AWS → Scaleway

Migrating from AWS to Scaleway

Scaleway is a French provider (Iliad group) with three EU regions and S3- and CNCF-compatible building blocks, so the commodity layer of an AWS migration is closer to configuration than rewrite. This page is concrete about what moves cleanly and what does not: the two managed services that have no Scaleway twin, where the S3 and SQS/SNS APIs actually diverge, and a realistic engineer-day range.

Which AWS services have a true Scaleway equivalent?

Row by row, graded conservatively. “Equivalent” means it repoints by configuration; “partial” means it works with real feature deltas; “no managed twin” means you self-host or reduce scope.

AWS services and their Scaleway equivalents, each with a migration verdict.
AWSScalewayVerdict
Amazon S3Scaleway Object Storage S3-compatible API; most SDKs, backups and static sites repoint by endpoint. Object Lock and fine-grained bucket policies cover fewer operations. Audit any WORM/compliance dependency.Equivalent
Amazon EBSScaleway Block Storage Network block volumes with snapshots. Direct equivalent.Equivalent
Amazon EC2Scaleway Instances General-purpose, GPU and dedicated instances across the three regions.Equivalent
Amazon EKSScaleway Kapsule CNCF-conformant managed control plane; portable manifests move with little friction.Equivalent
Amazon RDS / AuroraScaleway Managed Database (PostgreSQL) Standard PostgreSQL is well served; Aurora global database and Serverless v2 autoscaling have no twin, so plan around them.Partial
Amazon ElastiCacheScaleway Managed Database for Redis Managed Redis/Valkey; in-memory caching is well covered.Equivalent
Amazon SQS / SNSScaleway Messaging & Queuing SQS- and SNS-compatible endpoints, but fan-out, message filtering and FIFO guarantees do not map one-to-one; complex topologies need re-testing, not just re-pointing.Partial
AWS LambdaScaleway Serverless Functions HTTP and cron are fine; Lambda’s breadth of native event sources and cold-start tuning is not matched.Partial
Amazon DynamoDBNo managed twin (self-host ScyllaDB / Cassandra) No managed serverless wide-column NoSQL; you run the cluster and lose the pay-per-request model.No managed twin
Amazon RedshiftNo managed twin (self-host ClickHouse / Aiven) No sovereign serverless warehouse; ClickHouse is fast but a different operating model.No managed twin

Where do the APIs actually diverge?

The mature rows repoint cleanly. These are the specific edges where “S3-compatible” and “SQS-compatible” stop being drop-in.

  • Object Storage is S3-compatible, but Object Lock/immutability and fine-grained bucket policies cover fewer operations than S3. Verify any compliance-grade retention before you cut over.
  • Messaging & Queuing speaks SQS/SNS, yet SNS-style fan-out, message filtering and FIFO ordering do not map one-to-one; event topologies need re-testing.
  • There is no Kinesis-compatible stream. Event pipelines retarget to managed Kafka (Aiven) or self-hosted NATS.
  • DynamoDB and Redshift have no managed twin; those two AWS services drive most of the self-hosting effort.

Which of your services move cleanly?

Scoped to the rows Scaleway covers as managed services. Search or filter by verdict.

14 capabilities

  • Object storage

    Hyperscaler Amazon S3 Azure Blob Storage Google Cloud Storage
    EU-sovereign path Scaleway Object StorageOVHcloud Object Storage
    Verdict Mature

    S3-compatible API. A drop-in target for most SDKs, backups and static assets.

  • Block storage

    Hyperscaler Amazon EBS Azure Managed Disks Persistent Disk / Hyperdisk
    EU-sovereign path Scaleway Block StorageOVHcloud Block Storage
    Verdict Mature

    Network block volumes with snapshots. Standard building block, well covered.

  • Virtual machines

    Hyperscaler Amazon EC2 Azure Virtual Machines Google Compute Engine
    EU-sovereign path Scaleway InstancesOVHcloud Public CloudHetzner Cloud
    Verdict Mature

    General-purpose and dedicated instances are a solved problem in the EU.

  • Managed Kubernetes

    Hyperscaler Amazon EKS Azure AKS Google Kubernetes Engine (GKE)
    EU-sovereign path Scaleway KapsuleOVHcloud Managed Kubernetes
    Verdict Mature

    CNCF-conformant managed control planes. Portable workloads move with little friction.

  • Container registry

    Hyperscaler Amazon ECR Azure Container Registry Google Artifact Registry
    EU-sovereign path Scaleway Container RegistryOVHcloud Managed Private RegistryHarbor (self-hosted)
    Verdict Mature

    OCI registries are commodity. Self-hosted Harbor is a fully sovereign fallback.

  • Managed PostgreSQL

    Hyperscaler Amazon RDS / Aurora Azure Database for PostgreSQL Cloud SQL / AlloyDB for PostgreSQL
    EU-sovereign path Scaleway Managed DatabaseOVHcloud Managed DatabasesAiven for PostgreSQL
    Verdict Viable

    Standard PostgreSQL is well served. Aurora-specific features (e.g. global database, serverless v2 autoscaling) have no exact twin, so plan around them.

  • Managed cache (Redis/Valkey)

    Hyperscaler Amazon ElastiCache Azure Cache for Redis Memorystore for Redis
    EU-sovereign path Scaleway Managed Database for RedisAiven for Valkey/Redis
    Verdict Viable

    In-memory caching is well covered; Valkey is the emerging open successor to Redis.

  • Message queue / pub-sub

    Hyperscaler Amazon SQS / SNS Azure Service Bus Google Cloud Pub/Sub
    EU-sovereign path Scaleway Messaging & QueuingSelf-hosted NATS / RabbitMQ
    Verdict Viable

    Scaleway offers SQS/SNS-compatible endpoints; self-hosted NATS covers the rest.

  • Authoritative DNS

    Hyperscaler Amazon Route 53 Azure DNS Google Cloud DNS
    EU-sovereign path Scaleway Domains & DNSOVHcloud DNSBunny DNS
    Verdict Viable

    Managed authoritative DNS is well covered. Route 53's routing policies need a manual rebuild.

  • Transactional email

    Hyperscaler Amazon SES Azure Communication Services No first-party service (SendGrid via Marketplace)
    EU-sovereign path Scaleway Transactional EmailBrevo (FR)
    Verdict Viable

    Scaleway TEM and Brevo (French) cover sending, reputation and analytics from the EU.

  • Secrets & key management

    Hyperscaler AWS Secrets Manager / KMS Azure Key Vault Secret Manager / Cloud KMS
    EU-sovereign path Scaleway Secret ManagerHashiCorp Vault (self-hosted)
    Verdict Viable

    Managed secret storage exists; self-hosted Vault gives full control including HSM-backed keys.

  • Serverless containers

    Hyperscaler AWS Fargate Azure Container Instances Google Cloud Run
    EU-sovereign path Scaleway Serverless Containers
    Verdict Viable

    Scale-to-zero container hosting exists and is production-ready for request-driven workloads.

  • Serverless functions (FaaS)

    Hyperscaler AWS Lambda Azure Functions Cloud Run functions (Cloud Functions)
    EU-sovereign path Scaleway Serverless FunctionsOVHcloud Functions
    Verdict Partial

    The runtimes exist, but Lambda's breadth of native event sources and mature cold-start tuning is not matched. Fine for HTTP and cron; harder for deep event-driven fan-out.

  • Managed ML platform

    Hyperscaler Amazon SageMaker Azure Machine Learning Vertex AI
    EU-sovereign path OVHcloud AI Endpoints / AI TrainingScaleway GPU Instances + Managed Inference
    Verdict Gap

    GPU compute, training and inference endpoints exist, but nothing matches SageMaker's end-to-end breadth (pipelines, feature store, model registry, tuning) as one managed platform. You assemble it.

Verdicts are conservative and reflect managed, EU-jurisdiction offerings as of 2026. Provider feature sets move quickly, so we re-check on every engagement.

How many engineer-days does this take?

30–70 engineer-days

A directional range for a mid-sized estate, not a quote. The spread is driven by how much of your stack sits on the gap rows.

  • A mid-sized estate: ~20–40 VMs, two or three managed PostgreSQL instances, object storage, one or two Kubernetes clusters and CI/CD.
  • The low end assumes S3-, Kubernetes- and PostgreSQL-shaped workloads that repoint by configuration.
  • The high end assumes a DynamoDB or Redshift dependency you must re-platform onto ScyllaDB or ClickHouse.
  • Excludes application rewrites, data-gravity egress time and staff ramp-up on Scaleway tooling.

Where does Scaleway actually run?

Three European regions, each with three availability zones, enough for in-region HA and cross-region DR, but a smaller footprint than AWS’s global spread.

  • Paris (fr-par)Three availability zones (fr-par-1/2/3) for in-region HA.
  • Amsterdam (nl-ams)Three availability zones; a second EU jurisdiction for DR.
  • Warsaw (pl-waw)Three availability zones in central Europe.

What are the honest limitations?

  • A three-region footprint Paris, Amsterdam and Warsaw only, with no southern-EU or intercontinental regions. Latency-sensitive global users and wide multi-region DR need more planning than AWS demands.
  • No warehouse, DynamoDB or SageMaker twin Analytics- and ML-heavy estates self-host ClickHouse, ScyllaDB and a GPU/inference stack. That is real engineering on a different operating model.
  • Thinner serverless event ecosystem Serverless Functions handle HTTP and cron well, but Lambda’s deep native event-source catalogue is not matched.

What else does a CTO ask before committing?

Do our S3 and Kubernetes workloads move without code changes?
Largely yes. Object Storage is S3-compatible and Kapsule is CNCF-conformant, so object access and portable manifests repoint by endpoint. The rework concentrates on Aurora-specific features, DynamoDB, Redshift and deep Lambda event wiring.
Where will this migration actually cost time?
In the gap rows: a DynamoDB table, a Redshift warehouse or a complex SNS fan-out. Everything on the mature layer (storage, VMs, Kubernetes, PostgreSQL) is closer to configuration than rewrite.
Is Scaleway sovereign enough for our compliance posture?
Scaleway is French (Iliad group) with EU-only regions and no US parent, which clears most GDPR and data-residency requirements. For SecNumCloud-grade mandates, weigh the Outscale path as well.