Automatically Operating 100+ Lender APIs

In the loan comparison service, over 100 financial institution APIs are integrated to check users' loan limits and interest rates. Since even a single failure could cause a user to miss out on a good loan opportunity, it was necessary to be able to check in real time whether the APIs were actually working properly.

First, to secure Observability, the successes and failures of financial company API calls were aggregated as metrics. Using a count metric, the number of successes/failures was recorded for each call, and a dashboard was created to see the success rate by financial company at a glance.

However, the dashboard alone was not enough. Since a developer had to keep watching the screen, anomalies could be missed. For example, if the financial company reported an error first, or if a problem was not detected quickly after a network change, the response could be delayed.

So a Monitoring system was added.

• Automatically send an alert when the API success rate falls below a certain threshold
• Treat cases where requests themselves don't occur, such as during financial company server maintenance, as normal
• Include the related logs and a link to the success rate dashboard together in the alert

Now, through Slack alerts, developers can immediately check anomalies, proactively inquire with the financial company, or quickly apply error-handling logic. As a result, the speed and scope of proactively catching problems improved significantly, and the API success rate, which was 90% five months ago, rose to 99%.

To stably operate a rapidly growing service, this kind of automated monitoring and response system is needed. Banksalad operates several such systems, aiming to resolve issues before users even notice them.