IBM API Connect topology

Depending on what you want to use your API Connect cloud for, consider the topology that you want to implement.

The minimum server configuration that you can have in API Connect is one Management server and one Gateway server. However, this configuration does not provide any high availability.

By configuring two Management servers and two Gateway servers, your cloud benefits from high availability (HA). When you define more than one Management server, configure an external load balancer. Then, if one server is removed for planned maintenance or an unplanned failure, your users can continue to access the cloud. The more servers that you add the greater the ability of your cloud to withstand outages and continue to provide processing capability.

Consider whether you want to define a separate environment, within your cloud, for development work, functional testing, or staging. Separating your cloud usage can help to prevent problems when you are testing new APIs before you make them available to application developers.

Although you can populate your cloud with only virtual servers, you might want to use physical DataPower® appliances for your production Gateway servers. The physical Gateway servers provide improved performance throughput when compared with virtual Gateway servers.

Before you decide how many Management servers to add to your API Connect cloud, consider how the data is handled. Management servers control two different types of data:
  • Configuration (for example, Users, APIs, Products)
  • Analytics (for example, API usage, performance)

Configuration data is tightly controlled. At any time, only one Management server (the Primary) is allowed to write the data. The other servers (Secondary or RSS) maintain a complete local cache for performance and high availability (HA) reasons. Significant performance improvements are achieved by balancing connections across the Management servers.

When planning for high availability, consider the following:

Analytics data is handled differently. All of the data that is written to the Analytics system is replicated asynchronously on the back-end server. When Analytics data is sent from a Gateway server to a Management server, that data is balanced across the available Management servers. If there are two or more Management servers in the cloud, at least two of the Management servers hold each Analytics data record for redundancy purposes. Therefore, in this scenario, a minimum of two copies are kept for each data record.

When you configure a Management server, you can define a second hard disk to store the analytics data that is captured for your cloud. The size of the hard disk can vary from the advised minimum of 350 GB for a production cloud to a maximum of 2 TB. It is important to consider the disk size that you need to hold the analytics data. The second hard disk size can be modified in the settings of the deployed VM after deployment but before it is powered on for the first time.

Restriction: The maximum supported combined size of all disks is 2 TB per Management server.
Restriction: If you define two Management servers that both use a second hard disk, you do not have twice the disk space for the analytics data. With two Management servers, the analytics data is synchronized across the disks to provide high availability.

As part of your topology planning, consider the server configuration that you want to implement. For more information, see Firewall requirements.

Another consideration, is the IP address that you need to configure for the Gateway server.
Important: The actual IP address is determined from the interface definition on the DataPower appliance, and is typically accessible on the Internet. This address is also generated into the front side protocol handler of the API Connect service that mediates the client messages. If this DataPower appliance is part of a clustered configuration, this Ethernet interface is used for configuring self-balancing and the virtual IP address. This interface might be the same as the one used for the XML Management interface. If the server is added to a cluster that uses DataPower self-balancing, a Standby Control configuration is added automatically.