8 min read

Learning Objectives

By the end of this module, you will be able to:

  • Map the IONOS CLOUD product families to the architectural role each one plays
  • Explain why the Virtual Data Center (VDC) is the organising boundary and which decisions it fixes at creation
  • Distinguish the control surfaces and identify the environment that sits beside the VDC model
  • Frame the FinCorp scenario that the rest of this course resolves, decision by decision

Unit 1.1: The IONOS CLOUD Platform Map

Introduction

Before you choose a load balancer, a compute class or a replication mode, you need a frame: which products exist, how they relate, and which boundary holds them together. This unit draws that frame once. Every later module fills in one part of it.

IONOS CLOUD offers both Infrastructure as a Service (IaaS) and Platform as a Service (PaaS). The portfolio covers virtual and dedicated compute, block and object storage, and software-defined networking, alongside fully managed services for Kubernetes, databases and Artificial Intelligence (AI) workloads. Resources are provisioned on demand and billed on the consumption of compute, storage and network capacity as defined in the current price list. That on-demand model matters for design: almost every boundary you meet later is a consequence of where a resource lives and who controls it.

1. The Organising Boundary and Its Control Surfaces

1.1 The Virtual Data Center

A VDC is a repository for all infrastructure resources you order. As in a physical data center, you reach the resources in a VDC only through a corresponding network or internet connection. Within a VDC, the platform lets you distribute resources across availability zones. For compute, the zone choices are Zone 1, Zone 2 and Auto (default).

The VDC is more than a folder. It fixes the location of everything inside it, and that choice is permanent: you specify the location of a data center only when you first provision it, and you cannot change it afterwards. Several artifacts are bound to the region as well. Reserved IPv4 addresses, snapshots and uploaded private images are region-bound, so a design that assumes you can carry them to another location is wrong from the start. Treat the VDC location as a placement decision that you make before workloads land (Unit 1.4 turns this into a compliance filter).

Two further properties shape multi-VDC designs. A billing contract owner can transfer the ownership of a VDC, with all related rights and responsibilities, to another customer with a billing contract; the request goes to the 24/7 Enterprise Level Support. And VDCs in the same contract and the same region can be connected over private LANs with Cross Connect. Cross Connect does not join VDCs across regions (Module 3 covers hybrid and cross-VDC connectivity).

1.2 The Control Surfaces

The platform provides two interfaces to control resources: the Data Center Designer (DCD) and the Cloud API.

The DCD is a web application that runs in a modern browser. You drag virtual elements from the Palette into the Workspace, connect them, and configure them in the Inspector. Through the DCD you create, edit and delete VDCs, run the lifecycle of virtual servers and their storage, handle snapshots and private images, reserve static public IP addresses, build private and public LANs including firewall setups, manage SSH keys, and manage integrated services such as IONOS Cloud Object Storage, Managed Kubernetes and Backup Service.

One DCD behaviour deserves an architect's attention early. Your changes take effect when you provision them, and provisioning allocates billable resources and cannot be undone. The canvas is a design surface; the provision action is the commit.

The Cloud API gives full programmatic control over the same resources, so that you can automate provisioning, configuration and management. This course shows short API or CLI snippets only where they carry a design point. The automation treatment belongs to the Developer course.

One environment sits beside the VDC model. Private Cloud, the dedicated VMware offering, gives you access to the vSphere Client and NSX-T Manager, and you configure its virtual data center there rather than on the DCD canvas. A Private Cloud cluster includes a minimum of 3 hosts. Units 4.4 and 7.4 treat it in depth.

2. The Product Families

The following table groups the products by family and names the module where this course designs or builds them.

Family Products Role in the architecture Course module
Compute Compute Engine (Dedicated Core Server, vCPU Server, Confidential VM, Cloud GPU VM), Cubes, VM Auto Scaling The stateless and stateful tiers that run your workloads Module 4
Networking VDC Networking (LANs, IP addresses, IP failover), Managed Application Load Balancer, Managed Network Load Balancer, Managed NAT Gateway, VPN Gateway, Cross Connect, Cloud DNS, CDN Segmentation, traffic distribution, egress, hybrid links, name resolution Module 3
Network security Network Security Groups, NIC-based Firewall, DDoS Protect, Flow Logs Filtering on server NICs, attack mitigation, traffic evidence Module 3
Storage and protection Block Storage, Network File Storage, IONOS Cloud Object Storage, Images & Snapshots, Backup Service Disks, shared files, objects and archives, VM images, backup Modules 4 and 5
Data services DBaaS for PostgreSQL, DBaaS for MariaDB, MongoDB, DBaaS for In-Memory DB, Event Streams for Apache Kafka The managed data tier: relational, document, cache, streaming Module 5
Containers Managed Kubernetes, Private Container Registry Container orchestration and artifact storage Module 6
AI AI Model Hub Managed inference over an OpenAI-compatible API Module 6
Governance and operations Identity and Access Management, Activity Logs, Logging Service, Monitoring Service, Tracing Service, Certificate Manager, Cloud Savings Plans Identity, audit, telemetry, certificates, commitment pricing Modules 2 and 7
Dedicated VMware Private Cloud Single-tenant VMware estate beside the VDC model Modules 4 and 7

As shown above, the families map to layers, not to a flat catalogue. Two relationships recur through the course. First, managed services carry their own pricing logic: Managed Kubernetes is free of charge, and you pay only for the underlying infrastructure it uses. Second, some services are reached over an endpoint rather than placed on a LAN. The AI Model Hub is a serverless inference platform whose API endpoints are reachable over the public internet, while data processing and inference take place in Germany. Where a service lives relative to your LANs decides how you secure and route to it.

2.1 The Reference Picture

The following text diagram is the platform map that later modules fill in. Unit 1.2 turns it into the canonical layered shape.

   Clients / on-premises / other VDCs
        |                 |
   Cloud DNS, CDN    VPN Gateway, Cross Connect
        |                 |
 +-------------------- Virtual Data Center (one location) --------------------+
 |  Public edge:  Managed Application Load Balancer, Managed NAT Gateway      |
 |  App tier:     Compute Engine, Cubes, VM Auto Scaling,                     |
 |                Managed Kubernetes node pools                               |
 |  Data tier:    Managed Network Load Balancer (private), DBaaS engines,     |
 |                Event Streams for Apache Kafka, Network File Storage        |
 +----------------------------------------------------------------------------+
   Endpoint services: IONOS Cloud Object Storage, AI Model Hub,
                      Private Container Registry
   Governance plane:  IAM, Activity Logs, Logging, Monitoring, Tracing
   Beside the VDC:    Private Cloud (dedicated VMware, own management tools)

2.2 The FinCorp Scenario

FinCorp is a German financial-services firm. It operates under the General Data Protection Regulation (GDPR) and German Federal Office for Information Security (BSI) expectations, runs a large VMware estate that it must migrate, and is building an AI capability. Every unit in this course takes one slice of FinCorp's architecture and decides it.

On this map, FinCorp's first decisions are already visible. Its VDCs must sit in a German location, because the location cannot change after provisioning and the region-bound artifacts follow it. Its VMware estate splits between Private Cloud, for workloads that must keep their VMware operating model, and the VDC compute families, for workloads that can be rebuilt. Its AI capability starts from the AI Model Hub, whose processing location is part of the compliance argument (Module 6).

Decision Summary

Decision What it fixes When you make it Where this course decides it
VDC location Region of all resources, reserved IPv4 addresses, snapshots and uploaded images Before the first provisioning; it cannot change later Units 1.4 and 2.1
Number of VDCs Network and environment boundaries inside a contract At design time; join them with Cross Connect within one region and contract Units 2.1 and 3.6
Control surface DCD for design and builds; Cloud API for automation Per task Every build unit
Private Cloud or VDC compute VMware operating model versus rebuilt workloads Per workload, during migration planning Units 4.4 and 7.4
Endpoint service or LAN service How you secure and route to the service Per service Units 1.2 and 3.x

Summary

IONOS CLOUD is organised around the VDC. It fixes location, holds the LAN-attached resources, and is built in the DCD or through the Cloud API. Product families map to architectural layers. Endpoint services and the governance plane sit around the VDC, and Private Cloud runs beside it with its own management tools. FinCorp's location, migration and AI decisions all start from this map.

Key Points:

  • The VDC is the organising boundary, and its location cannot change after the first provisioning.
  • Reserved IPv4 addresses, snapshots and uploaded private images are region-bound.
  • In the DCD, provisioning allocates billable resources and cannot be undone; treat it as the commit step.
  • Some services attach to your LANs and others are reached over endpoints; that difference drives security and routing.
  • Private Cloud is managed through the vSphere Client and NSX-T Manager, not on the DCD canvas.