Table of Contents
2. SCALITY Software and Services
3.1. What type of information we collect
3.2. What information we do not collect
Specific Provisions — United States
1.3. Purpose of use and legal basis
1.6. California residents' privacy rights
1.7. Change to the Privacy Policy (Specific Provisions)
Specific Provisions — United Kingdom
3.4. How is your Personal Data collected?
3.5. How we use your personal data
3.6. Purpose of use and legal basis
3.9. Withdrawal of your consent
Home/Glossary
Scality glossary
Clear definitions for enterprise storage, AI infrastructure and data protection.
Browse by topic
- All topics
- Storage fundamentals
- AI & data infrastructure
- Data protection & recovery
- Cloud & architecture
- Security & governance
Explore more
All terms
A – ZNo terms match your search. Try a different word or clear the filters.
A
- Active-active disaster recoveryTwo or more live sites share production, so losing one leaves the others serving users.
- Agentic RAGRAG in which an AI agent decides what to search, where and how often before it answers.
- AI data architectureThe overall design for where AI data lives, how it moves and which systems can reach it.
- AI data infrastructureThe storage, networks and data services that store, move and protect data for AI.
- AI data managementThe ongoing work of keeping AI data findable, accurate, governed and protected.
- AI data pipelineThe chain of steps that moves raw data through preparation to AI training and live answers.
- AI data platformA product or integrated set of products giving AI one shared, governed data layer.
- AI data readinessHow well an organization's data and storage can support a specific AI use case now.
- AI knowledge baseAn organization's documents and data, prepared and indexed so AI assistants can search and cite them.
- AI storageStorage that holds and serves AI data, from raw collections and training sets to live answers.
- All-flash arrayShared storage built entirely on flash; fast, but its costs and limits shift at petabyte scale.
B
- Backup compressionRewriting backup data in a more compact form so it uses less storage and network time.
- Block storageRaw volumes of numbered blocks that servers format and use like a local disk.
- Breach containmentCutting off an attacker's access and stopping further damage once a breach is found.
- Business continuityKeeping key products and services running through a disruption, and returning to normal afterwards.
- Business impact analysis (BIA)A review of how outages harm each business process, used to set recovery priorities.
C
- Cloud archive storageLow-cost storage for data kept for years but rarely read, and its retrieval trade-offs.
- Cloud automationSoftware that creates, changes and removes infrastructure, storage included, without manual steps.
- Cloud cost managementTracking, assigning and reducing cloud spending, with storage bills in focus.
- Cloud migrationMoving applications and data into, between or back out of cloud environments.
- Cloud scalabilityA system's ability to grow with data and workload while performance and cost hold.
- Cloud storageStorage delivered as an on-demand network service, from public, private or sovereign clouds.
- Cloud storage gatewayAn appliance that lets file, block or tape software use cloud object storage.
- Cloud storage replicationAutomatic copying of stored data to another site, region or cloud for availability.
- Cloud storage securityThe controls that protect data in cloud and object storage from exposure, theft and deletion.
- Cold siteAn empty standby facility with power and network ready, but no equipment or data.
- Cold storageLow-cost storage for data that is rarely read but has to be kept, often for years.
D
- Data durabilityThe likelihood that stored data stays intact and readable over time despite hardware failures.
- Data fabricAn architecture that connects, governs and serves data across many systems, clouds and sites.
- Data gravityThe pull large datasets exert on applications, making the data hard to move.
- Data lakeA central store that keeps data of any type in its original form until it is needed.
- Data localityKeeping data physically or logically close to the applications and compute that use it.
- Data mobilityThe ability to move or reach data across systems, sites and clouds intact.
- Data replicationKeeping copies of data on several systems and updating them as the original changes.
- Disaster recoveryHow organizations restore IT systems and data after losing a site or the data it holds.
- Disaster recovery as a service (DRaaS)A provider keeps copies of servers and data and runs them on its infrastructure in a disaster.
- Disaster recovery orchestrationSoftware that runs a disaster recovery plan as an automated, repeatable workflow.
- Disaster recovery planThe written procedure for restoring IT systems after a disruption, step by step.
- Disaster recovery testingRehearsing a disaster recovery plan to measure real recovery times and data loss.
- Distributed file systemFiles spread across many servers but presented as one ordinary file system.
- Distributed storageData spread across many independent servers that act as a single storage system.
- Document chunkingCutting documents into short passages so AI search can find, cite and pass them to a model.
E
- Egress feesCharges a cloud provider applies for moving data out of its network.
- EmbeddingsLists of numbers an AI model produces to capture meaning, so similar content can be found.
- Enterprise storageStorage built for an organization's shared, critical data: what it covers and how to compare platforms.
- Erasure codingSplitting data into pieces plus spare pieces so it can be rebuilt after drives or servers fail.
- ExabyteA unit of a million terabytes, and what keeping that much data in one place demands.
F
- FailbackThe planned return of workloads and data to the primary site after a failover.
- FailoverThe switch of a workload to a standby when a component or site fails.
- Flash cacheFlash holding copies of busy data in front of disk; helps only when data is read again.
- Flash storage enduranceHow much data a flash drive can absorb before wearing out, rated in DWPD and TBW.
- Flash storage vs HDDHow flash and hard disks compare on speed, cost, power and fit for data at large scale.
G
H
- Hot siteA fully running standby data center with current data, ready to take over within minutes.
- Hot storageThe fastest, costliest storage tier, used for data that applications read and write constantly.
- Hybrid cloudPrivate and public clouds run together, with data and workloads moving between them.
- Hybrid flash arrayStorage combining flash and hard disks; fast only while the busy data fits on flash.
- Hybrid searchSearch that runs keyword and meaning-based search together and merges the results.
- Hyper-converged infrastructure (HCI)Compute, storage and virtualization combined on one cluster of servers with pooled local drives.
I
M
- Multi-cloudUsing more than one public cloud provider, and what that means for data.
- Multi-protocol storageStorage exposing the same data through NFS, SMB and S3 without keeping separate copies.
- Multi-region storageStoring data across two or more distant regions so it survives the loss of one.
- Multi-tenant storageOne storage platform shared by many tenants whose data and access stay separate.
N
- NAND flashThe memory chip inside every SSD; holds data as charge, erases in blocks and wears with use.
- NAS consolidationMoving file shares from many separate filers and file servers onto fewer shared platforms.
- Network-attached storage (NAS)File storage shared over a network, and where filers strain at petabyte scale.
- NVMeThe standard way servers talk to flash over PCIe, with many parallel queues instead of one.
- NVMe over Fabrics (NVMe-oF)NVMe stretched across a network so servers can use remote flash almost as if it were local.
O
P
- Persistent memoryMemory used like RAM that keeps its contents without power, used to make small writes safe quickly.
- Petabyte storageStoring a thousand terabytes or more in one system, and what that scale demands.
- Private cloudCloud-style infrastructure dedicated to one organization, on hardware it controls.
R
- RAG hallucinationWhen a RAG system gives an answer its retrieved sources do not support.
- RAG storageThe storage behind a RAG system: source documents, extracted text, indexes and answer records.
- RAG vs fine-tuningTwo ways to adapt an AI model to company data: look documents up, or train the model further.
- Read/write performanceWhy storage reads and writes run at different speeds, and how the gap affects sizing.
- Recovery point objective (RPO)The most recent data, measured in time, a system can afford to lose in a recovery.
- Retrieval-augmented generation (RAG)A way for AI models to answer from an organization's own documents, found by search at question time.
S
- S3 compatible storageStorage that applications use through the Amazon S3 API, running outside Amazon's cloud.
- SAN consolidationMerging separate block storage arrays and their networks into fewer, larger shared systems.
- Scale-out storageStorage that grows by adding nodes, each bringing drives, compute and network to one system.
- Secondary storageStorage for copies of production data: backups, replicas, archives and test clones.
- Semantic searchSearch that matches on meaning, finding documents that share no words with the question.
- Sequential vs. random I/OReading data in order versus jumping between scattered spots, and why the pattern shapes storage choices.
- SOC reportAn independent auditor's report on the controls a service provider runs.
- Software-defined storageStorage features run as software on standard servers, independent of any one hardware vendor.
- Solid-state drive (SSD)Flash-based drive with no moving parts; data center grades add endurance, power protection and steady speed.
- Storage area network (SAN)A dedicated network giving servers disk-level access to shared storage arrays.
- Storage automationSoftware, APIs and policies that provision, protect and repair storage without manual steps.
- Storage benchmarkingRunning fixed, repeatable workloads to compare storage systems, plus in-house tests on one platform.
- Storage compressionLossless encoding that stores data in fewer bytes and rebuilds the exact original on read.
- Storage consolidationMoving data from many separate storage systems onto fewer shared platforms or a single pool.
- Storage high availabilityKeeping stored data readable and writable through failures, maintenance and upgrades.
- Storage latencyThe wait between asking storage for data and getting it back, and where that time goes.
- Storage lifecycle managementPolicy-driven placement, protection, retention and deletion of data over its lifecycle.
- Storage managementThe ongoing work of provisioning, protecting, monitoring, moving and securing data on storage systems.
- Storage performanceHow fast storage completes work, measured as latency, IOPS and throughput, and why it degrades.
- Storage performance tuningFinding what limits storage speed and changing settings, layout or applications to remove it.
- Storage poolingCombining capacity from many drives or servers into one shared pool that applications draw from.
- Storage quality of service (QoS)Controls that cap, reserve or weight the performance each workload gets on shared storage.
- Storage sprawlUnplanned growth in storage systems, data copies and cloud resources across an organization.
- Storage throughputHow much data storage moves each second, and how that sets restore, rebuild and AI times.
- Storage tieringPlacing data on faster or cheaper storage as its use changes, from flash to archive.
- Storage virtualizationA software layer that presents physical storage as logical volumes, files or objects.