It’s 3 a.m. You’re dead asleep. The phone rings. You blunder around trying to locate your phone.
Oh thank God – it’s work! Only work. No one is dead.
It’s the support desk. Your most important customer’s service has gone down.
It might be the middle of the night where you are, but not where your customer is. If their service is down, it means they’re losing revenue – and if they’re pointing the finger at you, your company is going to be in a whole heap of pain.
“It’s OK,” says the DevOps guy a couple of minutes later, sounding irritatingly chipper for the dead of night. “It’s not us. I’m looking at the logs and everything is 200 OK. We haven’t seen anything other than a 200 for weeks.”
“So the API is working?” you ask.
“Absolutely. Completely nailed.”
So you call your customer. The problem’s not with us, you say, it’s with you. Our APIs working fine.
“No,” says the customer, “it’s not with us. It is with you. Your API is supposed to return 14 language options. It’s only returning 4. Which is pretty totally useless to all our customers who need those other 10 language options because they are the languages they need to use our product.”
You go back to DevOps – any way of seeing what payload the API is returning?
“At three-fifteen in the morning?” says DevOps. “It’s not a public website. We can’t see what they are seeing. It’s just a private page on their corporate intranet that makes a call to our API.”
“Couldn’t you put a trace on an API call or something and capture the payload that way?”
“A trace? This isn’t CSI, you know. You’d need to write a cURL script or something. But I’m the DevOps night shift guy. Our cURL expert’ll be at nine.”
After a sleepless six hours, you are back on the phone to DevOps. After she does all the other urgent things she has to do (what could be more urgent than your customer’s problem?), you finally get to speak to her.
“Well, yeah, guess I could write a script. Might take a few minutes. I know that API. Got a really weird authentication.”
After another couple of anxious hours, at last you get a call from DevOps cURL expert.
“Yeah, the payload only has 4 language options.”
“But it’s supposed to have 14! How could the list of languages get shortened?”
“We get that from an API. Will look into that.”
“How long has this been going on?”
“No way of telling. The API’s always up. Solid as a rock that API.”
“It’s not solid as a rock if it’s not returning what the customer needs.”
“Well, at least they’re always getting something back.”
“Just not what we promise.”
“Of course, if we monitored the API, we’d know if something went wrong. Or at least be able to see that it had.”
“There’s no way to monitor API?!”
All the other APIs that all your other customers are using. More sleepless nights. Your life flashes before your very eyes.
~Aaaaaaaaand scene.~
Look; we could give you our sales pitch. We could show you stats and graphs and charts until your eyes cross. But we’ve heard real-world stories just like this one from people who have since become our clients – because there IS a way to monitor API. We do it for companies every minute of every day. We can do it for you, too – so you can literally rest easy.
