Glossary
This glossary defines common terms used in VeloDB Cloud documentation. Terms are grouped alphabetically. For implementation details, follow the links to the related documentation.
A
Agent
An Agent is a VeloDB Cloud component deployed in your cloud account for a BYOC warehouse. It receives control instructions over a private connection and performs cluster creation, scaling, upgrades, and warehouse initialization. See Deployment Models.
Amazon Resource Name (ARN)
An Amazon Resource Name (ARN) is the standardized identifier AWS uses to specify a resource, such as a KMS encryption key. See Warehouse Encryption.
Apache Kafka
Apache Kafka is a distributed event-streaming platform for publishing, storing, and consuming streams of records for data integration and real-time processing. VeloDB Cloud can continuously ingest records from Kafka-compatible topics through an import job. See Import from Kafka.
API Key
An API key is a credential for authenticating requests to the VeloDB Cloud management API. Its permissions follow the organization role assigned when it is created, and the key is shown only once. See API Keys.
Aggregate Key Model
The Aggregate Key Model pre-aggregates rows that have the same key columns. It is designed for summary and reporting workloads where queries do not require the original detail rows. See Aggregate Key Model.
Amazon Resource Name (ARN)
An Amazon Resource Name (ARN) is the standardized identifier that AWS uses to uniquely identify a resource, such as an IAM role, instance profile, S3 bucket, or KMS key. When you register BYOC cloud resources, you paste several ARNs into the VeloDB Cloud console. See Credential Configuration.
Amazon EventBridge
Amazon EventBridge is an AWS event service that routes events to targets such as Lambda functions and can invoke them on a recurring schedule. See Automatically Sync IP Targets for Amazon RDS and Aurora Endpoints.
ARM
ARM is a CPU architecture available for AWS-based VeloDB Cloud clusters. ARM clusters require core version 4.0.4 or later. See View Cluster Details.
Arrow Flight SQL
Arrow Flight SQL is a protocol for executing SQL queries and transferring results in Apache Arrow format. For VeloDB Cloud availability and connection details, see Arrow Flight SQL Connection.
ASOF Join
An ASOF join matches each row from one table to the closest earlier or later row in another table based on an ordering column, instead of requiring an exact match. It is useful for time-series data, such as aligning each event with the most recent price. See ASOF Join.
Auto Scaling
Auto Scaling is a scaling strategy that automatically adjusts a cluster's compute capacity within configured minimum and maximum CPU limits as workload utilization changes. Cache capacity changes proportionally with the compute tier. See Auto Scaling.
Auto-suspend
Auto-suspend is a cluster setting that automatically suspends a cluster after a period of inactivity to reduce compute costs. The cluster resumes when it next receives a request. See Create a SaaS Warehouse.
Availability Zone
An availability zone is an isolated location within a cloud region where a warehouse or cluster can run. Choosing a zone close to your application can reduce network latency. See Create a SaaS Warehouse.
AWS CloudFormation
AWS CloudFormation is an AWS service that provisions cloud resources from declarative templates. VeloDB Cloud uses a CloudFormation template to create the resources for a BYOC warehouse in the New VPC and Existing VPC deployment options. See Deploy in a New VPC.
AWS Key Management Service (AWS KMS)
AWS Key Management Service, abbreviated as AWS KMS, is an AWS service for creating and managing the encryption keys that protect data and control cryptographic operations. For an AWS BYOC warehouse created with the Custom Infrastructure option, you register a customer-managed KMS key so the warehouse can use it for Transparent Data Encryption (TDE), EBS encryption, or both. See Encryption Keys Configuration.
AWS Lambda
AWS Lambda is a serverless compute service that runs code in response to events or scheduled invocations without requiring you to manage servers. In this workflow, it resolves a database endpoint and synchronizes the resulting IP addresses with a load balancer target group. See Automatically Sync IP Targets for Amazon RDS and Aurora Endpoints.
Azure Deployment Stacks
Azure Deployment Stacks is an Azure service that deploys and manages a group of related resources from a template. VeloDB Cloud uses it to provision the resources for a BYOC warehouse on Azure. See How to create BYOC Warehouse on Azure.
Azure Resource Manager
Azure Resource Manager is the Azure deployment and management service that provisions Azure resources from templates and deployment parameters. When you create a BYOC warehouse on Azure, you run a VeloDB-provided template through Azure Resource Manager to create the resources the warehouse needs. See How to create BYOC Warehouse on Azure.
B
Backend (BE)
The backend, abbreviated as BE, is an Apache Doris service that stores data and executes query plan fragments. In a distributed query, the frontend sends work to one or more BEs. See Query Profile Analysis.
Backup Plan
A backup plan is a warehouse configuration that defines a periodic or one-time backup operation. Creating or updating a plan replaces the warehouse's previous Import/Export backup plan. See Backups.
Broker Load
Broker Load is an asynchronous bulk-ingestion method that loads data into a VeloDB Cloud table from files in external storage. See Broker Load.
Bucket
A bucket distributes the data in a partition into multiple tablets. The bucketing process helps spread data across nodes and enables parallel reads and writes. See Partitioning and Bucketing.
BYOC
BYOC means bring your own cloud. In a BYOC warehouse, VeloDB Cloud installs and operates the warehouse in cloud resources in your own cloud account. See Create a BYOC Warehouse.
C
Cache Eviction
Cache eviction is the removal of cached data when available cache capacity decreases or space is required. Evicted data must be read from remote storage again before it can be cached. See Scaling Overview.
Cache Warmup
Cache warmup is a process that proactively loads table data into a compute group's cache before queries access it. See Cache Warmup.
CIDR Block
A CIDR block is a notation that represents a range of IP addresses with an address and prefix length, such as 192.0.2.0/24. See Public IP allowlist.
Catalog
A catalog is a metadata and connection definition that lets VeloDB Cloud access tables in an external data source. These catalogs are commonly used to query lakehouse systems and object storage without first copying the data into internal tables. See Catalogs and Lakehouse Catalog Overview.
CDC
CDC means change data capture. It records inserts, updates, and deletes in a source system so that changes can be synchronized incrementally to VeloDB Cloud. See PostgreSQL CDC.
Cold Cache
A cold cache is a cache state in which the required data has not yet been loaded into local cache, so a query reads it from the remote source. See Before You Start the POC.
Cloud Shell
Cloud Shell is a browser-based command-line environment provided by Google Cloud for running commands against your Google Cloud resources. When you create a BYOC warehouse on Google Cloud, you run a VeloDB-provided command in Cloud Shell to create the resources the warehouse needs. See How to create BYOC Warehouse on GCP.
Cloud Marketplace Deduction Channel
A cloud marketplace deduction channel is an active cloud marketplace payment arrangement used to deduct charges for eligible VeloDB Cloud resource usage. See Create a Cluster.
CloudWatch Logs
CloudWatch Logs is an AWS service for collecting, retaining, and inspecting log records produced by services and applications such as Lambda functions. See Automatically Sync IP Targets for Amazon RDS and Aurora Endpoints.
Cluster
A cluster is a group of compute resources inside a warehouse. A warehouse can contain multiple clusters, and the clusters share the warehouse's stored data. Cluster and Compute Group refer to the same concept in VeloDB. See Create a Cluster.
Compaction
In Apache Doris, compaction is a background process that merges data versions or files to reduce redundancy and improve storage and query efficiency. See Aggregate Key Model.
Data Cache
Data Cache stores recently accessed remote lakehouse data on local cluster storage to reduce latency for subsequent queries. It applies to Hive, Iceberg, Hudi, and Paimon tables. See Data Cache.
Data Skew
Data skew is an uneven distribution of data or tablet sizes across buckets or backend nodes that can reduce parallel query performance. See Before You Start the POC.
Compute Group
A Compute Group is a group of one or more backend (BE) nodes that provides physical workload isolation in a storage-compute separation architecture. Compute Group and Cluster refer to the same concept in VeloDB. Compute Groups share warehouse data in shared storage while using separate compute resources and caches. You can grant users access to a Compute Group, set a default group, and route workloads to a specific group. See Compute Group.
Compute-Storage Separation
With compute-storage separation, compute resources and storage resources can scale and be managed independently. In VeloDB Cloud, stored data is shared by the clusters in a warehouse, while cluster compute and cache can be adjusted for workload needs. See Create a Cluster.
Configuration Drift
Configuration drift is a difference between resources recorded in Terraform state and their current configuration because a change was made outside Terraform. Run terraform plan before each change to identify drift and proposed updates. See Terraform Provider.
Condition Cache
The Condition Cache stores the result of a filter condition over a segment of data and reuses it when the same filter runs again over unchanged data, so repeated queries that use that filter run faster. See Condition Cache.
Control Plane
The control plane is the shared VeloDB Cloud service layer for managing warehouses and organizations. It handles operations such as warehouse creation, deletion, scaling, settings, members, billing, and sign-in. See Console Data Plane and Control Plane.
Cooldown Period
A cooldown period is a 15-minute waiting interval after a successful scaling event during which Auto Scaling does not make another scaling change. See Auto Scaling.
Core Version
A core version is the version of VeloDB Core that runs in a warehouse. You can upgrade to a selected version manually and configure automatic upgrades for patch versions. See Upgrade Core Version.
CPU Architecture
CPU architecture is the processor instruction-set architecture used by a cluster's compute nodes, such as x86 or ARM. The architecture is selected when the cluster is created and cannot be changed later. See View Cluster Details.
Credential Configuration
A credential configuration is a reusable set of IAM roles that VeloDB Cloud uses to access a BYOC warehouse's S3 bucket and to deploy compute resources in your AWS account. You select a credential when creating a BYOC warehouse, or create one in advance from the Cloud Resources page in the console. See Credential Configuration.
Custom Infrastructure
Custom Infrastructure is an AWS-only BYOC deployment option in which you provide and manage the warehouse infrastructure, including networking, IAM roles, storage, and optional encryption keys. It is the only deployment option that supports customer-managed encryption keys. See Custom Infrastructure.
Custom Tags
Custom tags are user-defined key-value labels attached to the cloud resources VeloDB Cloud creates for a BYOC warehouse, so you can organize, identify, and manage those resources in your own cloud account. See Custom Tags.
Customer-Managed Key (CMK)
A customer-managed key, abbreviated as CMK, is an encryption key that you own and manage in AWS KMS and provide to VeloDB Cloud to encrypt warehouse data or EBS volumes. VeloDB Cloud uses the key but never stores or exports it, so you keep control of its lifecycle. See Warehouse Encryption.
D
Data Plane
The data plane carries requests that communicate with a warehouse to query or manage data. It is separate from the control plane, which manages organizations and warehouse resources. See Console data plane and control plane.
DNS Name
A DNS name is a human-readable network name that resolves to the address of a service or endpoint, allowing clients to connect without using a fixed IP address. See Access VeloDB from Your VPC on AWS.
Duplicate Key Model
The Duplicate Key Model retains every inserted row without deduplication or pre-aggregation. It is the default table model and is suitable for append-only data such as logs, user behavior, and transaction details. See Duplicate Key Model.
E
Endpoint Service
An endpoint service is a cloud-provider resource that exposes a service for private connectivity. In AWS, it lets a consumer VPC connect to a service provider through AWS PrivateLink. See Access VeloDB from Your VPC on AWS.
EBS Encryption
EBS encryption protects the EBS volumes used by warehouse nodes, including cache-related block storage. It is enabled by default using a cloud-provider-managed key. For AWS BYOC warehouses created with the Custom Infrastructure option, you can supply your own KMS key at warehouse creation, and this key cannot be rotated afterward. See EBS encryption.
Encryption at Rest
Encryption at rest protects data stored in persistent storage, such as cloud object storage and disks, so it cannot be read without the required encryption key. VeloDB Cloud provides storage-layer encryption by default and optional warehouse-layer encryption through Transparent Data Encryption (TDE). See Encryption at Rest.
Encryption in Transit
Encryption in transit protects data as it moves between clients, services, and network endpoints. In VeloDB Cloud, supported connection methods use protocol-level encryption such as TLS when enabled. See Encryption in Transit.
Encryption Key Configuration
An encryption key configuration is a reusable Cloud Resources configuration that registers one of your AWS KMS keys so a BYOC warehouse can use it for Transparent Data Encryption (TDE), EBS encryption, or both. See Encryption Keys Configuration.
External ID
An external ID is a value included in an IAM role's trust policy that restricts cross-account role assumption to the intended party. VeloDB Cloud provides the external ID for the cross-account role used in BYOC deployment, and you add it to the role's trust policy. See Credential Configuration.
External Table
An external table is a table definition that references data outside the warehouse, such as data in an object store or a lakehouse table format. External tables are accessed through a catalog and do not require the data to be loaded into internal tables first. See Lakehouse Overview.
F
Fe Compute
Fe Compute is the billable compute capacity assigned to a Frontend Group when its specification is scaled above the free Small baseline. See Scale Frontend Groups.
Frontend (FE)
The frontend, abbreviated as FE, is an Apache Doris service that receives SQL requests, maintains metadata, creates execution plans, and coordinates query execution. The FE sends execution fragments to backend services. See Query Profile Analysis.
Frontend Group
A Frontend Group is a group of Frontend (FE) nodes that provides frontend capacity for a warehouse. Each warehouse has a default FE group named meta. This feature is in private preview and must be enabled by VeloDB support. See Scale Frontend Groups.
Full-Text Search
Full-text search is a text-search capability that uses inverted indexes and the SEARCH() function to find matching text. See Full-Text Search.
G
Generated Column
A generated column is a table column whose value is computed automatically from an expression over other columns instead of being supplied by an insert or load. See Add a Generated Column.
Google Cloud project
A Google Cloud project is the administrative container that groups resources, configuration, permissions, and billing on Google Cloud. When you create a BYOC warehouse on Google Cloud, you run the deployment command in the project where the warehouse resources are created, which for an existing network is the project that contains your VPC and subnet. See How to create BYOC Warehouse on GCP.
Group Commit
Group Commit is a load-commit mechanism that batches compatible concurrent load requests into shared commits to improve ingestion efficiency. See Group Commit.
H
HNSW
HNSW means hierarchical navigable small world. It is a graph-based approximate nearest neighbor index used for high-performance vector retrieval. See HNSW.
I
IAM Instance Profile
An IAM instance profile is an AWS container that associates an IAM role with an EC2 instance so the instance can use the role's permissions. The EC2 instances that run a BYOC warehouse read and write your S3 bucket through the instance profile of the data-access role. See Credential Configuration.
IAM Policy
An IAM policy is a document that defines the permissions granted to an AWS identity, such as an IAM role, or to a resource. BYOC deployment uses a data-access policy for the warehouse's S3 access and a deployment policy for the compute VeloDB Cloud manages. See Credential Configuration.
IAM Role
An IAM role is an AWS identity with a set of permissions that a service or user can assume to access AWS resources. A BYOC warehouse uses IAM roles so that VeloDB Cloud can access your S3 bucket and deploy compute in your account. See Credential Configuration.
Iceberg Manifest
An Iceberg Manifest is an Apache Iceberg metadata file that records the data files associated with a table snapshot. See Iceberg Catalog.
Idempotent
Idempotent describes an operation that can be run multiple times with the same result as running it once. Repeating it after the desired state is reached makes no further changes. See Automatically Sync IP Targets for Amazon RDS and Aurora Endpoints.
Identity Provider (IdP)
An identity provider, abbreviated as IdP, authenticates users and sends identity assertions to an application. In VeloDB Cloud SAML single sign-on, your identity provider authenticates organization members, while VeloDB Cloud is the service provider. See SAML Single Sign-On.
Internet Gateway
An internet gateway connects a VPC to the public internet so that resources with the appropriate routing can send and receive internet traffic. For a BYOC warehouse in an existing VPC, the subnets need internet access through an internet gateway or a NAT gateway to reach AWS services. See Deploy in an Existing VPC.
Inverted Index
An inverted index maps terms or values to the rows that contain them. It accelerates text search and filtering without scanning every row. See Index Overview.
IP Allowlist
An IP allowlist is a network access policy that permits connections only from specified source IP addresses or CIDR blocks. See Public IP allowlist.
IVF
IVF means inverted file. It is an approximate nearest neighbor vector index that groups vectors into clusters so that a search scans only the most relevant clusters, using less memory than HNSW for large-scale vector search. See IVF.
J
JDBC
JDBC, or Java Database Connectivity, is a standard Java API that applications and tools use to connect to databases, submit SQL statements, and retrieve results. VeloDB Cloud supports MySQL-compatible JDBC connections to warehouses. See Connection Methods.
JSONB
JSONB is the binary storage format that VeloDB Cloud uses for the JSON data type. It keeps JSON objects and values in a parsed form so that path access and extraction are faster than parsing raw text. See JSON.
Just-in-Time (JIT) Provisioning
Just-in-time, abbreviated as JIT, provisioning automatically creates a VeloDB Cloud user and organization membership after a successful SAML sign-in, subject to the configured provisioning policy. See SAML Single Sign-On.
K
KMS Key Policy
A KMS key policy is a resource-based AWS policy attached to a KMS key that specifies which principals can use the key and which cryptographic actions they can perform. Before you register an encryption key configuration, the key's policy must include a statement that allows VeloDB Cloud to encrypt and decrypt with the key. See Encryption Keys Configuration.
L
Lakehouse
A lakehouse combines the open storage and table formats of a data lake with the query and analysis capabilities of a data warehouse. VeloDB Cloud can use catalogs to query data in systems such as Hive, Iceberg, Hudi, and Paimon. See Lakehouse Overview.
Log Explorer
Log Explorer is a VeloDB Cloud console tool for searching and inspecting warehouse logs. It supports keyword, phrase, and SQL searches. See Log Explorer.
LZ4
LZ4 is a lossless compression algorithm optimized for very fast compression and decompression at a moderate compression ratio. VeloDB Cloud can use it for column storage and for Parquet and ORC files written through lakehouse catalogs. See Column Compression.
M
Manual Scaling
Manual Scaling is a scaling strategy that lets an administrator make a one-time change to a cluster's compute or cache capacity immediately. See Manual Scaling.
MATCH_ANY
MATCH_ANY is a search operator that matches records containing one or more supplied keywords across the searched fields. See Log Explorer.
MATCH_PHRASE
MATCH_PHRASE is a search operator that matches a sequence of words as a phrase, typically entered within double quotation marks. See Log Explorer.
Maintenance Window
A maintenance window is the preferred recurring time period for scheduled maintenance and automatic software updates on a warehouse. See Warehouse Settings.
Management API
The Management API is the VeloDB Cloud API used to provision and manage warehouses, clusters, networking, and connection metadata programmatically. See Management API.
Materialized View
A materialized view stores the result of a query so that repeated queries can read precomputed data. VeloDB Cloud can use materialized views to accelerate queries and reduce repeated computation. See Materialized Views.
MCP Server
An MCP Server exposes VeloDB Cloud data and query capabilities to AI agents through the Model Context Protocol. The VeloDB Cloud MCP Server supports metadata discovery, SQL queries, and semantic-layer queries. See Get Started with the VeloDB MCP Service.
MCP Server URL
The MCP Server URL is the warehouse-specific URL that an AI client or agent uses to connect to the VeloDB MCP Service. Copy it from the MCP connection method for your warehouse. See MCP connection method.
MERGE INTO
MERGE INTO is a SQL statement that inserts, updates, and deletes rows in a target table in a single operation, based on whether each row matches a source. See MERGE INTO.
Meta Service
A Meta Service holds catalog and transaction metadata for a SaaS warehouse. Its isolation level can be shared across organizations, reserved for one organization, or dedicated to one warehouse. See Single-Tenant Isolation on Premium.
MPP
MPP means massively parallel processing. In an MPP architecture, a query is divided into tasks that run in parallel across multiple compute nodes. See Pipeline Execution Engine.
Metrics API
The Metrics API is an API for scraping warehouse and cluster metrics into monitoring systems such as Prometheus or Grafana. See Metrics API.
Model Context Protocol (MCP)
Model Context Protocol (MCP) is a protocol that lets AI applications and agents connect to servers that provide tools, data, and other capabilities. See VeloDB MCP Service Overview.
Multi-AZ
A Multi-AZ deployment places a warehouse and its clusters across multiple availability zones so the service keeps running through the loss of a single availability zone within a region. It provides high availability and is distinct from cross-region disaster recovery. See High Availability.
Multi-Factor Authentication (MFA)
Multi-factor authentication (MFA) is an account-security method that requires an organization member to provide a second verification factor when signing in to the VeloDB Cloud console. VeloDB Cloud supports an authenticator-app code or SMS code. See Multi-Factor Authentication.
MySQL-Compatible Connectivity
MySQL-compatible connectivity is an interface that lets MySQL-compatible clients and tools connect to a warehouse using supported MySQL-compatible protocols or drivers. See Database Connect.
N
NAT Gateway
A NAT gateway lets resources in a private subnet make outbound connections to external services while blocking unsolicited inbound connections. The New VPC deployment option creates a NAT gateway so warehouse resources can reach the AWS services they need. See Deploy in a New VPC.
Network Configuration
A network configuration is a reusable definition of the subnets, security group, and optional VPC endpoint where a BYOC warehouse runs. A network configuration is either Single-AZ or Multi-AZ, and you select one when creating a BYOC warehouse. See Network Configuration.
Network Load Balancer (NLB)
A Network Load Balancer is an AWS load-balancing service that forwards TCP and other network traffic to registered targets, including IP addresses, while providing health checks. See Automatically Sync IP Targets for Amazon RDS and Aurora Endpoints.
O
Object Storage
Object storage stores data as objects in containers, such as buckets, and is commonly used for durable, large-scale data storage. VeloDB Cloud uses cloud object storage as the shared storage layer for warehouse data. See Encryption at Rest.
ODBC
ODBC, or Open Database Connectivity, is a standard interface that applications and business intelligence tools use to connect to databases through database-specific drivers. VeloDB Cloud supports MySQL-compatible ODBC connections to warehouses. See Connection Methods.
Observability
Observability is the practice of monitoring a system through metrics, alerts, audit records, usage information, and other operational data to understand its health and behavior. See Monitoring.
Organization
An organization is the VeloDB Cloud administrative scope that contains members, organization roles, billing settings, warehouses, and clusters. These organization roles control console access and are separate from warehouse SQL roles. See Members.
Organization Role
An organization role controls what a member can see and manage in the VeloDB Cloud console, including organization and warehouse-management access. It does not grant SQL permissions in a warehouse. See Members.
P
Partition
A partition divides a table into smaller, mutually exclusive subsets based on column values such as time or region. Partition pruning can reduce the data scanned by a query and simplify data lifecycle management. See Partitioning and Bucketing.
Plan
A plan is an organization-level tier that sets the pricing and the set of capabilities available to every warehouse in the organization. VeloDB Cloud offers the Standard and Premium plans, and switching plans applies to the whole organization. See Choose Plans.
Private DNS
Private DNS resolves a service domain to a private endpoint address within a VPC or VNet. Use it when a private connection requires the VeloDB Cloud domain for TLS hostname verification. See Access VeloDB from Your VPC.
Private Preview
Private Preview is a feature availability status for capabilities that require contacting VeloDB Cloud support to use. Availability can depend on the deployment model and region. See Release Stages.
PrivateLink
PrivateLink is a private network path for connecting a client or application in your cloud environment to a SaaS warehouse without exposing the connection through the public internet. See Choose a Connection Path.
Private Service Connect
Private Service Connect is a Google Cloud networking capability that provides private connectivity from a VPC to a supported service without using the public internet. A Google Cloud BYOC warehouse uses it to connect its Agent to VeloDB Cloud. See Google Cloud Preparation.
Prefix Index
A Prefix Index is a sparse index built from the leading bytes of a table's sort key. It helps locate matching data blocks without scanning all rows and is maintained automatically. See Prefix Index and Sort Key.
Proof of Concept (POC)
A proof of concept (POC) is a limited, representative evaluation used to determine whether VeloDB Cloud meets the technical and operational requirements of a planned workload. See Before You Start the POC.
Public Preview
Public Preview is a feature availability status for capabilities available for evaluation. See Release Stages.
Public endpoint
A public endpoint lets authorized clients connect to a SaaS warehouse over the internet using its connection information. See SaaS Public Endpoint.
Q
Query Audit
Query Audit is a VeloDB Cloud tool for reviewing historical queries, including their details and available filtering dimensions. See Query Audit.
Query Cache
The Query Cache stores the intermediate aggregation result of a query and reuses it when the same aggregation runs again over unchanged data, so repeated aggregate queries run faster. See Query Cache.
Query ID
A Query ID identifies a query so that you can locate and inspect its execution and audit details. Failed and non-query statements do not have a Query ID. See Query Audit.
Query Profile
A query profile records query execution details, including operator metrics, elapsed time, row counts, and memory usage. It helps identify slow operators and other query performance bottlenecks. See Query Profile Analysis.
Quantization
Quantization is a technique that uses more compact vector representations to reduce memory requirements for large-scale vector search. See IVF vector search.
R
RAG
RAG means retrieval-augmented generation. A RAG application retrieves relevant information from a knowledge base and provides it as context to a generative AI model. VeloDB Cloud supports RAG workflows with vector search, text search, and SQL analytics. See RAG.
Recursive Common Table Expression (Recursive CTE)
A recursive common table expression is a common table expression that refers to its own result, which lets a query process hierarchical or graph-shaped data through the WITH RECURSIVE syntax. See Common Table Expression.
Region
A region is a cloud-provider geographic location where resources such as networks, subnets, and warehouses are deployed. For a BYOC warehouse, the region you select in VeloDB Cloud must match the region of the cloud resources it uses, especially the network, or the deployment cannot complete. See Create a BYOC Warehouse.
Resource Group
A resource group is an Azure container that organizes related cloud resources that share a lifecycle and access control. When you create a BYOC warehouse on Azure, you deploy the resource stack into a resource group whose region must match the warehouse region. See How to create BYOC Warehouse on Azure.
Reverse Private Endpoint
A Reverse Private Endpoint is a private connectivity option that establishes a multi-VPC private endpoint from VeloDB Cloud to a source service in another VPC. See Amazon MSK.
Routine Load
With routine load, a continuously running import job consumes streaming data, such as records from Kafka, and loads it into a VeloDB Cloud table. See Routine Load.
S
S3 Bucket
An S3 bucket is an Amazon S3 storage container. A BYOC warehouse stores its data in an S3 bucket in your AWS account, and we recommend a dedicated bucket for each warehouse. See Credential Configuration.
Snapshot Backup
A snapshot backup is a warehouse-level snapshot used as a recovery point and for disaster-recovery preparation. See Backups.
S3 Gateway Endpoint
An S3 gateway endpoint is an AWS VPC gateway endpoint that gives instances in a VPC private connectivity to Amazon S3, avoiding the public internet and reducing NAT-gateway charges when S3 traffic would otherwise use a NAT gateway. See Network Configuration.
SaaS
SaaS means software as a service. In a SaaS warehouse, VeloDB Cloud provides and manages the underlying cloud resources and services, so you can use the warehouse without managing its infrastructure. See Create a SaaS Warehouse.
SAML Single Sign-On (SSO)
SAML single sign-on lets organization members sign in to the VeloDB Cloud console through your identity provider using the SAML standard. It is a Premium capability and can be disabled at any time. See SAML Single Sign-On.
SASL/SCRAM
SASL/SCRAM is a password-based authentication mechanism that uses SASL with the Salted Challenge Response Authentication Mechanism (SCRAM). See Amazon MSK.
Security Group
A security group is an AWS virtual firewall that controls inbound and outbound traffic for the resources it is attached to, using a set of traffic rules. A BYOC network configuration includes an EC2 security group that VeloDB Cloud attaches to the warehouse nodes it creates. See Network Configuration. A security group is a virtual firewall that controls inbound and outbound network traffic for resources in a cloud network. A BYOC warehouse uses security groups to control traffic to warehouse resources and VPC interface endpoints. See Network Configuration.
Service Provider (SP)
A service provider, abbreviated as SP, is the application that relies on an identity provider to authenticate users in a SAML single sign-on flow. VeloDB Cloud is the service provider for its SAML SSO configuration. See SAML Single Sign-On.
Semantic Model
A semantic model describes the business meaning of tables, entities, dimensions, and measures in a structured format. The VeloDB Cloud MCP Server uses a committed semantic model to answer metric queries with consistent business definitions. See Semantic Model Concepts.
Session Management
Session Management is a VeloDB Cloud facility for viewing and managing active warehouse sessions, including their queries, execution time, users, and resource usage. See Session Management.
Sort Key
A sort key is the ordered set of columns used to organize table data and support prefix-index filtering and efficient query execution. See Prefix Index and Sort Key.
SQL Editor
SQL Editor is a VeloDB Cloud console tool for writing and executing SQL statements against a selected warehouse. See SQL Editor.
Stream Load
The stream load method is an HTTP-based data import method for loading local files or request bodies directly into a VeloDB Cloud table. It is suitable for small batches and real-time or near-real-time ingestion. See Stream Load.
Streaming Job
A Streaming Job is a continuously running job that synchronizes data from a source system into VeloDB Cloud, including snapshot and incremental CDC processing. See Continuous Load Overview.
Subnet
A subnet is a subdivision of a VPC or VNet that provides a specific network range for deploying cloud resources. For a BYOC warehouse deployed in an existing network, the subnet must be in the same region as the warehouse. See Network Configuration.
Switchover
A switchover is a planned high-availability operation that promotes a virtual cluster's standby cluster to become the active cluster. See Create and Manage a Virtual Cluster.
Symmetric Encryption Key
A symmetric encryption key is a cryptographic key used for both encrypting and decrypting data. When configuring a customer-managed key, create a symmetric key in AWS KMS so VeloDB Cloud can encrypt and decrypt warehouse data. See Warehouse Encryption.
Symmetric Key
A symmetric key is an encryption key that uses the same key material to encrypt and decrypt data. A warehouse created with the Custom Infrastructure option requires a customer-managed symmetric AWS KMS key for its encryption key configuration. See Encryption Keys Configuration.
T
Table Name Case Sensitivity
Table Name Case Sensitivity is a warehouse creation setting that determines whether table names are treated as case-sensitive. It cannot be changed after the warehouse is created. See Create a SaaS Warehouse.
Tablet
A tablet is a physical data shard created by bucketing a table partition. These tablets are distributed across backend nodes so that data can be stored and processed in parallel. See Partitioning and Bucketing.
Target Group
A target group is an AWS load-balancing resource that contains the backend targets and port to which a load balancer routes traffic. Targets can include IP addresses, instances, or other supported target types. See Automatically Sync IP Targets for Amazon RDS and Aurora Endpoints.
Terraform
Terraform is an infrastructure-as-code tool that uses declarative configuration files to provision and manage cloud resources and services. See Terraform Provider.
Terraform Provider
A Terraform provider is a plugin that lets Terraform create and manage resources in an external service. The VeloDB Terraform provider manages VeloDB Cloud control-plane resources through the Management API. See Terraform Provider.
Terraform State
Terraform state is the recorded representation of resources managed by a Terraform configuration. Terraform uses state to track deployed resources, detect changes, and determine which updates to apply. State can contain sensitive values, so protect it with appropriate access controls and encryption. See Terraform Provider.
Time-Based Scaling
Time-Based Scaling is a scaling strategy that changes a cluster's compute capacity at fixed daily times according to configured rules for predictable workload peaks and low periods. See Time-Based Scaling.
TLS
TLS, or Transport Layer Security, is a cryptographic protocol that encrypts network connections and authenticates the connected service. See Access VeloDB from Your VPC on AWS.
Transparent Data Encryption (TDE)
Transparent Data Encryption, abbreviated as TDE, is an optional warehouse-layer encryption that encrypts warehouse service data before it is written to storage and decrypts it when the data is read. It is available for AWS warehouses and can use either a VeloDB-managed key or a customer-managed key. See Transparent Data Encryption.
U
UDAF
UDAF means user-defined aggregate function. It is a user-defined function that combines multiple input rows into a single aggregate result, such as a custom statistical calculation. See Python User-Defined Function.
UDF
UDF means user-defined function. It is a function that you implement to add custom processing logic that can be called from SQL, written in a language such as Python or Java. See Python User-Defined Function.
UDTF
UDTF means user-defined table function. It is a user-defined function that returns a set of rows and columns that a query can read like a table. See Python User-Defined Function.
Unique Key Model
The Unique Key Model keeps one latest row for each set of key columns. Inserting a row with an existing key updates that row, making this model suitable for upserts and frequently changing dimension data. See Unique Key Model.
USAGE Privilege
The USAGE_PRIV privilege allows a warehouse user or role to use a Workload Group for workload execution. See Workload Groups.
Utilization Watermark
A utilization watermark is a threshold that Auto Scaling uses to determine whether resource usage should trigger scale-out, scale-in, or no change. See Auto Scaling.
V
vCPU-hour
A vCPU-hour is a unit of compute usage equal to one virtual CPU running for one hour. VeloDB Cloud meters BYOC compute service fees by the running compute cluster's vCPU-hours. See Create a BYOC Warehouse.
Variant
The Variant type is a semi-structured column type that stores values with varying or nested structure, so objects and arrays can be queried without defining a fixed schema for every field. See VARIANT.
Vector Index
A vector index accelerates similarity search over vector data. VeloDB Cloud supports approximate nearest neighbor indexes such as HNSW and IVF for vector retrieval. See Vector Index Overview.
Virtual Cluster
A virtual cluster pairs two physical clusters as active and standby across availability zones. VeloDB Cloud can automatically fail over to the standby cluster when the active cluster's availability zone fails. See Virtual Cluster.
Vended Credentials
Vended credentials are temporary storage-access credentials issued by a REST Catalog for accessing the data files of a table. See Iceberg REST Catalog.
VNet
A Virtual Network (VNet) is Azure's isolated virtual network. It is the Azure equivalent of a VPC, and the VeloDB Cloud console labels the network option VPC on every platform. See How to create BYOC Warehouse on Azure.
VPC
A virtual private cloud (VPC) is an isolated virtual network in a cloud provider account where a BYOC warehouse and its supporting resources run. On Azure the equivalent resource is a Virtual Network (VNet). See Network Configuration.
VPC Endpoint
A VPC endpoint provides private connectivity from resources in a VPC to a supported service without sending traffic over the public internet. A BYOC warehouse can use a VPC endpoint to reach the VeloDB Cloud service over PrivateLink. See Network Configuration.
Version Pile-Up
Version pile-up is the accumulation of many unmerged data versions caused by frequent small-batch ingestion, which can increase storage and query overhead. See Before You Start the POC.
W
Warehouse
A warehouse is the primary resource and SQL scope in VeloDB Cloud. It contains clusters, stored data, connection settings, and warehouse-level users and roles. See Warehouse Settings.
Warehouse Users and Roles
Warehouse Users and Roles are the warehouse-level accounts and SQL permissions used by applications, BI tools, and SQL clients. They are separate from organization members and organization roles used to access the VeloDB Cloud console. See Warehouse Users and Roles.
Warehouse-Specific Console Endpoint
A warehouse-specific console endpoint routes console data-plane requests for one warehouse through a private network path. It does not carry shared control-plane operations such as sign-in or billing. See Private connectivity to the warehouse-specific console endpoint.
Warmup Job
A warmup job preloads data into a cluster's local cache, optionally synchronizing cached data from another cluster, so workloads can avoid initially reading that data from remote storage. See Manage Warmup Jobs.
Window Function
A window function calculates a value across a set of related rows while keeping each row in the result. The set of related rows is defined by a window specification. See Window Function.
Workload Group
A Workload Group provides resource isolation within a backend (BE) process. It controls how queries use CPU, memory, concurrency, queue capacity, scan threads, and read throughput. See Workload Group.
Workload Policy
A Workload Policy is a workload-management rule that evaluates query or user conditions and backend runtime metrics to apply actions such as canceling a query. See Workload Policy.
Z
Zstandard (zstd)
Zstandard, abbreviated as zstd, is a lossless compression algorithm that provides a high compression ratio with fast decompression. VeloDB Cloud uses it as the default column compression and accepts zstd-compressed payloads for Stream Load. See Column Compression.