Network Load Balancer now supports IPv6</h1> <article> <h1>Network Load Balancer now supports IPv6</h1> <p>Thu, 24 Sep 2026 12:00:00 +0000</p> <p>Network Load Balancers now support IPv6. You can create an IPv6 Network Load Balancer in the <a href="https://portal.exoscale.com/login?next=%2Fcompute%2Fload-balancers%2Fadd%3F"target="_blank" rel="noopener">Exoscale Portal</a> or with the Exoscale CLI.</p> <p></p> <p>An IPv6 Network Load Balancer gets a public IPv6 address and distributes TCP and UDP traffic across the instances in your Instance Pools, just like an IPv4 Network Load Balancer.</p> <h2>Why use it<span class="hx:absolute hx:-mt-20" id="why-use-it"></span> <a href="#why-use-it" class="subheading-anchor" aria-label="Permalink for this section"></a></h2><ul> <li><strong>Lower cost:</strong> Elastic IPv6 addresses are free, so you only pay the load balancer fee. An IPv6 Network Load Balancer costs <strong>EUR/CHF/USD 0.02083 per hour</strong>, about 40% less than the IPv4 version. Check the <a href="https://www.exoscale.com/pricing/"target="_blank" rel="noopener">pricing page</a> for details.</li> <li><strong>Reach IPv6 users directly:</strong> Serve users on IPv6-only and IPv6-first networks, natively and without translation layers.</li> <li><strong>Go fully IPv6:</strong> Your instances already get a free public IPv6 address. Now the load balancer in front of them can be IPv6 too.</li> <li><strong>Nothing to give up:</strong> Same health checks, load balancing strategies and 200 service limit as an IPv4 Network Load Balancer.</li> </ul> <h2>Before you start<span class="hx:absolute hx:-mt-20" id="before-you-start"></span> <a href="#before-you-start" class="subheading-anchor" aria-label="Permalink for this section"></a></h2><p>The target Instance Pool must have IPv6 enabled (dual-stack). Instance Pools are IPv4 only by default, so set public IP assignment to dual-stack. Instances without an IPv6 address won’t receive traffic.</p> <p><a href="https://community.exoscale.com/product/networking/nlb/"target="_blank" rel="noopener">Learn more about Network Load Balancers →</a></p> </article> <article> <h1>Managed ClickHouse® is now generally available</h1> <p>Tue, 22 Sep 2026 06:00:00 +0000</p> <p class=""><b>Managed ClickHouse®</b> leaves early access and is now generally available in Exoscale European zones, with the full <a href="https://www.exoscale.com/sla/" target="_blank" rel="noopener">DBaaS SLA</a> and self-service plans from Startup to Premium.</p> <div class="medium-insert-images"><figure class=""> </figure></div> <p class=""><b>What changes</b></p> <ul> <li><p><b>Create and manage services from the <a href="https://portal.exoscale.com/" target="_blank" rel="noopener">Exoscale Portal</a></b>, alongside the rest of our managed databases, as well as from the API</p></li> <li><p><b>Script it from the command line</b> with the <code>exo dbaas</code> commands of the <a href="https://github.com/exoscale/cli/releases/tag/v1.100.0" target="_blank" rel="noopener">Exoscale CLI</a> from v1.100.0, including ClickHouse users</p></li> <li><p><b>Manage it as code</b> with the <a href="https://registry.terraform.io/providers/exoscale/exoscale/latest/docs/resources/dbaas" target="_blank" rel="noopener">Exoscale Terraform provider</a> from v0.72.0, including ClickHouse users through the <code>exoscale_dbaas_clickhouse_user</code> resource</p></li> <li><p><b>Self-service from Startup to Premium</b>, with no per-organization activation step; the largest configurations remain available on request</p></li> <li><p><b>Backed by the DBaaS SLA</b>, 99.95% on Startup and 99.99% on Business and Premium</p></li> </ul> <p class=""><b>What has not changed</b></p> <p class="">Everything the service already did: sub-second queries on billions of rows, vector search with the HNSW index, tiered storage to <a href="https://www.exoscale.com/object-storage/" target="_blank" rel="noopener">Exoscale Object Storage</a>, federated queries against <a href="https://www.exoscale.com/dbaas/postgresql/" target="_blank" rel="noopener">Managed PostgreSQL</a>, and 100% standard, open-source ClickHouse® with no proprietary fork, billed per second.</p> <p class="">Get started with the <a href="https://community.exoscale.com/product/dbaas/service-specific/clickhouse/quick-start/" target="_blank" rel="noopener">quick-start guide</a>, learn more on the <a href="https://www.exoscale.com/dbaas/clickhouse/" target="_blank" rel="noopener">product page</a>, or compare plans and per-second rates on the <a href="https://www.exoscale.com/pricing/#managed-clickhouse" target="_blank" rel="noopener">pricing page</a>. New to Exoscale? <a href="https://portal.exoscale.com/register/" target="_blank" rel="noopener">Register</a> and start with free credit.</p> <p class=""><i>ClickHouse® is a registered trademark of ClickHouse, Inc. Exoscale is not affiliated with, sponsored by, or endorsed by ClickHouse, Inc.</i></p> </article> <article> <h1>New Works with Exoscale Partner: Airlock Microgateway</h1> <p>Thu, 17 Sep 2026 08:00:00 +0000</p> <p><a href="https://www.airlock.com/en/microgateway-evaluation"target="_blank" rel="noopener">Airlock Microgateway</a> by Ergon Informatik AG is now certified as Works with Exoscale, bringing Kubernetes-native application and API protection to workloads running on Exoscale SKS.</p> <p>Built on open standards, Airlock Microgateway runs directly inside the Kubernetes cluster and combines the Kubernetes Gateway API with Web Application and API Protection (WAAP), identity-aware access control, reporting and monitoring. Routing rules and security policies are managed as code, allowing teams to integrate security directly into existing GitOps and DevSecOps workflows.</p> <p></p> <p><strong>Security that fits into Kubernetes</strong></p> <p>Airlock Microgateway gives platform and application teams a consistent way to manage routing and security without adding a separate security layer outside the cluster:</p> <ul> <li><strong>Kubernetes-native protection:</strong> Protect web applications and APIs directly at the cluster entry point.</li> <li><strong>Security as code:</strong> Define routing and security policies declaratively and manage them through existing GitOps workflows.</li> <li><strong>Identity-aware access:</strong> Control access using standards including OIDC, JWT, OAuth 2.0 Token Exchange and mTLS.</li> <li><strong>Application and API protection:</strong> Protect workloads against common web attacks and enforce API specifications with OpenAPI and GraphQL schema validation.</li> <li><strong>Built on open standards:</strong> Use a conformant implementation of the Kubernetes Gateway API together with established technologies including Istio and Cilium.</li> <li><strong>A path beyond Ingress NGINX:</strong> Migrate existing Ingress NGINX resources to Gateway API using the Kubernetes ingress2gateway tool, including its Airlock Microgateway emitter.</li> </ul> <p>Airlock Microgateway can be deployed directly on Exoscale SKS. Its Community Edition provides Kubernetes Gateway API capabilities and reverse proxy capabilities without a license or throughput limit, while advanced security capabilities are available with the Premium Edition.</p> <p><a href="https://www.exoscale.com/technology-providers/works-with-exoscale/airlock-microgateway/"target="_blank" rel="noopener">Protect your Kubernetes workloads →</a></p> </article> <article> <h1>New on the Marketplace: DigLM, Swiss AI for companies</h1> <p>Wed, 16 Sep 2026 08:00:00 +0000</p> <p><a href="https://diglm.ch/"target="_blank" rel="noopener">DigLM</a> is now available on the Exoscale Marketplace. It gives Swiss organisations an AI workspace for drafting, summarising, translating and analysing documents, with customer data hosted in Switzerland.</p> <p></p> <p><strong>What you can do with it</strong></p> <ul> <li>Draft, summarise and analyse documents without sending user content to an external AI provider</li> <li>Translate business content between French, German, English and Italian</li> <li>Use 13 built-in skills for Swiss law, HR, finance, strategy, sales, marketing, IT and translation</li> <li>Share conversations and company context across a team, with role-based administration</li> <li>Upload documents directly into a conversation for contextual analysis</li> <li>Integrate DigLM into existing workflows through an OpenAI-compatible API</li> </ul> <p><strong>Why it matters</strong></p> <p>For Swiss SMEs and regulated professions, AI adoption often creates a trade-off between productivity and control over sensitive data. DigLM brings both together with an accessible AI interface that can be deployed in minutes, using open-source models running on DigLM servers hosted on Exoscale infrastructure in Geneva and Zurich.</p> <p>Customer data is isolated per client, encrypted in transit and at rest, and is not used to train models. The API is stateless, so prompts and responses are not stored.</p> <p><a href="https://www.exoscale.com/marketplace/listing/diglm/"target="_blank" rel="noopener">Explore DigLM on the Exoscale Marketplace</a></p> </article> <article> <h1>New on the Marketplace: SOKKAN, an AI-development cockpit built on Exoscale</h1> <p>Tue, 08 Sep 2026 08:00:00 +0000</p> <p>SOKKAN by Geneva-based Ninabot Sàrl is now available on the Exoscale Marketplace.</p> <p>SOKKAN is an open-source cockpit for teams developing with AI agents. It brings parallel Claude Code sessions, persistent project memory, a dynamic Kanban and human-in-the-loop approvals into one environment, helping teams work with AI agents while keeping developers in control.</p> <p></p> <p>Available as both an open-source self-hosted setup and a fully managed cloud service, SOKKAN Cloud runs on Exoscale infrastructure in Geneva, with a dedicated environment for each customer.</p> <p><strong>What SOKKAN brings to development teams</strong></p> <p>SOKKAN is designed around a simple principle: AI agents work better when they understand what came before and humans remain responsible for what happens next.</p> <ul> <li><strong>Persistent project context:</strong> Project memory carries knowledge across tasks and sessions, so new agent sessions can start with the context they need.</li> <li><strong>Parallel AI workflows:</strong> Run multiple Claude Code sessions at the same time and manage work from a dynamic Kanban.</li> <li><strong>Human control:</strong> Sensitive actions can require explicit approval before an agent proceeds.</li> <li><strong>Dedicated environments:</strong> Each SOKKAN Cloud customer gets a dedicated VM and private network on Exoscale.</li> <li><strong>Automated infrastructure:</strong> Terraform handles provisioning and the infrastructure lifecycle, with optional Managed PostgreSQL. Daily snapshots of each environment are stored in Exoscale Object Storage in Geneva, so customer data stays in Switzerland.</li> <li><strong>Open source and self-hostable:</strong> Run SOKKAN on your own infrastructure or choose the fully managed Swiss-hosted service.</li> </ul> <p>With Exoscale as its infrastructure foundation, SOKKAN Cloud can provision complete customer environments in minutes while keeping each tenant isolated and the underlying infrastructure simple to operate.</p> <p><a href="https://www.exoscale.com/marketplace/listing/sokkan/"target="_blank" rel="noopener"><strong>Explore SOKKAN on the Exoscale Marketplace</strong></a></p> </article> <article> <h1>SKS: Kubernetes 1.37 Now Available</h1> <p>Mon, 31 Aug 2026 11:45:00 +0000</p> <p>Kubernetes 1.37 is now available on Exoscale Scalable Kubernetes Service (SKS) for new and existing clusters.</p> <p></p> <p>Named <em>Garhwal</em>, Kubernetes 1.37 includes 67 enhancements, with 16 features graduating to Stable. The release strengthens established Kubernetes APIs and improves the foundations for secure workloads and reliable cluster upgrades.</p> <h3>Key benefits<span class="hx:absolute hx:-mt-20" id="key-benefits"></span> <a href="#key-benefits" class="subheading-anchor" aria-label="Permalink for this section"></a></h3><ul> <li><strong>A stable resource Metrics API:</strong> The <code>metrics.k8s.io/v1</code> API now carries Kubernetes’ Stable API guarantees. This gives tooling that uses pod and node CPU and memory metrics a more durable integration target. Existing <code>v1beta1</code> clients remain supported during the transition, and using the new API requires a compatible metrics provider.</li> <li><strong>Stronger foundations for workload identity:</strong> Pod Certificates and Cluster Trust Bundles are now Stable, providing standard Kubernetes building blocks for issuing X.509 certificates and distributing trust material to workloads. This can simplify TLS and mutual TLS integrations when used with a compatible certificate signer.</li> <li><strong>Safer mixed-version operation:</strong> Node Declared Features are now Stable. Control-plane components can identify the features available on each node, improving behavior when clusters temporarily contain different node versions during upgrades.</li> <li><strong>Continued security and reliability improvements:</strong> Upgrading keeps clusters on an actively supported Kubernetes release and provides the latest upstream fixes and improvements included in Kubernetes 1.37.</li> </ul> <h3>Upgrade notes<span class="hx:absolute hx:-mt-20" id="upgrade-notes"></span> <a href="#upgrade-notes" class="subheading-anchor" aria-label="Permalink for this section"></a></h3><ul> <li>Existing SKS clusters are not upgraded automatically.</li> <li>Review the <a href="https://kubernetes.io/blog/2026/08/26/kubernetes-v1-37-release/"target="_blank" rel="noopener">Kubernetes 1.37 release announcement</a> and test the upgrade in a non-production environment before upgrading production clusters.</li> <li>Upgrade the control plane and node pools using the Exoscale Portal, CLI, or API. Kubernetes minor versions must be upgraded sequentially.</li> </ul> <p><a href="https://community.exoscale.com/product/compute/containers/operation/lifecycle-management/"target="_blank" rel="noopener">Learn how to upgrade and manage the Kubernetes version of an SKS cluster</a></p> </article> <article> <h1>New partner: CC-Interactive joins Works with Exoscale</h1> <p>Mon, 24 Aug 2026 08:00:00 +0000</p> <p>CC-Interactive is joining the Works with Exoscale partner ecosystem with SmartBoard, its AI-powered business intelligence platform built on a Data Lakehouse.</p> <p></p> <p><strong>What you can do with it</strong></p> <ul> <li>Consolidate data from your finance, operational and business systems into one validated foundation, instead of exporting a spreadsheet from each of them</li> <li>Replace recurring manual reporting with reports generated in one click, with a full audit trail for submission to authorities or compliance platforms</li> <li>Ask why a number moved and get the root cause and a written summary back, rather than digging through dashboards yourself</li> <li>Forecast each KPI from historical patterns and external variables, and test what-if scenarios before committing to a decision</li> <li>Give stakeholders secure role-based access to the current version of a dashboard, so reporting stops travelling by email attachment</li> <li>Connect the sources yourself from the admin panel, through REST APIs, webhooks or direct PostgreSQL, Microsoft SQL Server and MySQL connections, with incremental loading configured per data point and no custom code</li> </ul> <p><strong>Why it matters</strong></p> <p>ESG, finance and compliance reporting keeps growing while the teams producing it do not, and most of that effort goes into collecting and reconciling data rather than interpreting it. SmartBoard moves that work to a single structured foundation, which shortens the path from raw data to a decision and makes each figure traceable back to its source.</p> <p>SmartBoard runs on Exoscale infrastructure in European data centers, which matters when the data being reported on is financial, personal or regulated.</p> <p><a href="https://www.exoscale.com/technology-providers/works-with-exoscale/cc-interactive/"target="_blank" rel="noopener">Explore the CC-Interactive partner page</a></p> </article> <article> <h1>New partner: Lombiq joins Works with Exoscale</h1> <p>Fri, 21 Aug 2026 08:00:00 +0000</p> <p>Lombiq is joining the Works with Exoscale partner ecosystem. Lombiq specializes in Orchard Core, the open-source content management framework for .NET, and takes part in developing Orchard Core itself.</p> <p></p> <p><strong>What you can do with them</strong></p> <ul> <li>Have an enterprise content platform, customer portal or e-commerce site built on Orchard Core, with custom modules and themes instead of off-the-shelf compromises</li> <li>Hand over an existing Orchard Core project for long-term maintenance, including applications that have to keep running for a decade or more</li> <li>Get immediate help when a production Orchard Core instance breaks, rather than waiting for a generalist agency to ramp up</li> <li>Ship web applications that meet WCAG 2.2 Level AA, verified by automated accessibility tests with their UI Testing Toolbox</li> <li>Train your own developers and editors on Orchard Core so your team can take the project forward itself</li> <li>Start on DotNest, their Orchard Core SaaS, when you would rather not operate the platform yourself</li> </ul> <p><strong>Why it matters</strong></p> <p>Orchard Core is a capable open-source CMS, but the pool of teams with genuine depth in it is small, and content platforms tend to outlive the agency that built them. Lombiq covers the whole life of a project, from the first module to troubleshooting a decade later, and their open-source contributions go back into the framework itself.</p> <p>Their solutions are designed to run on Exoscale infrastructure, in European data centers, alongside Exoscale compute, networking, managed databases and object storage.</p> <p><a href="https://www.exoscale.com/technology-providers/works-with-exoscale/lombiq/"target="_blank" rel="noopener">Explore the Lombiq partner page</a></p> </article> <article> <h1>Managed ClickHouse®: Real-Time Analytics, Fully Managed in Europe</h1> <p>Thu, 20 Aug 2026 08:00:00 +0000</p> <p class="">Real-time analytics shouldn't mean running your own database cluster. <b>Managed ClickHouse®</b> brings the fastest open-source column-oriented (OLAP) database to Exoscale as a fully managed service: run sub-second SQL queries on billions of rows while provisioning, scaling, backups, monitoring and upgrades are handled for you, on <a href="https://www.exoscale.com/compliance/" target="_blank" rel="noopener">GDPR-compliant</a> European infrastructure.</p> <p class=""></p> <p class=""><b>What you get</b></p> <ul> <li><p><b>Sub-second queries on billions of rows</b>, a vectorized, columnar engine built for dashboards, observability and user-facing analytics</p></li> <li><p><b>Vector search built in</b>, approximate nearest-neighbor search with the HNSW index, generally available as a ClickHouse 25.8 feature, for RAG pipelines, recommendations and semantic search</p></li> <li><p><b>Tiered storage</b>, recent data stays on fast local disk while historical data moves to S3-compatible <a href="https://www.exoscale.com/object-storage/" target="_blank" rel="noopener">Exoscale Object Storage</a> in the same European zone, keeping long-term retention affordable</p></li> <li><p><b>Federated queries</b>, read from <a href="https://www.exoscale.com/dbaas/postgresql/" target="_blank" rel="noopener">Managed PostgreSQL</a> and query S3-compatible <a href="https://www.exoscale.com/object-storage/" target="_blank" rel="noopener">Object Storage</a> directly in SQL, and build dashboards with Managed Grafana</p></li> <li><p><b>Fully managed and highly available</b>, automated operations with replication, failover and a 99.99% uptime SLA on Business and Premium plans</p></li> <li><p><b>European and open source</b>, 100% standard ClickHouse® running the 25.8 LTS release, no proprietary fork, in the country of your chosen zone, billed per second</p></li> </ul> <div style="border-left:4px solid #DA291C;background:#fdf6f5;border-radius:6px;padding:12px 18px;margin:18px 0"> <p class="" style="margin:0">Managed ClickHouse® is in <b>early access</b> in Exoscale European zones. <b>Startup</b> and <b>Business</b> plans are self-service; larger and <b>Premium</b> plans are enabled per organization on request. Create your service via the Exoscale CLI or API by following the <a href="https://community.exoscale.com/product/dbaas/service-specific/clickhouse/quick-start/" target="_blank" rel="noopener">quick-start guide</a>; Portal integration arrives with general availability. New to Exoscale? <a href="https://portal.exoscale.com/register/" target="_blank" rel="noopener">Register</a> and start with free credit.</p> </div> <p class=""><b>How it fits your stack</b></p> <p class=""></p> <p class="">Spin up a cluster in minutes and put your data to work. Learn more on the <a href="https://www.exoscale.com/dbaas/clickhouse/?utm_source=changelog.exoscale.com&utm_medium=changelog&utm_campaign=managed-clickhouse&utm_content=clickhouse_changelog" target="_blank" rel="noopener">Managed ClickHouse® product page</a>, read the <a href="https://community.exoscale.com/product/dbaas/service-specific/clickhouse/" target="_blank" rel="noopener">documentation</a>, or compare plans and per-second rates on the <a href="https://www.exoscale.com/pricing/#managed-clickhouse" target="_blank" rel="noopener">pricing page</a>.</p> <p class="" style="font-size:0.85em;color:#6b7280"><i>ClickHouse® is a registered trademark of ClickHouse, Inc. Exoscale is not affiliated with, sponsored by, or endorsed by ClickHouse, Inc.</i></p> </article> <article> <h1>SKS: Container registry mirroring for Karpenter nodes</h1> <p>Wed, 19 Aug 2026 08:00:00 +0000</p> <p>Exoscale SKS now supports configuring OCI registry mirrors for nodes provisioned by Karpenter. Registry mirror endpoints and pull credentials can be defined in an <code>ExoscaleNodeClass</code> and applied consistently to new nodes created from that class.</p> <p><strong>What this enables</strong></p> <ul> <li><strong>Faster and more reliable image pulls:</strong> Route image requests through private mirrors located closer to your workloads.</li> <li><strong>Lower egress costs:</strong> Reduce repeated downloads from public registries by caching commonly used container images.</li> <li><strong>Fewer upstream limits:</strong> Avoid relying solely on public registries and reduce the impact of their rate limits or availability issues.</li> <li><strong>Greater supply-chain control:</strong> Use private registries, authenticated mirrors, and trusted TLS configurations for container images.</li> </ul> <p><strong>Why it matters</strong></p> <p>Registry mirroring gives teams more control over how container images reach their SKS clusters. It improves reliability and supports production environments that require private access, predictable image delivery, or tighter control over external dependencies.</p> <p><a href="https://community.exoscale.com/product/compute/containers/how-to/container-registry-mirroring/"target="_blank" rel="noopener">Learn how to configure OCI registry mirroring for Karpenter nodes.</a></p> </article> <article> <h1>New on the Marketplace: BunkerWeb Community</h1> <p>Tue, 18 Aug 2026 08:00:00 +0000</p> <p>BunkerWeb Community is now available on the Exoscale Marketplace. It is an open-source, cloud-native Web Application Firewall (WAF) and reverse proxy from Bunkerity that you put in front of your websites, applications and APIs.</p> <p></p> <p><strong>What you can do with it</strong></p> <ul> <li>Add a security layer in front of an existing application without touching its code, by routing traffic through BunkerWeb as a reverse proxy</li> <li>Block the common attack classes of the OWASP Top 10 before they reach your application</li> <li>Slow down credential stuffing, scraping and brute-force attempts with rate limiting, bot protection, IP filtering and GeoIP filtering</li> <li>Serve everything over HTTPS with automated certificate issuance and renewal instead of managing TLS by hand</li> <li>Protect Linux, Docker, Swarm and Kubernetes workloads with the same tool, configured from a web interface or driven by API</li> <li>Extend it with plugins and wire it into your existing deployment and automation pipelines</li> </ul> <p><strong>Why it matters</strong></p> <p>Internet-facing applications need a filtering layer, and the usual choice is between a managed service that routes your traffic through a third party and a WAF you assemble yourself. BunkerWeb Community is a ready-to-run alternative that stays inside your own Exoscale environment, in the zone you choose, with no traffic leaving to an external provider and no per-request pricing.</p> <p>It is licensed under the AGPLv3, so only the instance it runs on is billed. If you later need tailored SLAs, BunkerWeb PRO is available from Bunkerity via subscription.</p> <p>To get started, point a DNS A record to the IP address of your instance, then open the setup wizard at your fully qualified domain name.</p> <p><a href="https://www.exoscale.com/marketplace/listing/bunkerweb-community/"target="_blank" rel="noopener">Explore BunkerWeb Community on the Exoscale Marketplace</a></p> </article> <article> <h1>SKS: CPU Manager policies for Karpenter-provisioned nodes</h1> <p>Mon, 17 Aug 2026 08:00:00 +0000</p> <p>Exoscale SKS now supports configuring Kubernetes CPU Manager policies for nodes provisioned by Karpenter. You can define the kubelet CPU Manager configuration in an <code>ExoscaleNodeClass</code> and apply it consistently to every node created from that class.</p> <p><strong>What this enables</strong></p> <ul> <li><strong>More predictable performance:</strong> Assign exclusive CPUs to eligible workloads that are sensitive to latency or CPU contention.</li> <li><strong>Consistent node configuration:</strong> Apply the selected CPU policy automatically to new nodes provisioned by Karpenter.</li> <li><strong>Flexible workload isolation:</strong> Reserve CPU capacity for Kubernetes and system components while keeping application resources separate.</li> </ul> <p><strong>Why it matters</strong></p> <p>CPU Manager policies give teams greater control over how processor resources are allocated across their SKS workloads. This is especially useful for high-performance, real-time, and other compute-intensive applications that benefit from stable CPU placement.</p> <p><a href="https://community.exoscale.com/product/compute/containers/how-to/cpu-manager-policies/"target="_blank" rel="noopener">Learn how to configure CPU Manager policies with Karpenter.</a></p> </article> <article> <h1>New on the Marketplace: Anapaya EDGE for secure SCION connectivity</h1> <p>Fri, 14 Aug 2026 08:00:00 +0000</p> <p><a href="https://www.anapaya.net/edge"target="_blank" rel="noopener">Anapaya EDGE</a> is now available on the Exoscale Marketplace, bringing secure and resilient SCION connectivity directly to workloads running on Exoscale.</p> <p>Anapaya EDGE is a virtual network appliance that connects existing networks and services to the SCION Internet. Standard IP traffic is carried over SCION without requiring changes to applications or end-user devices, making it possible to introduce SCION into existing infrastructure without redesigning the application environment.</p> <p></p> <p><strong>More control over critical network paths</strong></p> <p>For organizations where network availability, security, and control matter, Anapaya EDGE adds capabilities that go beyond traditional Internet connectivity:</p> <ul> <li><strong>Resilient connectivity:</strong> Multipath routing provides sub-second failover to alternative network paths in case of disruption.</li> <li><strong>End-to-end path control:</strong> Traffic paths can be selected based on criteria such as latency, provider preference, cost, or geopolitical boundaries.</li> <li><strong>Controlled access:</strong> Critical services can be restricted to trusted groups of peers, reducing their exposure to the public Internet.</li> <li><strong>EDGE-to-EDGE encryption:</strong> Traffic between locations can be encrypted and authenticated without introducing a traditional VPN setup.</li> <li><strong>Integration with existing environments:</strong> Existing applications and end-user devices can continue using standard IP connectivity.</li> </ul> <p>Anapaya EDGE is available across all Exoscale zones as a Bring Your Own License (BYOL) Marketplace appliance. A 30-day try-and-buy period is available before a license is required.</p> <div class="hx:overflow-x-auto hx:mt-6 hx:flex hx:flex-col hx:rounded-lg hx:border hx:py-4 hx:px-4 hx:border-gray-200 hx:contrast-more:border-current hx:contrast-more:dark:border-current hx:border-blue-200 hx:bg-blue-100 hx:text-blue-900 hx:dark:border-blue-200/30 hx:dark:bg-blue-900/30 hx:dark:text-blue-200"> <p class="hx:flex hx:items-center hx:font-medium"><svg height=16px class="hx:inline-block hx:align-middle hx:mr-2" xmlns="http://www.w3.org/2000/svg" fill="none" viewBox="0 0 24 24" stroke-width="2" stroke="currentColor" aria-hidden="true"><path stroke-linecap="round" stroke-linejoin="round" d="M13 16h-1v-4h-1m1-4h.01M21 12a9 9 0 11-18 0 9 9 0 0118 0z"/></svg>Note</p> <div class="hx:w-full hx:min-w-0 hx:leading-7"> <div class="hx:mt-6 hx:leading-7 hx:first:mt-0"><p>SCION connectivity through a SCION-enabled network service provider is required.</p></div> </div> </div> <p><a href="https://www.exoscale.com/marketplace/listing/anapaya-edge/"target="_blank" rel="noopener">Explore Anapaya EDGE on the Exoscale Marketplace</a></p> </article> <article> <h1>SKS: GPU MIG partitioning support</h1> <p>Tue, 21 Jul 2026 00:00:00 +0000</p> <p>Exoscale SKS now supports GPU MIG partitioning for MIG-capable GPU nodepools.</p> <p>With Multi-Instance GPU (MIG), customers can split a supported physical GPU into multiple isolated GPU instances, each with dedicated compute and memory resources. This makes it easier to run several smaller GPU workloads on the same SKS node instead of assigning a full GPU to each workload.</p> <p></p> <h3>Availability and setup<span class="hx:absolute hx:-mt-20" id="availability-and-setup"></span> <a href="#availability-and-setup" class="subheading-anchor" aria-label="Permalink for this section"></a></h3><ul> <li>MIG partitioning is available for Exoscale <a href="https://www.exoscale.com/gpu/a30/"target="_blank" rel="noopener">A30</a>, <a href="https://www.exoscale.com/gpu/rtx6000/"target="_blank" rel="noopener">RTX Pro 6000</a> and <a href="https://www.exoscale.com/gpu/b300/"target="_blank" rel="noopener">B300</a> GPUs.</li> <li>Standard SKS nodepools can be configured with MIG profiles through the Exoscale Portal, CLI, or Terraform provider.</li> <li>Karpenter-managed GPU nodepools can use MIG through the <code>ExoscaleNodeClass</code> configuration.</li> </ul> <h3>Why use MIG?<span class="hx:absolute hx:-mt-20" id="why-use-mig"></span> <a href="#why-use-mig" class="subheading-anchor" aria-label="Permalink for this section"></a></h3><ul> <li>Run multiple smaller GPU workloads on one physical GPU.</li> <li>Improve GPU utilization and reduce underused capacity.</li> <li>Keep workloads isolated at the GPU partition level.</li> <li>Better accommodate AI inference, development and testing, and smaller batch workloads.</li> </ul> <blockquote> <p><strong>Note:</strong> Existing nodes must be cycled before a changed MIG profile takes effect. Billing is based on the selected GPU node instance type, not on individual MIG partitions.</p> </blockquote> <p>For setup instructions, see:</p> <p><a href="https://community.exoscale.com/product/compute/containers/how-to/gpu-mig-partitioning/"target="_blank" rel="noopener">Configure GPU MIG partitioning on SKS</a></p> </article> <article> <h1>Archival Storage: Cold Data at 77% Below Standard Pricing</h1> <p>Mon, 06 Jul 2026 00:00:00 +0000</p> <p class="">Cold data shouldn't cost the same as hot data. <b>Archival Storage</b> stores cold data on Simple Object Storage (SOS) at <a href="https://www.exoscale.com/object-storage/classes/" target="_blank" rel="noopener">€0.0042 per GB per month, 77% below Standard pricing</a>, with free retrieval, on the same S3-compatible infrastructure you already use. It is built for genuine archival workloads: cold archives, long-term backups, and retention data you rarely access.</p> <p class=""><b>How it works</b></p> <ul> <li><p>Request activation through the <a href="https://app.formbricks.com/s/cmpy13efand7qsm01pd7oo85s?utm_source=changelog.exoscale.com&utm_medium=changelog&utm_campaign=storage-classes&utm_content=archival_changelog" target="_blank" rel="noopener">activation form</a>, one request per bucket, reviewed case by case</p></li> <li><p>Once approved, your usage is billed at the Standard rate, then the difference down to the Archival rate is credited as a discount in the following billing period</p></li> <li><p>Same buckets, same API, same European zone, nothing changes for your applications</p></li> </ul> <div style="border-left:4px solid #0b7285;background:#f1f8fa;border-radius:6px;padding:12px 18px;margin:18px 0"> <p class="" style="margin:0"><b>Sizing: each bucket is billed for at least 10 TB.</b> A bucket on Archival Storage is billed for a minimum of 10 TB per month, metered as 9,313 GiB since Object Storage bills in GiB, even if it holds less. That is €42 per month, against €184 for the same 10 TB on Standard. Archival becomes cheaper than Standard from roughly 2.3 TB in a bucket, so it fits large, genuinely cold datasets rather than small ones. Estimate yours in the <a href="https://calculator.exoscale.com/storage/" target="_blank" rel="noopener">storage calculator</a>.</p> </div> <p class=""><b>Key conditions</b></p> <ul> <li><p><b>At least 10 TB billed per bucket</b>, whatever the bucket actually holds</p></li> <li><p>No object lock or retention on enrolled buckets: both arrive with the Cold Storage Class</p></li> <li><p>When the Warm and Cold Storage Classes launch in early 2027, a 30-day window is open to move each bucket to the Class of your choice. Cold Storage keeps the same €0.0042 per GB per month, without the 10 TB minimum. Buckets not migrated within the window return to Standard pricing</p></li> </ul> <p class=""><b>Timeline Overview</b></p> <p class=""></p> <p class="">If you keep large, rarely-accessed datasets on Standard storage, Archival pricing can substantially lower their cost. See the full details on the <a href="https://www.exoscale.com/object-storage/classes/" target="_blank" rel="noopener">Storage Classes page</a>, <a href="https://app.formbricks.com/s/cmpy13efand7qsm01pd7oo85s?utm_source=changelog.exoscale.com&utm_medium=changelog&utm_campaign=storage-classes&utm_content=archival_changelog" target="_blank" rel="noopener">request activation</a> for your buckets, or estimate your savings with the <a href="https://calculator.exoscale.com/storage/" target="_blank" rel="noopener">storage calculator</a>.</p> </article> <article> <h1>Exoscale Key Management Service (KMS) is now available</h1> <p>Thu, 02 Jul 2026 00:00:00 +0000</p> <p><strong>Exoscale Key Management Service (KMS)</strong> is now available, bringing centralized management to encryption keys across cloud workloads, applications, and integrations.</p> <p>This fully managed service enables organizations to create keys, control their lifecycle, replicate them across zones, and perform cryptographic operations.</p> <h3>What’s new<span class="hx:absolute hx:-mt-20" id="whats-new"></span> <a href="#whats-new" class="subheading-anchor" aria-label="Permalink for this section"></a></h3><ul> <li>Create and manage <strong>cryptographic keys</strong> and use them for your cryptographic workloads.</li> <li>Control key lifecycle, including **rotation schedules, deletion schedules and key status.</li> <li>Replicate keys across <strong>Exoscale zones</strong> for multi-zone deployments.</li> <li>Benefit from predictable <a href="https://www.exoscale.com/pricing/#security"target="_blank" rel="noopener">per-key pricing</a>, with all operations included.</li> </ul> <h3>What this means in practice<span class="hx:absolute hx:-mt-20" id="what-this-means-in-practice"></span> <a href="#what-this-means-in-practice" class="subheading-anchor" aria-label="Permalink for this section"></a></h3><p>Exoscale KMS gives you greater control over how encryption keys are secured and utilized across sensitive workloads. It helps meet compliance requirements, satisfy internal security policies, and protect regulated workloads through centralized lifecycle management. This also introduces expanded options and flexibility for choosing and managing the exact keys used for Encryption-at-Rest across your Exoscale environment.</p> <p><a href="https://community.exoscale.com/product/security/kms/"target="_blank" rel="noopener">Explore Exoscale KMS</a></p> </article> <article> <h1>New on the Marketplace: Anantis TrapEye</h1> <p>Thu, 18 Jun 2026 00:00:00 +0000</p> <p class=""><a href="https://anantis.io/" target="_blank" rel="noopener">Anantis TrapEye</a> is now available on the Exoscale Marketplace. It adds deception-based threat detection to environments running on Exoscale.</p> <p class="">Trapeye works differently from traditional security tools. Instead of hunting for threats hidden in billions of events, it deploys controlled decoys inside your infrastructure. Any interaction with these decoys is malicious by design, resulting in high-fidelity alerts with near-zero false positives.</p> <p class=""></p> <p class=""><b>What does this mean in practice?</b></p> <ul> <li><p>Detect attackers at their first interaction and reduce detection time (Mean Time to Detect, MTTD) significantly</p></li> <li><p>High-confidence alerts with almost no false positives</p></li> <li><p>Visibility into lateral movement and attacker behavior</p></li> <li><p>Works across cloud, hybrid, and on-prem environments</p></li> </ul> <p class=""><b>Why this matters?</b></p> <p>Modern attackers don’t cause immediate visible damage and often remain undetected for months. By the time they are discovered, it’s often too late.</p> <p class="">TrapEye shifts the power balance back to the defender by turning your infrastructure into a hostile environment for intruders, exposing them instantly as they move through your network.</p> <p><a href="https://www.exoscale.com/marketplace/listing/anantis-trapeye/"target="_blank" rel="noopener">Explore deployment details and launch Anantis Trapeye directly from the Marketplace</a></p> </article> <article> <h1>Manage Exoscale infrastructure with Crossplane</h1> <p>Wed, 17 Jun 2026 00:00:00 +0000</p> <div class="medium-insert-images"> <figure class="resized" style="max-width: 350.968px; max-height: 108.972px;"> </figure> </div> <p class="">Exoscale now offers two official Crossplane providers, so platform teams can manage Exoscale infrastructure declaratively from Kubernetes. Two official Crossplane providers are now available, both built with Upjet:</p> <ol> <li><b>provider-exoscale</b>: compute instances, SKS clusters, managed databases (DBaaS), IAM, and networking.</li> <li><b>provider-exoscale-sos</b>: S3-compatible SOS object storage, covering buckets, versioning, ACLs, CORS, and cross-zone replication.</li> </ol> <p class="">Crossplane runs as a controller and reconciles continuously: it detects configuration drift and brings resources back to their declared state, where a Terraform plan only checks when a run is triggered.</p> <p><a href="https://community.exoscale.com/reference/crossplane/"target="_blank" rel="noopener">Get started with Crossplane</a></p> </article> <article> <h1>AI: Dedicated Inference is Now Generally Available</h1> <p>Mon, 15 Jun 2026 00:00:00 +0000</p> <p><a href="https://www.exoscale.com/ai-cloud-infrastructure/dedicated-inference/"target="_blank" rel="noopener">Dedicated Inference</a> is now generally available.</p> <p>Deploy AI models on dedicated NVIDIA GPUs through managed inference endpoints hosted in European data centers, without managing Kubernetes clusters, inference frameworks, or supporting infrastructure.</p> <p>With this release:</p> <ul> <li>Dedicated Inference is covered by a <strong>99.95% SLA</strong>.</li> <li>Documentation has been expanded with deployment scaling guidance, updated CLI references, service boundaries, and model compatibility requirements.</li> <li>Support and operational processes are now part of the standard Exoscale service lifecycle.</li> </ul> <p></p> <p>Dedicated Inference is designed for production AI workloads, including LLMs, embeddings, RAG applications, AI agents, and custom inference APIs running on dedicated GPU resources.</p> <p><a href="https://community.exoscale.com/product/concrete-ai/dedicated-inference/quick-start/"target="_blank" rel="noopener">Get started with Dedicated Inference</a></p> </article> <article> <h1>New on the Marketplace: Zephior AI and its AI Agent Ziva</h1> <p>Fri, 05 Jun 2026 15:30:00 +0000</p> <p>We’re happy to welcome Zephior to the Exoscale Marketplace.</p> <p>Zephior helps proposal, procurement, legal, compliance, and sales teams handle RFPs, security questionnaires, public tenders, and complex procurement workflows in one workspace.</p> <p>Teams can monitor public procurement portals such as SIMAP and EU Tender, score opportunities against their company profile, review strengths and gaps, and convert a qualified opportunity into a response project.</p> <p></p> <p>The platform combines:</p> <ul> <li>Tender discovery and fit scoring across selected public procurement portals</li> <li>AI-assisted drafting for RFPs, questionnaires, and procurement documents</li> <li>Cited responses generated from approved company knowledge</li> <li>Traceable review, approval, and version history</li> </ul> <p>For organizations with stricter sovereignty or compliance requirements, Zephior offers a Switzerland-based deployment option on Exoscale infrastructure.</p> <p><a href="https://www.exoscale.com/marketplace/listing/zephior-ai-procurement/"target="_blank" rel="noopener">Explore Zephior on the Exoscale Marketplace</a></p> </article> </main></body></html>