Most of the Tink endpoints show a sharp rise in overall latency, beginning on about 27 November. We then get a further sharp rise from about 6 December, which won’t get be flagged up in Insights.
We can see this very clearly on the API.expert heatmap, where we the initial rise on Monday, 28 November and then a further rise on Tuesday, 6 December.
The cause seems to be mostly SSL Handshake Time. This was pretty stable at around just above 150 ms until it suddenly shot up by about 120 ms from 27 November to 1 December when it stabilized before jumping again to about 320 m from 5 December to 6 December.
Tink’s Processing Time also went up by about 30 ms from 6 to 7 December.
If we look at the raw data for the Tink endpoints, we can see that it an huge increase in the number of outliers from 28 November that is causing the SSL Handshake Time to increase.
The decrease in CASC score seen in the week beginning 28 November can be ascribed to a step change in the SSL Handshake Time of about 120 ms.
