Resources
Understand the terms used when choosing and managing hosting infrastructure. Browse definitions for servers, cloud platforms, storage, databases, security, and recovery. Links lead to explanatory resources or clearly labelled hosting services.
168 terms across 21 letters
Artificial intelligence is the field of developing computer systems that perform tasks such as recognizing patterns, understanding language, making predictions, or generating content. It includes machine learning and other approaches. In hosting, AI workloads can involve training models or running inference, with resource requirements determined by the model and application.
An AI agent is a software system that uses artificial intelligence to pursue a goal by selecting actions and interacting with tools, data, or other systems. It may retrieve information, call an API, or carry out a multistep task. Permissions, monitoring, and human approval controls determine which actions it can safely perform.
An AI data pipeline moves and prepares data for model training, evaluation, or inference. Stages can include collection, validation, cleaning, transformation, and delivery to storage or a model. Reliable pipelines track data sources and processing steps, apply access controls, and detect failures or quality problems before unsuitable data reaches the application.
AI inference is the process of using a trained model to produce an output from new input. Examples include generating a response to a prompt, classifying an image, or predicting demand. The model, response-time target, and request volume determine the computing resources required.
AI infrastructure is the combination of computing hardware, storage, networking, and software used to develop and operate artificial intelligence systems. It supports activities such as preparing datasets, training models, and serving predictions. Resource requirements vary widely: some workloads run on CPUs, while others benefit from GPUs, high-capacity memory, and fast connections between servers.
AI model training is the process of using data to adjust a model's parameters so it can perform a task. Depending on the approach, training may use labeled examples, patterns within unlabeled data, or feedback. Computing requirements depend on the model and dataset, while separate evaluation helps establish how well the trained model handles new inputs.
An AI server is a physical or virtual server configured to run artificial intelligence workloads. It combines suitable processors, memory, storage, and software for tasks such as model training or inference. GPUs or other accelerators may be included, but the appropriate configuration depends on the model, dataset, application, and required response times.
An AI workload is a computing task or set of tasks used to build or operate an artificial intelligence application. Examples include preparing data, training a model, and processing inference requests. Each places different demands on processors, memory, storage, and networking, so infrastructure should be sized around the actual task and service requirements.
Anycast is a network addressing method in which the same IP address is announced from multiple locations. Routing directs traffic to a destination selected by network paths and policy, which is not necessarily the geographically closest site. DNS providers and content delivery networks use anycast to distribute requests and support resilience across their infrastructure.
Apache HTTP Server is open-source web server software maintained by the Apache Software Foundation. It serves website content over HTTP and HTTPS and can pass requests to application services. Administrators use its modules and configuration files to control functions such as authentication, logging, and request routing.
An application programming interface defines how software components communicate and request functions or data from one another. Hosting APIs can automate tasks such as creating servers, managing storage, or retrieving account information. Available operations, authentication requirements, request limits, and response formats are described in the provider's API documentation.
Auto scaling automatically adjusts the number or capacity of computing resources according to configured rules, schedules, or demand measurements. A hosting platform might add application instances as traffic increases and remove them when demand falls. Minimum and maximum capacity, startup time, application design, and service limits affect how effectively scaling responds.
Backup as a Service delivers backup capabilities through a service provider, usually with subscription or usage-based pricing. The service can include backup software, storage, scheduling, monitoring, and restoration support. Customers should confirm which systems are protected, how long copies are retained, where data is stored, and who performs and tests restores.
A backup is a recoverable copy of data kept for use when the original is lost, damaged, or unavailable. Backups can cover files, databases, or entire systems. Their usefulness depends on the recovery points retained, protection against deletion or attack, and successful restoration testing.
Backup hosting provides infrastructure for storing backup copies away from the systems they protect. It may supply a backup repository, dedicated storage, or a managed backup environment. The agreement should identify supported backup software, retention, access restrictions, available capacity, and restoration support. Separate credentials and suitable storage protection help reduce the risk of losing both production data and backups.
Bandwidth is the data-carrying capacity of a network connection, usually expressed in megabits or gigabits per second. Actual transfer speeds also depend on congestion, latency, and the systems at each end. Hosting plans sometimes use “bandwidth” to describe a monthly data-transfer allowance, so check whether a stated limit measures capacity or total usage.
A bare metal server is a physical server allocated to one customer, with direct access to its hardware resources. The customer can install an operating system or virtualization software. Hosting providers often use “bare metal server” and “dedicated server” for closely related offerings.
Border Gateway Protocol (BGP) exchanges routing information between networks known as autonomous systems, including internet service providers and large organizations. Routers use this information and configured policies to select paths for traffic. BGP supports internet connectivity and multihoming, but routing choices do not automatically identify the fastest path or verify that every advertised route is legitimate.
Big data describes datasets whose volume, speed of arrival, or variety makes them difficult to process with conventional tools. Examples include large collections of application logs, transaction records, and sensor readings. Big data systems often distribute storage and processing across multiple machines to support analysis, reporting, or machine learning.
Block storage presents data to a server as individually addressable blocks, usually within a volume that the operating system treats as a disk. The server can format that volume with a file system or use it for database storage. Performance depends on the underlying storage and connection.
Business continuity is an organization's ability to maintain essential operations during and after a disruption. Planning covers people, facilities, suppliers, communications, and technology, including alternative ways to deliver critical services. Disaster recovery supports business continuity by restoring IT systems, while the broader continuity plan addresses how the organization continues operating.
A content delivery network (CDN) service uses distributed servers to deliver website content closer to users. It can cache files such as images, stylesheets, and videos, reducing requests to the original hosting server. Some CDNs also accelerate dynamic traffic or provide security features. Results depend on caching rules, content type, and network coverage.
CJIS compliance means meeting applicable requirements of the FBI's Criminal Justice Information Services Security Policy when handling criminal justice information. Controls address areas such as access, authentication, encryption, personnel, and incident response. Hosting arrangements must also satisfy the responsible agency's agreements and oversight; a provider's marketing claim alone does not establish compliance.
Cloud computing delivers computing resources, such as processing, storage, networking, and software, as services accessed over a network. Resources can be provisioned on demand and adjusted as requirements change. Public, private, and hybrid environments use different operating models, while responsibilities for security, maintenance, and data protection depend on the services selected.
Cloud hosting runs websites or applications using computing resources supplied through a cloud platform. Resources can be provisioned and adjusted as workloads change. Infrastructure as a Service (IaaS) is one cloud service model; cloud hosting can also include services that manage more of the application environment.
Cloud infrastructure is the collection of hardware and software that supports cloud computing services. It includes servers, storage, networking, virtualization, and management systems used to allocate resources to workloads. This infrastructure can support public, private, or hybrid cloud environments, with capacity and availability shaped by the underlying architecture.
Cloud migration is the process of moving applications, data, or other IT resources into a cloud environment or between cloud providers. A migration may transfer an existing system largely unchanged or redesign parts of it for new services. Planning includes dependencies, security, data transfer, testing, service cutover, and a rollback approach.
Cloud native describes an approach to building and operating applications that uses cloud capabilities such as automated deployment, flexible scaling, and service-based architecture. Containers, APIs, and microservices are common components, though the implementation varies. Cloud-native applications can run in public, private, or hybrid environments; moving an existing application to cloud hosting alone does not make it cloud native.
A cloud platform brings together computing services and the tools used to provision and manage them. It may offer virtual servers, storage, networking, databases, and application services through a control panel or API. Available features, supported software, billing methods, and the division of management responsibilities vary between platforms.
Cloud portability is the ability to move applications and data between cloud environments while preserving their intended operation. It depends on compatible runtimes, data formats, APIs, and supporting services. Containers and open standards can reduce migration work, while provider-specific features and dependencies can make a move more difficult.
A cloud provider is a company that supplies computing services such as virtual servers, storage, networking, or hosted applications. The provider operates the infrastructure needed to deliver its services. Customer responsibilities for software, access controls, data protection, and recovery depend on the service model and the terms of the agreement.
Cloud storage holds data on infrastructure accessed through a cloud service over a network. It can provide object storage, block volumes, or shared file systems, each suited to different access patterns. Capacity, performance, data location, access controls, and charges for requests or transfers depend on the selected service.
A cluster is a group of connected computers or service instances, called nodes, that work together on a workload. Clusters can distribute processing, pool resources, or help services remain available when a node fails. Their behavior depends on the software, networking, storage, and failure-handling design; having multiple nodes alone does not guarantee uninterrupted service.
Colocation hosting places customer-owned servers in a provider’s data center. The provider supplies agreed facilities such as rack space, power, cooling, connectivity, and physical security. The customer retains ownership of the equipment, while hardware maintenance and any additional management services follow the contract.
A container is an isolated environment for running an application and its required files and dependencies. Operating-system containers share the kernel of the system on which they run, rather than each requiring a separate guest operating system. Their portability and isolation depend on compatible platforms, runtime configuration, resource limits, and security controls.
Container orchestration automates the deployment and operation of containers across computing resources. An orchestration platform can schedule workloads, maintain a desired number of instances, restart failed containers, and coordinate networking and storage. Kubernetes is one example. Application configuration, persistent data, security policies, and the underlying infrastructure still require administration.
A container registry stores and distributes container images used to deploy applications. Developers or build pipelines push images to the registry, and servers retrieve them when starting containers. Registries can be public or private. Access permissions, image scanning, retention, and references such as tags or immutable digests help control which software reaches an environment.
cPanel is a web-based control panel for managing hosting accounts on supported Linux servers. It provides tools for tasks such as managing website files, domains, email accounts, databases, and backups. Its companion product, WebHost Manager (WHM), supports server-level and account administration. Available features depend on the hosting plan and administrator's configuration.
Database hosting supplies the infrastructure on which a database system, such as MySQL or Microsoft SQL Server, runs. The customer may administer the database themselves or purchase management services. Responsibilities for patching, backups, replication, and recovery depend on the agreed service scope.
Database replication copies database changes to other database instances to support availability, read capacity, or recovery. Synchronous replication waits for specified confirmation from another instance; asynchronous replication allows changes to arrive later. Replication lag, failover behavior, and consistency depend on the database and configuration. Replicated deletion or corruption can affect multiple copies, so separate backups remain important.
A data center is a facility that houses servers, storage systems, and networking equipment. It supplies supporting infrastructure such as electrical power, cooling, connectivity, and physical security. Data centers may serve a single organization or multiple customers. Redundancy, maintenance procedures, and the services included vary between facilities.
A data lake stores large volumes of structured, semistructured, and unstructured data for later processing or analysis. It often retains data in its original form and supports multiple tools for analytics and machine learning. Cataloging, access controls, data quality, and lifecycle management are needed to keep the stored information usable and appropriately protected.
Data residency describes the geographic locations where data is stored or processed. In hosting, this can include production databases, replicas, backups, logs, and supporting services. Choosing a server region alone may not establish where every copy or processing activity resides, so organizations should review the provider's architecture, contractual commitments, and available location controls.
Data sovereignty concerns the laws and governmental authority that apply to data and its handling. Storage location, the organizations processing the data, and relevant jurisdictions can affect these obligations. Hosting decisions should consider applicable legal requirements and access arrangements as well as physical location; keeping data in one country does not resolve every jurisdictional issue.
A data warehouse is a system designed to bring together data from multiple sources for reporting and analysis. Data is typically organized and prepared for consistent queries across business activities and historical periods. Its design prioritizes analytical workloads, while ingestion processes, data quality, permissions, and refresh frequency determine how useful the results are.
A distributed denial-of-service attack attempts to make a system unavailable by overwhelming its network, services, or application with traffic from multiple sources. Attacks can exhaust bandwidth, connection capacity, or application resources. Protection may combine upstream traffic filtering, rate limits, application controls, and a response plan suited to the services being protected.
DDoS protection detects and mitigates distributed denial-of-service attacks that attempt to overwhelm a network or application. Techniques include traffic filtering, rate limiting, and routing traffic through scrubbing infrastructure. Coverage differs for network, protocol, and application-layer attacks. Effective protection depends on the services covered, available mitigation capacity, configuration, and the provider's response procedures.
Dedicated hosting is a service that provides one or more physical servers for a single customer's use. The customer receives access to the server's allocated hardware resources without sharing that server with other hosting customers. The service may be managed or self-managed, with maintenance, monitoring, backups, and security responsibilities defined in the agreement.
A dedicated hosting provider supplies physical servers reserved for individual customers, together with agreed data center facilities and connectivity. Providers differ in their hardware options, deployment locations, support arrangements, and management services. The contract should identify responsibilities for hardware replacement, operating system maintenance, security controls, backups, and incident response.
A dedicated server is a physical server reserved for one customer’s use. Its processor, memory, and local storage are allocated to that customer, who can run applications directly or install virtualization software. Operating system maintenance, backups, and monitoring depend on the hosting agreement.
Disaster recovery is the planned process of restoring IT systems and data after a serious disruption. It covers recovery priorities, people, procedures, and the infrastructure needed to resume service. Backups can support recovery, while testing establishes whether applications and their dependencies can return within agreed targets.
The Domain Name System is a distributed naming system that associates domain names with records used by internet services. These records can identify server IP addresses, mail servers, and other configuration information. DNS resolvers retrieve and cache the records, so changes may not reach every user immediately when an existing cached record remains valid.
DNS hosting provides authoritative servers that publish a domain's DNS records, such as website addresses and mail-routing information. It is separate from registering the domain or hosting its website, although one provider may supply all three services. Availability, record-management access, DNSSEC support, and change procedures are important considerations when choosing the service.
Docker is a set of tools for building, distributing, and running containerized applications. Developers package application code and dependencies into images, which a compatible runtime uses to create containers. Docker can make development and deployment environments more consistent. Persistent data, networking, image updates, and access permissions still require configuration and management.
Edge computing processes data near the users, devices, or locations that produce or consume it, rather than sending every task to a central data center. Examples include local analysis of sensor data and regional application processing. It can reduce latency and network traffic, while distributed equipment still needs security, updates, monitoring, and recovery planning.
Elastic computing allows computing resources to expand and contract as demand changes. Capacity may be added or released automatically or through provisioning tools, helping match resource use to workload requirements. Elasticity depends on available capacity, service quotas, deployment speed, and application design; it does not mean that scaling is unlimited or instantaneous.
Electronic protected health information is protected health information created, received, maintained, or transmitted electronically by HIPAA-regulated entities. Examples can include patient records in a database and identifiable health information in electronic messages or backups. The HIPAA Security Rule requires applicable administrative, physical, and technical safeguards to protect its confidentiality, integrity, and availability.
Encryption at rest protects stored data by converting it into an unreadable form without the appropriate cryptographic key. It can apply to disks, database files, objects, and backups. Its effectiveness depends on encryption scope, key management, and access controls. It does not prevent an authorized application or compromised account from reading data that it is permitted to decrypt.
Encryption in transit protects data while it moves between systems over a network. Protocols such as TLS can provide confidentiality, integrity, and authentication for a connection. Administrators must configure trusted certificates and protect each relevant connection; encryption between a browser and a proxy does not automatically protect the separate connection from that proxy to an application server.
Endpoint detection and response monitors activity on devices such as servers and workstations to identify and investigate suspicious behavior. EDR tools collect endpoint data and may support actions such as isolating a device or stopping a process. Coverage depends on supported operating systems, deployed agents, detection rules, and the people or services responsible for responding.
Extended detection and response combines security information from multiple sources, such as endpoints, networks, email, identities, and cloud services, to investigate threats and coordinate responses. It aims to connect related activity that individual tools might assess separately. Supported integrations, detection coverage, automation, and analyst oversight vary between products and deployments.
FedRAMP, the Federal Risk and Authorization Management Program, is a U.S. government program for assessing and monitoring cloud services used by federal agencies. Its security requirements apply to a defined cloud service offering. FedRAMP certification does not replace an agency's own authorization decision or the customer's responsibility to configure and operate its workloads securely.
File storage organizes data into files and directories that applications access through a file system. Network file storage makes those files available to multiple clients using protocols such as NFS or SMB. Permissions, locking, concurrent access, and performance depend on the service and configuration, making it important to check compatibility with the intended application.
A firewall allows or blocks network traffic according to configured rules. It can run on a server, in a network appliance, or as a hosted service. Rules may consider addresses, ports, connection state, or application information. Effective protection depends on appropriate configuration and ongoing maintenance.
File Transfer Protocol is a network protocol for transferring files between a client and a server. Standard FTP does not encrypt credentials or file contents in transit. Encrypted alternatives include FTPS, which adds TLS protection to FTP, and SFTP, a separate file-transfer protocol that operates over SSH.
The General Data Protection Regulation (GDPR) is an EU law governing the processing of personal data. It establishes requirements for lawful processing, appropriate security, accountability, and individuals' rights. It can apply to organizations outside the EU in specified circumstances. Hosting location, processor contracts, access arrangements, and international data transfers all need consideration.
Generative AI uses trained models to produce content such as text, images, audio, video, or software code in response to input. Its output reflects patterns learned during training and the context supplied when it is used. Hosting requirements depend on the model and demand, while accuracy, privacy, and appropriate use require evaluation and controls.
A graphics processing unit (GPU) is a processor designed to perform many calculations in parallel. GPUs handle graphics rendering and can accelerate workloads such as machine learning and scientific computing. Available memory, software support, and the workload’s processing pattern affect whether a particular GPU is suitable.
GPU cloud services provide access to GPU computing resources through a cloud platform. Customers provision GPU-enabled environments for tasks such as AI training, inference, rendering, or scientific computing without buying the underlying hardware. GPU allocation, availability, billing, storage, and networking vary by service, so the deployment model and workload requirements should be checked before provisioning.
A GPU cluster connects multiple GPU-equipped servers so they can work together on distributed workloads. High-speed networking, shared or coordinated storage, and scheduling software help distribute tasks and data across the nodes. Training large models is one use, but applications must support distributed execution, and performance depends on communication overhead as well as GPU capacity.
GPU hosting supplies hosted servers or computing environments with graphics processing units for accelerated workloads. It can include dedicated physical servers or cloud-based resources used for AI, rendering, and scientific applications. GPU type, memory capacity, isolation, software compatibility, and management responsibilities determine whether a particular hosting service suits the workload.
Helm is a package manager for Kubernetes that uses packages called charts to define related application resources. Charts combine templates, configuration values, and other files to support installation and updates. Administrators still need to review the resulting configuration, manage secrets and persistent data, and check that chart versions are compatible with their cluster and application.
High availability describes a system designed to remain accessible despite certain component failures or maintenance events. Common techniques include redundant servers, health checks, replication, and automated failover. Availability depends on the complete service and its dependencies, including networking and storage; redundant hardware alone does not guarantee uninterrupted operation or replace disaster recovery planning.
A HIPAA Business Associate Agreement (BAA) sets out a business associate’s obligations when handling Protected Health Information (PHI) for a covered entity or another business associate. It addresses permitted uses, safeguards, and reporting duties. Signing a BAA is one part of meeting applicable HIPAA requirements.
HIPAA compliance means meeting the requirements of the U.S. Health Insurance Portability and Accountability Act and its implementing rules that apply to an organization. For covered entities and business associates, obligations include protecting health information and following applicable privacy, security, and breach-notification requirements. A hosting service or signed agreement alone does not make the customer's entire operation compliant.
The Health Information Technology for Economic and Clinical Health (HITECH) Act is a U.S. law enacted in 2009 that promoted electronic health information technology and strengthened HIPAA protections and enforcement. It expanded certain business associate obligations and introduced breach-notification requirements. Healthcare hosting arrangements should account for the resulting duties as implemented through applicable HIPAA rules.
HITRUST is an organization that provides a security and privacy control framework, assessments, and certification programs. Its framework brings together requirements from multiple standards and regulations to support risk management and assurance. Certification applies to a defined assessment scope and program; it should not be treated as a blanket guarantee that every system or legal obligation is covered.
A host is a physical or virtual computer that runs software or provides services on a network. In virtualization, the host is the system that runs the hypervisor and its guest virtual machines. In web hosting, the term can also refer to the company supplying hosting services.
Hybrid cloud connects public cloud services with private cloud or on-premises infrastructure so that applications and data can work across environments. Organizations may use it to retain existing systems while adding cloud-based capacity or services. Effective operation requires suitable connectivity, identity management, security controls, monitoring, and clearly defined responsibilities across the connected systems.
Hyper-V is Microsoft's hardware virtualization technology, available in Windows Server and supported editions of Windows. It creates virtual machines with their own operating systems, virtual disks, and network connections. Server deployments can use features such as live migration, clustering, and replication, subject to the Windows edition, hardware, and configuration.
A hypervisor is the software layer that creates and runs virtual machines while managing their access to physical computing resources. Type 1 hypervisors operate directly on hardware, while Type 2 hypervisors run through a host operating system. Resource scheduling, isolation, device support, and management features differ between implementations.
An inference server is software or a server environment that accepts requests, runs a trained AI model, and returns results to an application. It may provide APIs, request scheduling, batching, and model management. Hardware requirements and response times depend on the model, supported runtime, number of simultaneous requests, and deployment configuration.
IT infrastructure is the combination of hardware, software, networks, and supporting facilities used to operate technology services. It includes components such as servers, storage, operating systems, switches, power, and cooling. Infrastructure may be located on premises, supplied by a hosting provider, or distributed across several environments.
Infrastructure as a Service supplies processing, storage, and networking resources that customers use to run software. The provider manages the underlying cloud infrastructure; the customer controls operating systems and deployed applications. Additional management services can change who performs particular tasks, so responsibilities should be documented.
Internet Information Services is Microsoft’s web server software for Windows. It serves web content and hosts applications through a configurable request-processing system. Application pools provide boundaries between groups of applications, but several applications can share a pool and its worker process, depending on configuration.
An intrusion detection system monitors network traffic or host activity for signs of attacks or policy violations. It generates alerts that help administrators investigate suspicious events. Detection can use known signatures, behavior analysis, or other methods. An IDS generally identifies and reports activity; blocking and incident response require additional controls or integrated capabilities.
An intrusion prevention system detects suspicious activity and can take action to stop it, such as dropping network traffic or blocking a connection. Network-based systems often inspect traffic inline. Rules and thresholds need careful tuning because false positives can disrupt legitimate services, while encrypted traffic and deployment position affect what the system can inspect.
An Internet Protocol address identifies a network interface and helps direct traffic to its destination. IPv4 uses 32-bit addresses, while IPv6 uses 128-bit addresses. Public addresses can be routed across the internet; private addresses are used within private networks. A device can have multiple addresses, and an address does not necessarily identify one person or website.
Internet Protocol version 4 (IPv4) provides addressing and packet delivery across IP networks. It uses 32-bit addresses, commonly written as four decimal numbers separated by periods. Hosting services may offer public or private IPv4 addresses, and network address translation allows multiple devices to share a public address. Routing and firewall rules determine which services are reachable.
Internet Protocol version 6 (IPv6) uses 128-bit addresses, usually written in hexadecimal groups separated by colons, to provide a much larger address space than IPv4. Hosting deployments need IPv6 support across the relevant network, operating system, DNS, and security controls. Dual-stack environments support both protocols; enabling IPv6 does not automatically make IPv4-only services compatible.
ISO/IEC 27001 is an international standard specifying requirements for an information security management system. It uses a risk-based approach to establishing, maintaining, and improving information security processes and controls. Certification concerns the organization's defined management-system scope, so customers should review the certificate's covered services, locations, and validity rather than assume every operation is included.
Kubernetes is an open-source platform for managing containerized workloads across a cluster of machines. It uses declared configurations and controllers to schedule applications, maintain desired states, and coordinate resources such as networking and storage. Kubernetes can support resilient deployments, but cluster maintenance, application security, persistent data, and recovery still require deliberate configuration and management.
Kernel-based Virtual Machine is an open-source virtualization technology built into the Linux kernel. It uses hardware virtualization support to run virtual machines with their own operating systems and resources. KVM is commonly combined with QEMU, which provides virtual hardware and other supporting functions, and management tools that handle provisioning, networking, and storage.
A large language model is an AI model trained on substantial amounts of data to process and generate language. Applications use LLMs for tasks such as drafting text, answering questions, summarizing documents, and producing code. Memory, processing requirements, and response times depend on the model and deployment, and generated answers can contain factual errors.
A load balancer distributes incoming traffic across multiple servers or application instances. Depending on its configuration, it can check backend health and direct requests to available destinations. It supports capacity and availability, while session handling, application design, and the load balancer’s own resilience still need consideration.
Machine learning infrastructure supplies the computing resources and software needed to prepare data, train and evaluate models, and operate them in production. It can include experiment tracking, model storage, deployment pipelines, and monitoring alongside servers and databases. Consistent environments, versioned artifacts, and appropriate access controls help teams reproduce results and manage changes.
Managed hosting includes agreed administration tasks alongside the hosting infrastructure. These may cover operating system updates, monitoring, backups, or security controls. Coverage varies between providers and plans. The service agreement should identify who maintains applications, responds to alerts, restores data, and handles work outside the standard scope.
Managed PostgreSQL is a hosting service in which a provider performs agreed administration tasks for a PostgreSQL database. These may include provisioning, updates, monitoring, backups, and failover. Customers usually retain responsibility for their data, queries, permissions, and application behavior. Supported extensions, maintenance windows, recovery options, and service boundaries vary between providers.
MariaDB is an open-source relational database management system originally developed from MySQL. It stores structured data in tables and supports SQL queries, transactions, and applications such as websites and business systems. Although the databases share a history, features and compatibility differ between versions, so migrations should test application behavior, drivers, and database functions.
Microservices are an application architecture in which smaller services handle distinct functions and communicate through defined interfaces. Teams can develop, deploy, and scale those services separately when the design supports it. This approach introduces operational requirements around service communication, authentication, monitoring, deployment coordination, and data consistency across the application.
Microsoft SQL Server is a relational database management system that stores structured data and supports queries using Transact-SQL (T-SQL). Applications use it to read and update records, enforce data relationships, and process transactions. Available features and deployment options depend on the selected version and edition.
Machine learning is a branch of artificial intelligence that develops models by learning patterns from data. These models can support tasks such as forecasting demand, detecting unusual transactions, or classifying images. Training develops the model, while inference applies it to new inputs. Accuracy depends on data quality, model design, and evaluation.
MongoDB is a document database that stores records as BSON documents within collections. It supports flexible document structures, indexes, and queries, with replication and sharding available for suitable deployments. Applications still need an intentional data model, access controls, backups, and monitoring. Availability and consistency behavior depend on topology, configuration, and how the application performs reads and writes.
Multi-cloud describes the use of cloud services from more than one cloud provider. Organizations may choose different providers for particular capabilities, geographic coverage, or business requirements. Using several clouds does not automatically create an integrated or portable environment; identity, networking, security, data movement, monitoring, and costs must be managed across the services involved.
Multi-factor authentication requires evidence from at least two different categories: something you know, something you have, or something you are. A password combined with a security key is one example. Two passwords belong to the same factor category. Protection varies with the authentication method and account-recovery process.
A multi-GPU server contains more than one GPU within a single physical server. Applications can assign separate tasks to different GPUs or distribute a supported workload across them. Performance depends on software parallelism, GPU memory, connections between devices, and data delivery; installing multiple GPUs does not automatically combine their memory or multiply application speed.
MySQL is an open-source relational database management system that uses SQL to store, retrieve, and update structured data. It organizes information into tables and supports applications such as WordPress. Hosting a MySQL database also requires decisions about access controls, backups, software updates, and recovery.
In Kubernetes, a namespace groups and names resources within a cluster, helping separate teams, applications, or environments. Administrators can apply access permissions and resource quotas at namespace level. Namespaces do not cover every resource type and do not provide complete security isolation by themselves; network policies and other controls may also be required.
Network latency is the time taken for data to travel across a network, commonly measured in milliseconds. Measurements may describe one-way delay or the round-trip time between systems. Distance, routing, congestion, and network equipment affect latency. Application response times also include server processing, so a slow request is not necessarily caused by the network alone.
Network File System is a protocol that allows a computer to access files and directories stored on another system over a network. It is commonly used for shared storage in Linux and Unix environments, with support also available on other platforms. Permissions, identity mapping, network access, and protocol configuration determine how clients can use a share.
In cybersecurity, “NIST Framework” commonly refers to the NIST Cybersecurity Framework (CSF), guidance for managing and communicating cybersecurity risk. Its functions cover Govern, Identify, Protect, Detect, Respond, and Recover. NIST also publishes other frameworks and standards, so procurement or compliance requirements should identify the specific publication and scope rather than request an undefined “NIST certification.”
A NoSQL database uses a data model outside the traditional relational-table approach, such as documents, key-value pairs, graphs, or wide columns. These systems suit different access patterns and scaling requirements. “NoSQL” covers many technologies, so query capabilities, transactions, consistency, schema controls, and backup methods must be assessed for the selected database rather than assumed.
NVIDIA Blackwell is a GPU architecture used in a family of products, including data center accelerators for AI and other computing workloads. It introduces capabilities for accelerating model training and inference, with features that vary by product. A hosting specification should identify the actual GPU, memory, server configuration, and software support rather than use the architecture name alone.
NVIDIA H100 is a data center GPU based on NVIDIA's Hopper architecture, designed to accelerate AI training, inference, and high-performance computing. It is available in different configurations with varying memory and connection options. Suitability depends on the server design, model requirements, supported software, and workload, so hosting comparisons should identify the specific H100 configuration offered.
NVIDIA H200 is a Hopper-based data center GPU that uses HBM3e high-bandwidth memory to support demanding AI and high-performance computing workloads. Its larger memory capacity and bandwidth compared with H100 configurations can help memory-intensive applications. Actual performance depends on the model, software, server connections, and workload, rather than memory specifications alone.
Non-Volatile Memory Express is a storage interface and protocol designed for efficient access to non-volatile storage, especially solid-state drives. Local NVMe drives commonly connect through PCI Express, while NVMe over Fabrics extends access across a network. Performance depends on the drive, connection, queue handling, and workload, rather than the NVMe label alone.
Object storage holds data as objects, each containing the data itself, metadata, and an identifier. Applications usually access objects through an API within a bucket or similar container. It is commonly used for images, videos, backups, and large datasets. Object storage has different access behavior from a mounted block volume or shared file system.
A one-click application is a preconfigured software deployment available through a hosting platform or control panel. It automates steps such as installing a web server, database, or content management system. Users still need to review credentials, configure the application, and arrange updates, security controls, and backups after deployment.
Red Hat OpenShift is an application platform built around Kubernetes for developing, deploying, and operating containerized applications. It combines Kubernetes with additional tools for administration, security, networking, and application delivery. It is available through different deployment and service options, with infrastructure responsibilities, supported features, and maintenance arrangements depending on the selected offering.
An operating system manages a computer's processor, memory, storage, devices, and access to hardware. It also provides the environment in which applications run. Server operating systems include Linux distributions and Windows Server. Software compatibility, licensing, security updates, administration tools, and support requirements influence which operating system suits a workload.
Open Virtual Network (OVN) and Open vSwitch (OVS) are related open-source networking technologies. OVS provides software switching that forwards traffic between virtual interfaces and networks. OVN adds logical switches, routers, and access policies that are translated into configurations for OVS. Together, they support software-defined networking for virtualized and cloud environments.
The Payment Card Industry Data Security Standard (PCI DSS) specifies security requirements for environments that store, process, or transmit payment card data, and systems that can affect their security. It covers technical controls and operational practices. A hosting provider's assessed services can support compliance, but customers must establish their own scope, responsibilities, and required validation.
PCI hosting describes hosting services intended to support Payment Card Industry Data Security Standard (PCI DSS) requirements. The provider’s assessed services and responsibilities should be clear. A customer’s compliance also depends on its applications, access controls, processes, and the full scope of its payment environment.
Platform as a Service provides an environment for deploying applications while the provider manages the underlying infrastructure, operating systems, and supported runtime components. Customers manage their application code and data. Supported languages, database services, configuration options, and operational responsibilities vary between platforms.
Plesk is a hosting control panel available for supported Linux and Windows servers. It provides a web interface for managing websites, domains, email, databases, certificates, and hosting subscriptions. Extensions can add application and administration tools. Available features depend on the operating system, license, installed extensions, and settings chosen by the administrator.
A Pod is the smallest deployable unit in Kubernetes. It contains one or more containers scheduled together on the same node, sharing a network context and any configured shared volumes. Pods can be replaced rather than repaired in place, so applications need suitable controllers and persistent storage when data must survive an individual Pod's lifecycle.
PostgreSQL is an open-source object-relational database management system that uses SQL to store, query, and manage data. It supports transactions, constraints, extensible data types, and features for complex applications. A hosted PostgreSQL deployment also requires decisions about version support, access permissions, backups, replication, performance monitoring, and recovery procedures.
Private cloud hosting provides cloud infrastructure for the exclusive use of one organization. It can be operated on the organization’s premises or by a hosting provider. Resources remain available to multiple internal users or teams, with cloud capabilities such as self-service provisioning and resource pooling.
Private hosting is a broad marketing term whose meaning varies between providers. It may describe a virtual private server, dedicated hardware, or a private cloud. Check what is reserved for your organization: a VPS can provide a separate operating system while sharing the physical server with other customers.
Protected health information is individually identifiable health information protected by HIPAA when held or transmitted by a covered entity or business associate, subject to the rule's exclusions. It can concern health conditions, care, or payment and can exist in electronic, paper, or oral form. Not every item of health-related data is automatically PHI under HIPAA.
Proxmox Virtual Environment, commonly called Proxmox VE, is an open-source platform for managing virtual machines and containers. It combines KVM virtualization and LXC containers with tools for storage, networking, backups, and clustering. Administrators use its web interface or command-line tools to manage hosts and workloads, with availability features dependent on the deployment design.
Public cloud provides computing services offered by a provider to many customers. Customers can provision resources such as virtual servers, storage, or managed applications without owning the underlying infrastructure. Physical resources are often shared, with access controls separating customer environments. “Public” describes service availability and does not mean that a customer's data is publicly accessible.
Public key infrastructure is the combination of technologies, policies, and processes used to manage public-key certificates and related trust relationships. Certificate authorities issue certificates that associate public keys with identities, supporting functions such as TLS authentication and digital signatures. Secure operation requires appropriate certificate issuance, private-key protection, renewal, and revocation procedures.
Retrieval-augmented generation supplies a generative AI model with relevant information retrieved from documents, databases, or other sources when answering a request. The retrieved material becomes part of the model's context without necessarily changing its trained parameters. RAG can improve relevance, but source quality, access permissions, retrieval accuracy, and answer validation remain important.
RAID, commonly expanded as Redundant Array of Independent Disks, combines multiple drives into a logical storage arrangement. Different levels use techniques such as striping, mirroring, or parity to change performance, usable capacity, and tolerance of drive failures. RAID 0 provides no redundancy, and even fault-tolerant levels do not replace backups against deletion, corruption, or wider system loss.
Recovery point objective describes the point in time to which data must be recovered after an outage. It is commonly expressed as an acceptable data-loss interval. An RPO of 15 minutes means planning to recover data to a point no more than 15 minutes before disruption.
Recovery time objective is the target time for restoring a system or service after disruption. An RTO of two hours means planning to resume the agreed service within that period. Recovery procedures must account for dependencies, data restoration, checks, and the people needed to complete the work.
Redis is an in-memory data store used for functions such as caching, session storage, messaging, and fast access to structured values. It supports data types including strings, hashes, lists, and sorted sets. Persistence and replication options can help protect data, but durability, recovery behavior, and memory requirements depend on the deployment and configuration.
Replication maintains copies of data or system state in more than one location or instance. It can support availability, distribute access, or assist recovery after a failure. Updates may be copied synchronously or asynchronously, with different consistency and data-loss tradeoffs. Because unwanted changes can also replicate, replicated copies should not automatically be treated as independent backups.
A reverse proxy accepts client requests on behalf of backend servers and forwards them to the appropriate destination. Depending on its configuration, it can provide TLS termination, caching, load balancing, or access controls. Backend connections, forwarded headers, timeouts, and the proxy's own availability need consideration because it becomes part of the application's request path.
A software-defined wide area network (SD-WAN) uses centrally managed policies to control connectivity and traffic across locations. It can combine different connection types and select paths according to application requirements and measured link conditions. Security features and routing behavior vary by solution, while the quality and resilience of the underlying connections still matter.
Security information and event management collects and analyzes security-related logs and events from multiple systems. A SIEM platform helps correlate activity, generate alerts, investigate incidents, and retain evidence. Its effectiveness depends on the data sources connected, detection rules, log quality, retention settings, and the analysts or response processes that act on its findings.
Server migration moves a server's applications, data, or complete operating environment to another physical or virtual server. It may involve a hardware replacement, operating system change, or move to another hosting provider. The migration plan should cover compatibility, data synchronization, configuration changes, testing, service cutover, and recovery if the new environment does not work as expected.
Server virtualization uses a software layer called a hypervisor to run multiple virtual machines on physical hardware. Each virtual machine has its own operating system and allocated resources. This allows workloads to share a server, while performance and isolation depend on resource allocation and configuration.
A service mesh is an infrastructure layer that manages communication between application services. It can provide traffic routing, service-to-service authentication, encryption, and communication monitoring without requiring every application to implement those functions itself. Implementations use proxies or other networking components, and introduce their own configuration, resource requirements, and operational responsibilities.
Single sign-on allows a user to authenticate through an identity system and access multiple connected applications without signing in separately to each one. Applications rely on configured trust relationships and identity information to establish access. SSO does not itself guarantee multifactor authentication or appropriate permissions; those depend on identity policies and application configuration.
A snapshot records the state of a storage volume or virtual machine at a particular time. It can provide a recovery point before a change or help create another environment. How snapshots are stored and restored varies between platforms. A snapshot is not automatically an independently protected backup; its usefulness depends on consistency, storage location, and retention.
SOC 2 is an independent examination and report on a service organization's controls against applicable AICPA Trust Services Criteria. It addresses security and, where included, availability, processing integrity, confidentiality, or privacy. The report's scope, reporting period or date, exceptions, and customer responsibilities matter; “SOC 2 compliance” should not imply a blanket security certification.
SOC as a Service provides outsourced security operations center capabilities, such as monitoring alerts, investigating suspicious activity, and supporting incident response. The provider combines analysts, processes, and security tools to cover agreed systems. Customers should confirm monitoring hours, response authority, escalation arrangements, data access, and which remediation tasks remain their responsibility.
Software as a Service delivers a provider-hosted application that customers access through a browser, client, or API. The provider manages the underlying infrastructure and application software. Customers typically control their data and available settings. Pricing may use subscriptions, usage charges, or other commercial arrangements.
SSD hosting uses solid-state drives to store hosted applications and data. These drives use electronic storage rather than rotating disks and can provide lower storage-access latency and higher random I/O performance than hard disk drives. Actual hosting performance also depends on drive type, storage architecture, resource sharing, and the application's workload.
A storage area network connects servers to shared block storage through a specialized network. Technologies such as Fibre Channel and iSCSI carry storage traffic, allowing servers to access volumes as disks. Access controls and suitable file systems are needed when multiple servers use shared storage.
Throughput is the amount of data successfully transferred or work completed during a measured period. Network throughput is commonly expressed in bits per second, while application throughput might use requests or transactions per second. Actual throughput depends on bottlenecks, protocol overhead, congestion, and processing capacity, and may be lower than a connection's advertised bandwidth.
A vector database stores and searches numerical representations of data called vectors, often generated by an embedding model. It can find items with similar representations, supporting applications such as semantic search, recommendations, and retrieval-augmented generation. Search quality depends on the embedding model, indexing method, filters, and underlying data; numerical similarity does not guarantee factual relevance.
Veeam is a provider of backup, recovery, and data-protection software and services. Its products protect supported virtual, physical, and cloud workloads, with capabilities such as scheduled backups, replication, and file-level restoration. Supported platforms and recovery options vary by product and license. A Veeam deployment still requires appropriate storage, access controls, monitoring, and restoration testing.
Virtualization uses software to represent physical computing resources as logical resources that can be allocated and managed separately. It can apply to servers, storage, networks, or desktops. Server virtualization uses hypervisors to run virtual machines, while other forms use different techniques. Resource sharing can improve utilization, but capacity planning and isolation remain important.
A virtual machine is a software-defined computer with virtual processors, memory, storage, and network interfaces. It runs its own operating system and applications, with a hypervisor managing access to the host's hardware. Multiple virtual machines can share one physical server, although their performance and isolation depend on the resources allocated and the virtualization configuration.
A virtual private server is a virtual machine used to host websites or applications. It has its own operating system and allocated computing resources, while the physical server may also host other customers’ virtual machines. The hosting plan determines which resources are shared and which are reserved.
A virtual local area network (VLAN) logically separates an Ethernet network into distinct broadcast domains while allowing devices to share physical networking equipment. Switch configuration and VLAN tags determine which traffic belongs to each segment. Communication between VLANs normally requires routing, with firewall or access-control policies needed to enforce the intended restrictions.
VMware is a brand of virtualization and infrastructure software. Its technologies include the ESXi hypervisor, which runs virtual machines on physical servers, and vCenter Server, which manages virtualized environments. Specific capabilities, licensing, and supported configurations depend on the products and versions selected.
A virtual private network (VPN) creates a logical connection between devices or networks over another network. Common remote-access and site-to-site VPNs use encrypted tunnels to protect traffic between their endpoints. Authentication, routing, encryption settings, and endpoint security determine the protection provided; a VPN does not automatically make connected systems trustworthy or users anonymous.
vSphere is VMware's server virtualization platform, built around the ESXi hypervisor and centralized management through vCenter Server. It provides tools for managing virtual machines, hosts, storage, and networking. Capabilities such as live migration, high availability, and resource scheduling depend on the selected product entitlement, hardware compatibility, and configuration.
Vulnerability management is the ongoing process of identifying, assessing, prioritizing, and addressing security weaknesses in systems and software. It combines asset visibility, scanning or other discovery methods, risk assessment, remediation, and verification. Priorities should consider exposure and exploitation risk as well as severity, with ownership and tracking for weaknesses that cannot be fixed immediately.
A web application firewall inspects HTTP and HTTPS traffic to help identify and block harmful requests to a web application. Rules can address threats such as common injection attacks or abusive request patterns. Deployment position, encrypted-traffic visibility, and rule tuning affect coverage; a WAF supports application protection but does not replace secure development or patching.
Web server hosting supplies an environment for software such as Apache HTTP Server, NGINX, or IIS to deliver website content over HTTP and HTTPS. The environment may use shared hosting, virtual servers, or dedicated hardware. Responsibility for configuring and maintaining the web server depends on the service.
A website is a collection of related web pages and resources, usually accessed through a domain name. Its content may consist of stored files or be generated by applications and databases when a visitor makes a request. Hosting supplies the environment that delivers the site, while DNS helps visitors locate it.
WordPress is an open-source content management system used to create and manage websites. It provides an editing interface, themes for site design, and plugins for additional functionality. Self-hosted WordPress runs on a compatible web server with PHP and a MySQL or MariaDB database. Updates, access controls, and backups require ongoing attention.
Zero trust security is an approach that does not grant implicit trust solely because a user or device is inside a network or owned by an organization. Access decisions consider identity, device state, requested resources, and policy. Implementations combine authentication, least-privilege access, monitoring, and other controls; zero trust is an architecture and operating approach, not a single product.
No terms match that search. Try a shorter word, or clear the search to browse all 168 terms.
From single servers to fully managed clusters, we'll help you build the ideal infrastructure for your business.
Call or email us now.
Contact an advisor at 866-618-DATA (3282), email sales@atlantic.net, or fill out the form below to get started.
Recognized with the 2026 Excellence in Customer Service Award from the Business Intelligence Group, Atlantic.Net is committed to delivering industry-leading support and an exceptional customer experience.
