The application runs in your own data center or with a European cloud provider. Yet its public entry point still depends on another platform: that is where DNS settings, certificates, routing rules, and security functions reside.
Anyone considering a change of provider first needs to understand which responsibilities this layer handles for their application.
With the ayedo Edge Cloud, public access can be designed independently of the application's location. Organizations can continue using their existing compute infrastructure and evaluate the upstream edge layer separately.
How do you prepare for such a transition? Which dependencies need to be considered? And how can an operations team tell whether the new entry point is ready for production?
Why the edge is an infrastructure decision of its own
Several technical decisions sit between a user and an application. A domain name must be resolved, a connection accepted, encrypted traffic processed, and a request forwarded to a suitable backend.
These tasks shape application availability. Selecting edge infrastructure therefore belongs in architecture planning.
The questions are similar to those involved in selecting a compute provider:
- Who operates the infrastructure?
- Which functions does the application actually need?
- How can configurations and changes be traced?
- Who handles operations and responds to incidents?
- What might a subsequent provider change look like?
The ayedo Edge Cloud combines Anycast DNS, Layer 4 and Layer 7 load balancing, TLS termination, a web application firewall, DDoS protection, and health checks. It is built on our own network infrastructure, our own autonomous system, and multiple active points of presence in Germany.
We describe this infrastructure in more detail in “DNS, Routing, Anycast: Why Europe's Cloud Also Needs Its Own Edge”. Migration adds another question: how does this become a reliable entry point for your application?
Start by identifying the functions actually in use
A Cloudflare alternative should be evaluated against the functions an organization uses in production.
The first step is therefore an inventory. Which domains are managed? Which applications sit behind a proxy? Which rules affect requests and responses? Where are certificates provisioned?
It is worth examining configurations that have grown over the years. A redirect for an older domain, an exception for a business partner, or special handling of individual API paths may still be relevant.
For planning, we recommend an overview like this:
| Area | Clarify before switching |
|---|---|
| DNS | Zones, records, delegations, and automated changes |
| TLS | Certificates, renewal, and encryption to the backend |
| Routing | Hostnames, paths, redirects, and backend mappings |
| Security rules | WAF rules, exceptions, and access requirements |
| Availability | Health checks, backend pools, and failover behavior |
| Additional services | For example, caching, edge code, or identity-based access, where used |
| Operations | Logs, alerting, permissions, and responsibilities |
This inventory provides the basis for comparison with the target platform. Every required function gets a specific implementation and a test. Additional services need a separate assessment of whether they should be transferred, implemented differently, or retained initially.
This turns a general search for an alternative into an architecture decision that can be evaluated.
The application can remain on its existing infrastructure
A key principle of the ayedo Edge Cloud is the separation of edge and compute.
The application behind the edge can run in the ayedo Cloud, on your own infrastructure, or with another provider. This requires an agreed and supported connection between the edge and the backend.
For an organization, this means the decision about public access can be made independently of moving the application.
An existing Kubernetes cluster, for example, stays with its current operator. The edge handles public ingress and forwards requests to the intended cluster endpoints. Responsibilities for workloads, services, and application routing remain within the cluster.
This division makes the scope of a migration more manageable. The team can initially focus on application access. A later change of compute infrastructure becomes a separate decision.
We also explain this separation for Kubernetes in “Using the Edge Cloud with Your Own Kubernetes Clusters”.
Plan the backend connection deliberately
Entering a target address is not enough to establish the connection.
Before migration, clarify which backend endpoints will be used, how they can be reached from the edge, and which network rules apply. Equally important is determining which information the application needs about an incoming connection.
Examples include the original client address, the requested hostname, and whether a user accessed the application over HTTPS. Forwarding and interpreting this information must be coordinated between the edge and downstream components.
TLS processing also belongs in this plan: where is the public connection terminated? How is the subsequent connection to the backend secured? Which certificates and trust relationships are required there?
Our recommendation is to document this connection path fully before implementation. Network, platform, and application teams can then check their respective settings against the same target architecture.
Test the new entry layer before switching traffic
The new configuration should be checked before regular user traffic is moved to it.
We recommend an agreed test access method. It must reproduce the relevant properties of the eventual production connection, especially hostnames, TLS behavior, routing, and security rules. An arbitrary test hostname alone does not automatically meet these requirements.
Testing should cover complete application workflows:
- Signing in and out
- Accessing protected areas
- API requests and webhooks
- File uploads and downloads
- Redirects
- Persistent connections, where used
- Error responses and behavior when backends are unavailable
The required cases depend on the application. Sign-in may be critical for a customer portal. API requests and incoming webhooks may be the priority for an integration platform.
For each case, define the expected result and who grants approval.
Align health checks with the application
The edge needs to determine which backends should be available for new requests.
We recommend defining health checks together with the application team. A successful connection to a port tests a different property from an application-specific readiness check.
The appropriate depth depends on the service. The check should determine whether a backend can perform its intended task. It also needs defined behavior during brief disruptions and planned maintenance.
Before approval, teams should therefore test failure scenarios: a backend is taken out of service, a service returns errors, or the connection to the target environment fails.
Acceptance testing should show how the configured forwarding behaves in these situations and which notifications the operations team receives.
Plan DNS changes and traffic changes separately
In an edge migration, teams should distinguish between two changes:
Changing the authoritative DNS provider affects the management and serving of the DNS zone.
Changing application access affects the destination of public requests.
Which changes are needed together depends on the existing integration and target architecture. They should be described separately in the migration plan so their effects remain clear.
DNS caches are part of the schedule. A record's TTL affects how long an answer can be cached. A shorter TTL does not retroactively invalidate answers already cached. Preparation therefore needs to start early enough. (Source: Cloudflare documentation on DNS TTL)
If nameserver delegation also changes, the existing DNSSEC configuration must be considered. Signing, delegation, and the information registered with the registrar require a coordinated transition. Cloudflare's documentation explicitly warns of potential availability problems when nameserver and DNSSEC changes are not coordinated. (Source: Cloudflare documentation on changing nameservers)
Give the cutover clear decision points
For the production cutover, we recommend a procedure with named owners, checkpoints, and criteria for stopping.
Establish the following before starting:
| Question | Required decision |
|---|---|
| Who carries out the change? | Owners for DNS, edge, and backend |
| How is success measured? | Application checks, error rates, and response times |
| When is the change stopped? | Agreed technical and business criteria |
| Who decides on rollback? | Named person or responsible team |
| How long is the observation period? | A period that adequately covers normal operations |
During the transition, teams should observe both access paths. This makes it possible to see where requests are still arriving and whether errors can be isolated to a particular path.
The previous configuration is retired only after agreed acceptance. Until then, it must be clear how changes are kept consistent and which conditions are necessary for rollback.
Direct backend access belongs in acceptance testing
The target architecture also needs to specify the paths through which a backend should be reachable.
If security and access rules are enforced at the edge, backend connectivity must align with them. Existing firewall rules, previously allowed source networks, and direct public access should therefore be checked explicitly.
The specific safeguards are agreed with the network and platform teams. They belong both in preliminary tests and in the final checks after migration.
Acceptance therefore goes beyond a successful request through the new edge. It also checks whether the intended access boundaries have been implemented.
Operations must remain understandable after migration
The cutover marks the start of regular operations on the new entry layer.
This includes understandable logs, suitable alerting, and a clear troubleshooting path. When a user reports an error, the team should be able to determine whether the request reached the edge, how it was processed, and whether the backend returned a response.
Configuration maintenance is equally important. Changes to DNS, routing, or backend pools need a defined process. Permissions should match the responsibilities of the teams involved.
We also recommend incorporating migration results into operating documentation: target architecture, responsibilities, acceptance tests, and recovery procedures. This information will be needed again during the next application change or maintenance window.
The path to the ayedo Edge Cloud starts with requirements
A Cloudflare alternative must be assessed against the requirements of the specific application. These include the necessary functionality, connectivity to existing infrastructure, and an operating model with clear responsibilities.
The ayedo Edge Cloud makes it possible to organize public access in front of applications in different compute environments. Organizations can select their edge infrastructure separately and prepare a transition in manageable steps.
The first step is a joint inventory: which responsibilities does the existing platform handle? Which are required? And how should implementation, migration, and acceptance be organized for the target environment?
Would you like to assess how your applications can connect to the ayedo Edge Cloud? Talk to us about your existing architecture and a suitable migration path.