

Reverse proxy

A server placed in front of applications to terminate TLS, route requests, cache, filter or hide backend topology.

WAF

Web Application Firewall filtering HTTP traffic to reduce common web attacks. It complements secure code and patching, not replaces them.

IDS

Intrusion Detection System that detects suspicious activity and raises alerts without necessarily blocking traffic.

IPS

Intrusion Prevention System that can actively block detected malicious traffic according to policy.

VPN

Encrypted network tunnel for administrators, branches or systems. It should use MFA and limited access scopes.

IPsec

A VPN technology often used for site-to-site encrypted tunnels between networks.

WireGuard

A modern VPN protocol known for simple configuration and good performance.

Peering

Direct traffic exchange between networks, often used to reduce latency, improve paths and lower transit dependency.

IXP

Internet Exchange Point where operators exchange traffic. It can improve routing quality and reduce dependency on upstream transit.

IP transit

Paid Internet connectivity through an upstream provider that carries traffic to and from the global Internet.

Uplink

A network connection from a server, rack or switch to an upstream switch, router or provider network.

LACP

Link Aggregation Control Protocol used to combine multiple physical links into one logical bundle for capacity or redundancy.

MTU

Maximum transmission unit, the largest packet size on a link. Wrong MTU can cause hard-to-diagnose connectivity issues.

QoS

Quality of Service rules that prioritise selected traffic types, useful when voice, backups and production traffic share links.

Monitoring

Continuous observation of availability, performance, logs and alerts. It should be connected with escalation and responsibility.

Syslog

A standard way to send logs from systems and network devices to a central place for analysis and retention.

SIEM

Security Information and Event Management platform that correlates logs and alerts from many systems.

Bastion host

A hardened jump host used to administer internal systems without exposing every server directly to the Internet.

MFA

Multi-factor authentication requiring more than a password. It is important for panels, VPN, admin access and privileged accounts.

Zero Trust

A security model based on explicit verification, least privilege and continuous evaluation instead of trusting a network location.

Patch management

Controlled process of applying security and stability updates to operating systems, applications and devices.

NIS2

EU cybersecurity directive affecting risk management, incident reporting and security governance for many important organisations.

GDPR / RODO

Personal data protection rules. Infrastructure projects must consider data location, access, backups, logs and processor agreements.

Connectivity, transport and link parameters

Carrier, switching and link-performance terms used in colocation, IP transit and interconnection projects.

Backbone

Backbone describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

Transit

Transit describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

Public peering

Public peering describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

Private peering

Private peering describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

Interconnection

Interconnection describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

Interconnect

Interconnect describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

Dark fiber

Dark fiber describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

Metro Ethernet

Metro Ethernet describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

Ethernet handoff

Ethernet handoff describes how networks or carrier services connect, affecting routing, latency, redundancy and transport cost.

WAN

WAN is a network architecture term used for separation, tunnelling, site connectivity or private networks.

LAN

LAN is a network architecture term used for separation, tunnelling, site connectivity or private networks.

VXLAN

VXLAN is a network architecture term used for separation, tunnelling, site connectivity or private networks.

MPLS

MPLS is a network architecture term used for separation, tunnelling, site connectivity or private networks.

SD-WAN

SD-WAN is a network architecture term used for separation, tunnelling, site connectivity or private networks.

Edge router

Edge router is a data center network topology element; its role affects bandwidth, redundancy, failures and traffic paths.

Core router

Core router is a data center network topology element; its role affects bandwidth, redundancy, failures and traffic paths.

Aggregation switch

Aggregation switch is a data center network topology element; its role affects bandwidth, redundancy, failures and traffic paths.

Top-of-rack switch (ToR)

Top-of-rack switch (ToR) is a data center network topology element; its role affects bandwidth, redundancy, failures and traffic paths.

Spine switch

Spine switch is a data center network topology element; its role affects bandwidth, redundancy, failures and traffic paths.

Leaf switch

Leaf switch is a data center network topology element; its role affects bandwidth, redundancy, failures and traffic paths.

Dedicated Internet Access (DIA)

Dedicated Internet Access (DIA) describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Internet uplink

Internet uplink describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Network uplink

Network uplink describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Backhaul

Backhaul describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Symmetric bandwidth

Symmetric bandwidth describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Asymmetric bandwidth

Asymmetric bandwidth describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Port speed

Port speed describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Access port

Access port describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Trunk port

Trunk port describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Burst bandwidth

Burst bandwidth describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Committed bandwidth

Committed bandwidth describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

CIR

CIR describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

EIR

EIR describes a link parameter or commercial model and should be separated from real application performance and SLA guarantees.

Throughput

Throughput is a transmission quality metric affecting backup, VoIP, replication, web applications and data migration.

Bandwidth

Bandwidth is a transmission quality metric affecting backup, VoIP, replication, web applications and data migration.

Latency

Latency is a transmission quality metric affecting backup, VoIP, replication, web applications and data migration.

Packet loss

Packet loss is a transmission quality metric affecting backup, VoIP, replication, web applications and data migration.

Jitter

Jitter is a transmission quality metric affecting backup, VoIP, replication, web applications and data migration.

DDoS protection

DDoS protection means protection against volumetric and application attacks, planned with routing, filtering and incident procedure.

Power, cooling, resilience and operations

Facility operations, redundancy, monitoring, certification and remote-service terms used in professional data centers.

Dual power feed

Dual power feed is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

A feed

A feed is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

B feed

B feed is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Battery backup

Battery backup is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Diesel generator

Diesel generator is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

ATS

ATS is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

STS

STS is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Rack PDU

Rack PDU is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Busway

Busway is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Power strip

Power strip is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Grounding

Grounding is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Earthing

Earthing is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Power redundancy

Power redundancy is a data center power term, important for A/B paths, emergency operation, safety and power measurement.

Cooling

Cooling describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

Precision cooling

Precision cooling describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

CRAC

CRAC describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

CRAH

CRAH describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

In-row cooling

In-row cooling describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

Liquid cooling

Liquid cooling describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

Direct-to-chip cooling

Direct-to-chip cooling describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

Immersion cooling

Immersion cooling describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

Airflow management

Airflow management describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

Containment

Containment describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

Free cooling

Free cooling describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

HVAC

HVAC describes a cooling or environmental control technique, important for high rack density, GPU, NVMe and 24/7 operation.

Access control

Access control belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

Multi-factor authentication

Multi-factor authentication belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

Biometrics

Biometrics belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

Video surveillance

Video surveillance belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

Security guard

Security guard belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

Mantrap

Mantrap belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

Visitor management

Visitor management belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

Audit trail

Audit trail belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

Intrusion detection

Intrusion detection belongs to physical or access security and helps control who can access infrastructure, when and under which rules.

N

N describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

N+1

N+1 describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

N+2

N+2 describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

2N

2N describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

2N+1

2N+1 describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Fault tolerance

Fault tolerance describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

High availability

High availability describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Resilience

Resilience describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Failover

Failover describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Disaster recovery

Disaster recovery describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Business continuity

Business continuity describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Geo redundancy

Geo redundancy describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Active-active

Active-active describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Active-passive

Active-passive describes a redundancy level or continuity model and should be linked with tests, monitoring and recovery procedures.

Tier I

Tier I is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

Tier II

Tier II is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

Tier IV

Tier IV is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

Uptime Institute

Uptime Institute is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

ANSI/TIA-942

ANSI/TIA-942 is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

ISO 27001

ISO 27001 is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

ISO 22301

ISO 22301 is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

PCI DSS

PCI DSS is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

SOC 1

SOC 1 is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

SOC 2

SOC 2 is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

SOC 3

SOC 3 is a standard, tier or compliance report used to describe security, availability, continuity or operational controls.

Smart hands

Smart hands is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Rack and stack

Rack and stack is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Hardware replacement

Hardware replacement is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Cable management

Cable management is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Equipment installation

Equipment installation is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Equipment removal

Equipment removal is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Asset management

Asset management is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Inventory management

Inventory management is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Migration

Migration is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Relocation

Relocation is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

Decommissioning

Decommissioning is a data center operational service supporting installation, maintenance, migration or removal of customer equipment.

DCIM

DCIM relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

Environmental monitoring

Environmental monitoring relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

SNMP

SNMP relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

Telemetry

Telemetry relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

Power monitoring

Power monitoring relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

Temperature monitoring

Temperature monitoring relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

Humidity monitoring

Humidity monitoring relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

Capacity planning

Capacity planning relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

Alerting

Alerting relates to monitoring or facility capacity planning, helping detect issues before they affect production services.

Storage, cabling, equipment and commercial parameters

Storage, cabling, network-equipment, SLA metric and pricing terms that often appear in colocation offers.

SAN

SAN describes a storage type or use case, affecting performance, availability, access protocols and backup strategy.

NAS

NAS describes a storage type or use case, affecting performance, availability, access protocols and backup strategy.

DAS

DAS describes a storage type or use case, affecting performance, availability, access protocols and backup strategy.

File storage

File storage describes a storage type or use case, affecting performance, availability, access protocols and backup strategy.

Backup storage

Backup storage describes a storage type or use case, affecting performance, availability, access protocols and backup strategy.

Archive storage

Archive storage describes a storage type or use case, affecting performance, availability, access protocols and backup strategy.

Router

Router is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

Switch

Switch is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

Gateway

Gateway is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

Appliance

Appliance is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

Network interface card (NIC)

Network interface card (NIC) is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

Fibre Channel

Fibre Channel is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

SFP

SFP is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

QSFP

QSFP is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

DAC cable

DAC cable is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

Patch panel

Patch panel is a network or transport hardware element; selection affects bandwidth, compatibility, redundancy and serviceability.

Structured cabling

Structured cabling is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Patch cord

Patch cord is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Fiber optic cable

Fiber optic cable is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Single-mode fiber

Single-mode fiber is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Multi-mode fiber

Multi-mode fiber is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Copper cable

Copper cable is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Cat6

Cat6 is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Cat6A

Cat6A is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Cat7

Cat7 is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

MPO

MPO is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

LC connector

LC connector is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

SC connector

SC connector is a data center cabling term, important for distance, bandwidth, rack order and troubleshooting.

Availability

Availability is an SLA or incident-handling metric that should be clearly defined in the contract and measured by monitoring systems.

Uptime

Uptime is an SLA or incident-handling metric that should be clearly defined in the contract and measured by monitoring systems.

Downtime

Downtime is an SLA or incident-handling metric that should be clearly defined in the contract and measured by monitoring systems.

MTBF

MTBF is an SLA or incident-handling metric that should be clearly defined in the contract and measured by monitoring systems.

MTTR

MTTR is an SLA or incident-handling metric that should be clearly defined in the contract and measured by monitoring systems.

Service credit

Service credit is an SLA or incident-handling metric that should be clearly defined in the contract and measured by monitoring systems.

Response time

Response time is an SLA or incident-handling metric that should be clearly defined in the contract and measured by monitoring systems.

Resolution time

Resolution time is an SLA or incident-handling metric that should be clearly defined in the contract and measured by monitoring systems.

U (rack unit)

U (rack unit) is a data center unit or infrastructure metric used for planning space, power, cooling and billing.

kW

kW is a data center unit or infrastructure metric used for planning space, power, cooling and billing.

kVA

kVA is a data center unit or infrastructure metric used for planning space, power, cooling and billing.

Ampere

Ampere is a data center unit or infrastructure metric used for planning space, power, cooling and billing.

Volt

Volt is a data center unit or infrastructure metric used for planning space, power, cooling and billing.

Watt

Watt is a data center unit or infrastructure metric used for planning space, power, cooling and billing.

WUE

WUE is a data center unit or infrastructure metric used for planning space, power, cooling and billing.

Power density

Power density is a data center unit or infrastructure metric used for planning space, power, cooling and billing.

Power allocation

Power allocation is a billing or service-scope item that should be checked before comparing colocation offers.

Cross-connect fee

Cross-connect fee is a billing or service-scope item that should be checked before comparing colocation offers.

Remote hands service

Remote hands service is a billing or service-scope item that should be checked before comparing colocation offers.

IP transit service

IP transit service is a billing or service-scope item that should be checked before comparing colocation offers.

How to use the terms in a purchase decision

Glossary FAQ