--- title: "Glossary" description: "Glossary of Exoscale platform terms — instances, networking, storage, IAM and AI — each defined where the product documentation uses it." url: https://community.exoscale.com/platform/glossary/ section: platform last_updated: 2026-09-29 --- > For AI agents: the documentation index is at https://community.exoscale.com/llms.txt. Every page is available as markdown at `index.md` or with `Accept: text/markdown`. # Glossary Terms used across the Exoscale documentation, with links to the pages that define and use them. Anti-Affinity Group : A placement rule that distributes member instances across different physical hypervisors, reducing the impact of a host failure. See the [Compute instances overview](https://community.exoscale.com/product/compute/instances/overview/index.md). API key : A credential attached to an IAM role, responsible for authentication when accessing Exoscale programmatically. See the [IAM quick start](https://community.exoscale.com/product/security/iam/quick-start/index.md). DBaaS : Exoscale's Database as a Service deploys fully managed data stores on demand, with enterprise-grade automation, European data residency and clusters with up to 99.99% uptime SLAs. See the [DBaaS overview](https://community.exoscale.com/product/dbaas/overview/index.md). Dedicated Inference : A managed AI service within the Concrete AI product suite that deploys and runs AI models as production-ready inference endpoints on Exoscale's dedicated, sovereign GPU infrastructure, exposed through an OpenAI-compatible API. See the [Dedicated Inference overview](https://community.exoscale.com/product/concrete-ai/dedicated-inference/overview/index.md). Deploy Target : Also called a Deployment Target; a virtual abstraction layer that lets you provision Compute instances onto dedicated (single-tenant) infrastructure using a specific Deployment Target ID, typically a pool of dedicated hypervisors managed by Exoscale. See the [Dedicated Hypervisors overview](https://community.exoscale.com/product/compute/dedicated-hypervisors/overview/index.md). Elastic IP (EIP) : A static IP address that can be persistently attached to multiple Compute instances within the same zone, allowing seamless transition and consistent IP configuration across instances. See the [Elastic IP overview](https://community.exoscale.com/product/networking/eip/overview/index.md). IAM policy : A set of permissions attached to an identity that defines what actions it can perform on which resources. See the [IAM overview](https://community.exoscale.com/product/security/iam/overview/index.md). IAM role : A collection of policies that can be assigned to an identity to streamline permission assignments for specific needs or job functions. See the [IAM overview](https://community.exoscale.com/product/security/iam/overview/index.md). Instance Pool : A managed group of Compute instances used to maintain capacity and support automated scaling and replacement. See the [Compute instances overview](https://community.exoscale.com/product/compute/instances/overview/index.md). Instance Type : The resource configuration of a Compute instance — number of vCPUs, amount of RAM, and local SSD storage. Exoscale offers instance families optimized for CPU, memory, storage, or GPU workloads. See the [Compute instances overview](https://community.exoscale.com/product/compute/instances/overview/index.md). Key Management Service (KMS) : A centralized, fully managed security service that creates, controls, and manages the lifecycle of cryptographic keys, providing data encryption across Exoscale workloads, custom applications, and external integrations. See the [KMS overview](https://community.exoscale.com/product/security/kms/overview/index.md). Network Load Balancer (NLB) : A Layer 4 (TCP/UDP) load balancer that distributes incoming traffic to Compute instances managed by an Instance Pool, with round robin, source-hash, and MagLev strategies plus health checks. See the [NLB overview](https://community.exoscale.com/product/networking/nlb/overview/index.md). Organization : The context in which any action on the Exoscale platform is performed; an organization groups users, resources, and billing. See the [Organization documentation](https://community.exoscale.com/platform/organization/index.md). Private Connect : Connects on-premises or hosted infrastructure with one or more private networks from a single Exoscale organization. See the [Private Connect overview](https://community.exoscale.com/product/networking/private-connect/overview/index.md). Private Network : A private network segment shared only with your own instances; each instance may provision one or more additional network interfaces bound to it. See the [Private Network overview](https://community.exoscale.com/product/networking/private-network/overview/index.md). Scalable Kubernetes Service (SKS) : A fully managed Kubernetes offering that simplifies the deployment, scaling, and management of containerized applications, with a production-ready control plane and full Kubernetes ecosystem support. See the [SKS overview](https://community.exoscale.com/product/compute/containers/overview/index.md). Security Group : A set of stateful ingress and egress firewall rules, managed at the hypervisor level, that control traffic to and from an instance. See the [Security Group overview](https://community.exoscale.com/product/networking/security-group/overview/index.md). Simple Object Storage (SOS) : A highly durable, S3-compatible object storage service that offers data accessibility across multiple locations over HTTP(s) and supports a wide range of tools, libraries, and programming languages. See the [SOS overview](https://community.exoscale.com/product/storage/object-storage/overview/index.md). Template : A pre-configured disk image used to launch instances; templates define the base OS and system configuration, and can be public (Exoscale-provided) or custom (uploaded by you). See the [Compute instances overview](https://community.exoscale.com/product/compute/instances/overview/index.md). Zone : A location in which an instance and its zone-scoped resources are created; an instance cannot be moved to another zone after creation. See the [Compute instances overview](https://community.exoscale.com/product/compute/instances/overview/index.md) and [Data Center Zones](https://community.exoscale.com/platform/dc-zones/index.md).