API Management Tools: Thoughts on High Availability
It is common for API Management platforms to be powered by multiple components, the most common ones being the API Manager Server, the API Developer Portal Server, the API Gateway Server, the API Analytics Server.
While ideally your organization may want to have highly available components all-around, there could be constraints that prevent it from happening, such as:
- Not enough personnel to manage the additional servers required by HA topology;
- No budget to account for a potential increase in license costs;
- Not enough time to fit (sometimes) a more complex deployment in existing project timelines.
Some organizations however may see these challenges as opportunities for investment. They realize that having a highly available environment may result in a more stable environment that is capable of seamlessly handling failures, while keeping trustiness levels of their users high.
While it may be quite obvious why an organization may be looking into having a highly available API Gateway component, as that is the component that serves the API calls, the need for HA is often overlooked for the other components that compose the API Management platform. Below I list a few good reasons to consider the additional investment towards an all-around highly available API Management platform:
API Manager: In a single instance deployment, we never know what could go wrong, perhaps it has nothing to do with the platform itself, but the infrastructure supporting it, the virtual machine, the network. We also never know WHEN it is going to happen, it could be on a Monday morning when everyone is in the office, or it could be right before a production maintenance window on a Saturday night. Should it be the latter, it may impact the production deployment timeline of an entire project, since you won’t be able to publish the APIs supporting such project in your runtime environment.
API Developer Portal: Would it hurt your brand if a portion of your website is down? If application developers trying to leverage your external APIs can’t connect to your Developer Portal for a number of hours or days? Developers may be very vocal about these things in external forums (remember, users of this platform may not be internal users only). Your API Developer Portal being down may raise questions about the health of your whole API strategy.
API Analytics: An outage of this component will result in metrics being lost. It is very likely that there is a number of business units, owners of API Products, that care about API analytics. They may anticipate analyzing analytics reports at daily basis. Would it be okay if one day the report doesn’t make it into their inboxes? Or if it does make it, would it be okay if it is missing data?
There could be other components in your API Management platform, such as an API Test Server, an API Billing Server, and others. Similar considerations have to be made about each one of them. Using the “What If” approach may quickly help you consider situations that the additional investment towards highly available components may pay for itself.
