Technology Glossary

Plain definitions of the terms that come up in infrastructure, cloud, security and compliance. Each one is written to answer the question someone looking the term up actually has.

150 terms

A

Active database

The production database serving real-time read and write traffic. What separates it from a standby or archive copy is that it carries the live workload.

Microsoft's directory service. It defines and authorises network resources — user accounts, computers, printers and groups — from one place, and is where sign-on, group policy and access control are centralized in most enterprise environments.

Agility

The ability to adapt quickly to changing business needs. On the IT side that means standing up a new environment in hours rather than weeks — and being able to reverse a decision once it is made.

Antivirus

Security software that detects, blocks and removes malicious code. Modern products go beyond signature matching to behavioral analysis.

AP controller

The device or software that manages wireless access points centrally — channel selection, power levels, client roaming and configuration distribution all run from one place.

API

Application programming interface — the defined contract through which two pieces of software talk to each other. Most integrations are built on APIs today, which is why API security has become a heading of its own.

Autoscaling

Automatically adding and removing resources in response to demand. Configured well it opens capacity at peak and lowers cost off-peak; configured with the wrong thresholds it does neither.

Fault management

The process of detecting, classifying, resolving and recording a fault. In a mature practice the real output is not the closed ticket but the root cause found and the recurrence prevented.

Network security

Protecting data and systems on a network against unauthorised access, attack and malicious software. Firewalling, segmentation, monitoring and access control are the core tools under this heading.

B

B2B (business to business)

Commerce between businesses. What separates it from consumer sales is that the buying decision belongs to a committee rather than a person, and the cycle is long.

Back end

The part of an application the user never sees: server logic, database and integrations. The interface is the front end; the layer feeding it is the back end.

Backup-as-a-Service (BaaS)

Backup taken as a service. Rather than building and running the infrastructure, you define retention and recovery targets and the provider operates it. The clause to insist on is reporting of restore tests.

BDDK (Turkish banking regulator)

Türkiye's Banking Regulation and Supervision Agency. For banks, procuring cloud counts as outsourcing, and only a private cloud or a community cloud serving banks alone may be used.

Read furtherSee also: Private cloud

Big data

Datasets too large, too fast-moving or too varied to process with conventional tools. The definition is less about size than about existing methods no longer being enough.

Bug

A defect that makes software behave differently from what was expected. Catching one before release is far cheaper than after — which is the case for test automation.

Cloud computing

Compute, storage and application resources delivered over the internet, on demand and paid for by consumption. In Türkiye the cloud decision also carries a data localization dimension.

Cloud infrastructure

The compute, storage, network and virtualization layer that carries cloud services. The user sees virtual resources; underneath sit physical servers, arrays and switches.

Cloud server

A server running on virtualized resources and managed remotely, with no hardware to buy. Being able to create and destroy it in minutes is what separates it from a physical server.

Cloud storage

Storing data in a service reached over the internet. It brings cost and scale advantages; the question to ask when deciding is which country and which facility the data sits in.

Read further

Information processing cycle

The path from collecting data to processing, storing and turning it into meaningful output. Each step carries its own security and retention requirements.

SAP Basis management

Managing the technical layer of SAP systems: installation, system copy, transport routes, performance tuning, patching and user administration. It owns the infrastructure side of SAP, not the functional side.

C

Cache

Holding frequently accessed data temporarily in a faster layer. Used well it delivers the largest performance gain available; used with the wrong invalidation rule it produces the hardest bugs.

CDN

Content delivery network. It serves static content from the server geographically closest to the user, lowering latency — and reducing both load and attack surface on the origin.

Cookies

Small pieces of data websites store in the browser. Uses range from session management to tracking, and because they can carry personal data they may require notice and consent under data protection law.

CPU

Central processing unit — the component that executes instructions. In virtualized environments the ratio between physical and virtual cores directly determines performance.

Multi-cloud

Using more than one cloud provider together. It reduces dependence on a single vendor but raises operational complexity and makes cost tracking harder; without a clear reason it does not pay off.

D

Data lake

A pool where structured and unstructured data is stored raw, with no schema imposed up front. The flexibility is the advantage; ungoverned, it turns into a swamp nobody can describe.

Data migration

Moving data from one system or storage platform to another. The risk lies less in the transfer than in the cutover window and in verifying integrity once it is done.

Read further

DDoS

A distributed denial of service attack, making a target unreachable with traffic from many sources. The defense has to be configured before the attack, not during it.

Deep learning

The branch of machine learning that uses multi-layer neural networks to learn patterns from large datasets. It is the method behind the leap in vision, speech and language processing, and training it demands serious GPU capacity.

Read further

DevOps

A way of working that removes the wall between development and operations. It rests on automation, small frequent releases, shared ownership and measurement — it is an operating model, not a tool.

Read further

Digital transformation

Rebuilding business processes around technology. Digitising an existing process is not transformation; transformation is the process itself changing because the technology made a different one possible.

Disaster recovery

Bringing systems and data back after a major outage, data loss or disaster. Having a plan is not enough; what counts in an audit and in a real event is a recovery time measured by exercise.

Disaster recovery as a service (DRaaS)

Disaster recovery taken as a service. Instead of building and running a secondary site, replication and failover capability come from the provider. What to measure is not the monthly fee but the recovery time achieved in an exercise.

Domain name

The human-readable form of a web address. Its expiry, DNS records and ownership belong in the corporate asset inventory — a domain allowed to lapse can be a brand lost.

Hardware

The physical components of a computing system: servers, storage, network devices, end-user devices. Managing their lifecycle — warranty, firmware, refresh and disposal — is part of security too.

E

Edge computing

Processing close to where data is produced rather than in a central data center. It is chosen where latency is critical and moving raw data to the center is expensive.

Elasticity

A system's ability to adapt to changing demand and requirements. It differs from scalability in covering not just growth but contraction and change of direction.

Email management

Operating corporate mail: account lifecycle, mail flow security, quotas, archiving and retention. Getting SPF, DKIM and DMARC right is the most basic step in stopping mail spoofed in your name.

Integration

Connecting different systems at the data and process level. The part most often skipped in design is not the happy path but error handling and retry policy.

F

Disaster recovery site

The secondary facility that takes over workloads when the primary data center becomes unusable. Its value is set not by distance or hardware but by the measured time it takes to fail over to it.

Flexi cloud

A cloud consumption model where resources flex up and down with need, built on variable consumption rather than a fixed capacity commitment.

Physical server

A server running directly on hardware, not virtualized. Still preferred in certain high-performance and licensing scenarios, at the cost of any capacity change requiring a hardware purchase.

Ransomware

Malware that encrypts data and demands payment. What determines recovery is not that a backup exists but that it was stored where an attacker could not delete it.

Read furtherSee also: Backup

G

Firewall

A system that inspects network traffic against defined rules and blocks unauthorised access. The most common problem in the field is a rule set that accumulates until nobody dares touch it.

H

Hosted Exchange

Microsoft Exchange delivered as a hosted service. Rather than running the server in-house, the mail infrastructure runs on the provider's side.

Hosting

Running a website, application or data on a provider's infrastructure. What to compare is not price but guaranteed resources, redundancy and the support model.

HTTPS

Web traffic encrypted with TLS. It prevents data being read or altered in transit, and is now expected as the default rather than the exception.

Hybrid cloud

A model combining on-premises infrastructure with cloud. In Türkiye the most common rationale is keeping regulated data local while the rest runs in the cloud.

Hypervisor

The layer that lets several virtual machines run on one physical server. VMware ESXi, KVM and Hyper-V are the common examples; today the choice is less a strategic decision than a choice of carrier.

I

Business continuity

Critical business processes continuing to run through a disruption or crisis. Broader in scope than disaster recovery: it covers people, suppliers and process as well as technology.

IaC (infrastructure as code)

Defining and building infrastructure in code rather than by hand. It removes drift between environments and makes a change reviewable before it is applied.

Read further

IDE

Integrated development environment — code editing, debugging and build tooling brought together in one interface.

Infrastructure-as-a-Service (IaaS)

Infrastructure — servers, storage, network — delivered as a service. The operating system and everything above it stays the customer's responsibility, and where that boundary sits belongs in the contract.

Internet

The global network connecting devices and networks over shared protocols.

Internet gateway

The point that governs traffic between a network and the internet — where address translation, routing and usually security inspection are applied.

Internet of Things (IoT)

Sensors and devices producing and managed over a network. The real security difficulty is that most of them cannot be patched and stay in service for years.

ISO 20000

The international standard for IT service management. It defines a management system for how services are designed, delivered and improved.

ISO 27001

The information security management system standard. It mandates no particular product or configuration; it requires you to assess risk, select controls and evidence that you applied them. The certificate runs three years, with an annual surveillance audit.

Read further

IT (information technology)

The technologies used to collect, process, store and transmit information.

ITIL

The widely used good-practice framework for IT service management, defining incident, problem, change and release processes. It is guidance rather than a standard.

Requirements analysis

Measuring what is actually needed before a project starts. Skipping it does not produce the wrong solution so much as the right solution applied to the wrong problem.

J

JavaScript

The programming language that makes web pages interactive. It runs in the browser and, through Node.js, on the server as well.

K

Capacity planning

Measuring current resource use and projecting future need. Decisions taken because a disk filled up always cost more than decisions taken on schedule.

Container

A lightweight unit of isolation that packages an application with its dependencies and shares the host kernel. Unlike a virtual machine, it carries no operating system of its own.

Kubernetes

The open-source platform that deploys, scales and manages containerised applications. Its operational weight matches its power; without the capability to run it, it creates more problems than it solves.

KVKK (Turkish data protection law)

Türkiye's personal data protection law. It ties processing to a legal basis and conditions transfer abroad. For organizations using cloud, the point missed most often is where backups are held.

Read furtherSee also: Data privacy

Mixed workload

Applications of different character sharing the same infrastructure. This is where the noisy neighbour problem comes from: an I/O-heavy workload degrading a latency-sensitive one beside it.

L

Linux

The open-source, Unix-like kernel and the distributions built around it. It is the dominant server platform, and the choice of distribution matters for support lifetime and patch policy.

LLM (large language model)

An AI model trained on very large text corpora that can understand and generate language. In enterprise use the question is less the model's capability than which data goes where, and what that means for compliance.

Read further

Load balancer

The component distributing incoming traffic across several servers. It does more than spread load: health checks take a failing node out of rotation, which raises availability.

Logging

Recording system and application events. Three things get asked in an audit: coverage, the integrity of the record, and how quickly it can be produced on request.

Read further

LUN

A logical unit presented to a server by a storage system. It abstracts the physical disks; the server sees it as a single volume.

M

Cost advantage

A solution delivering the same outcome at a lower total cost than the alternative. Comparisons built on the license or hardware line alone are usually misleading.

Machine learning

Methods that predict and classify by learning from data rather than being explicitly programmed. Model quality depends almost entirely on data quality.

Monitoring

Continuous observation of systems, networks and applications. Collecting and monitoring are not the same thing: a metric nobody looks at will pass an audit and prevent nothing.

MSP (managed service provider)

A provider that takes on the operation of IT infrastructure under contract. Ask before price: what the SLA measures, who is on call at night, how you access records, and what happens when you want to leave.

MTTR (mean time to repair)

The average time from an incident starting to it being resolved. Not to be confused with response time: opening a ticket is a response, not a resolution.

N

NAT

Network address translation — mapping private network addresses to a public address on the way out to the internet.

Network

Devices connected so they can exchange data. Network design carries a security dimension as much as a performance one: segmentation is the most effective limit on how far an incident spreads.

Network virtualization

Abstracting physical network resources in software and presenting them as logical networks — which is what makes micro-segmentation and per-workload policy possible.

Node.js

The runtime that executes JavaScript server-side. Its event-driven architecture lets it carry large numbers of concurrent connections efficiently.

O

Operational burden

The ongoing cost to the team of keeping a system running. When choosing a technology this belongs in the calculation alongside the license fee.

Operations

Everything that keeps systems running day to day: monitoring, incident response, patching, backup and change management.

Optimization

Reworking a system or process to deliver the same result with fewer resources. Optimising without measuring usually means speeding up the wrong thing.

Scalability

A system's ability to absorb increasing load. Vertical scaling grows the same machine, horizontal scaling adds machines — and the architecture decides up front which is available to you.

P

PaaS (Platform-as-a-Service)

A platform for building and running applications, delivered as a service. Infrastructure management stays with the provider; in exchange, platform lock-in rises.

Private cloud

A cloud model where resources are allocated to a single organization. Under Turkish banking regulation it is one of the two models available to banks; the other is a community cloud serving banks only.

Proxy server

A server that mediates between client and destination, used for access control, caching and traffic logging.

Public cloud

A cloud model where resources are shared across customers and delivered over the internet. For regulated workloads in Türkiye its use can be restricted.

Q

Queue

A structure where work accumulates to be processed in order. It buffers integrations between systems — and an unmonitored queue depth turns into a backlog that grows silently.

R

Replication

Continuously copying data from a source to one or more targets. A replica being up is not enough — what needs measuring is how far behind it is.

Reseller hosting

A model that lets purchased hosting resources be resold to others.

Router

The device that forwards packets between networks, deciding which path a packet takes from its routing table.

RPO (recovery point objective)

The maximum data loss, measured in time, acceptable in an outage. It sets replication frequency, and it is calculated from business cost rather than from what the technology can do.

RTO (recovery time objective)

How quickly a system must be back after an outage. In most organizations the RTO on paper and the RTO measured in an exercise are different numbers; the measured one is the real one.

S

Continuous monitoring

Uninterrupted observation of systems with automatic alerting on threshold breaches. For an alert to be meaningful it has to be tied to impact rather than to a symptom.

Cybersecurity

Protecting information systems, networks and data against attack. In Türkiye, organizations within scope of Law No. 7545 carry live obligations to provide logs, report vulnerabilities and report incidents without delay.

Read further

Penetration testing

Controlled testing of a system's weaknesses using a real attacker's methods. It differs from vulnerability scanning in showing whether what was found is actually exploitable.

SaaS (Software-as-a-Service)

Software used over the internet with nothing to install. The point most often missed on the data protection side is that most SaaS tools are hosted abroad, which makes their use a data transfer.

SD-WAN

Software-defined management of the wide area network. It uses multiple links together and steers traffic by application priority, bringing cost and flexibility gains on branch connectivity.

Service desk

The single point of contact for user requests and incident tickets. The same issue arriving repeatedly is not a support problem but a root cause left unclosed.

SLA (service level agreement)

The contract annex defining the service level a provider commits to. "99.9%" says nothing on its own: ask what counts as up, whether planned maintenance is excluded, and what happens on breach.

Snapshot

A record of a system or volume at a point in time. It is not a substitute for backup: it lives on the same storage, and if that storage is lost the snapshot goes with it.

SSL / TLS

The protocol that encrypts communication over the internet. Although the name SSL persists, what is actually in use today is TLS. Certificate expiry is among the most common causes of outage.

Switch

The network device that forwards frames between devices on the same network. Managed models add VLANs, port security and traffic prioritisation.

Virtual server

A server with its own operating system running on physical hardware through a hypervisor. It shares resources but runs isolated.

Virtualization

Abstracting physical resources in software and presenting them as several logical ones. It is the basis of server consolidation, and platform choice becomes a live question again whenever a licensing model changes.

T

Technology

The methods, tools and systems used to get work done.

Time to first byte (TTFB)

The time between a request being sent and the first byte arriving. It is the most direct indicator of server-side latency and feeds the speed metrics search engines assess.

Tokenisation

Replacing sensitive data with a token that carries no meaning of its own. Widely used to protect card data; unlike encryption, the real value never sits in the system at all.

U

Application management

Managing an application across its whole lifecycle, from deployment to retirement: versions, configuration, performance, access and support.

Upload

Transferring data from a device to a remote system.

Uptime

The proportion of time a system stays available. A percentage alone in a contract is not enough: ask what counts as up, and whether planned maintenance is excluded.

V

Data

Raw, unprocessed information. The moment it contains personal data, storing, processing and transferring it require a legal basis.

Data center

A facility housing server, storage and network hardware with power, cooling and physical security. Dense GPU racks need their power and cooling calculated separately.

Data privacy

Protecting personal and sensitive data against unauthorised access and disclosure. Under Turkish data protection law, transfer abroad requires its own legal basis.

Read further

Data storage

Holding data persistently. Tiering by access frequency is how performance and cost get managed together rather than traded off.

Data warehouse

A structured store where data from different sources is gathered for analysis. Unlike a data lake, the schema is applied on load.

Database

A system where data is stored in structured form and can be queried. Relational and non-relational families are designed for different problems; the choice follows the workload.

Read further

Database operations services

Database installation, monitoring, backup, tuning and version management taken as a service. The output to require in the contract is the restore test report.

Efficiency

Producing a given output with the least resource required.

VDS

Virtual dedicated server — running on shared physical hardware but with resources allocated to a single customer.

Virtual machine (VM)

A software-created equivalent of a physical computer, running its own operating system. Unlike a container, it carries a full OS.

Virtual private cloud (VPC)

A logically isolated network space inside a public cloud. Because the isolation is logical, it may not substitute for a private cloud under regulation.

Read furtherSee also: Private cloud

Virtualization

Abstracting physical resources and presenting them as logical ones.

VPN

Technology that establishes an encrypted private connection over a public network. Common for remote access — and who used it and when is a question that comes up in audits.

VPS

Virtual private server — a partition of a physical server with its own operating system and allocated resources.

W

Web

The interlinked documents and applications reached over the internet.

Web 3.0

A vision of the web built on decentralised protocols and user ownership.

Web application firewall (WAF)

A security layer protecting web applications against application-layer attacks. The critical step at deployment is running rules in monitoring mode first to clear false positives; the reverse order stops production.

X

XaaS (anything-as-a-service)

The umbrella term for delivering any IT capability as a subscription service.

XML

A tag-based format for structuring data in transport and storage, still widespread in enterprise integration.

Z

Zero trust

A security model in which being inside the network confers no trust. Every access request is re-verified against identity, device posture and context — "never trust, always verify".