Customer metrics availability

Overengineering or not? #

Often enough I share an idea that all new serverless platforms as well as simple cloud providers like Digital Ocean do not support customer metrics. All the time I hear it’s not needed for non-enterprise customers and I ask for the feature due my overengineering mindset.

But today I saw a blogpost in my RSS reader from Internal Tech Emails:

From: Bill Gates Sent: Thursday, October 19, 2000 3:02 PM To: Will Poole Cc: Jim Allchin; Eric Rudder Subject: Quality experience and streaming media

I don’t know how unique I am but my experience in clicking on streaming media stuff and seeing if it works is still less than 50%.

For example the MSN home page talk about a golf match going on called the President’s cup. I click on that and then I click on this link:

http://stream1.[…].com/[…]/101700192418001.rm

I get a message saying: “This document is not a RealAudio or RealVideo Document” “This is error 11. To learn more go to a real URL.

This is so typical. I always get error messages when I click on Audio and Video.

Just for fun I go to the home page and grab an MSNBC video clip:

<http://go.msn.com/CM/10002/default.asp?target=http://www.msnbc.com/[…]>

I click and bizarrely I get a browser page and a video page - the lack of integration of the browser and the player is crazy.

The browser page is fine but the video says buffering and 5 minutes later nothing has happened.

We should go on a quality campaign. We should really collect error message reports from our servers.

Something is very hard to set up or hard to track in this area. The quality is giving us a bad name.

I guess I can keep sending all my failures along - the key point is that I haven’t seen any improvement.

I have yet to have a good video experience.

Where Bill Gates asks for a solution to collect logs from different servers to monitor them. It was ~24 years ago, Microsoft that is already big company with many customers and many servers, and they still don’t have a solution for this.

The CEO experiences an issue with a production product, but there is not simple solution to look at metrics/logs/etc. to identify the reason of the issue. I’m sure it’s not the case in current Microsoft’s infrastructure.

Moreover, there is no such cases even in small per projects or startups that just deployed a single serverless function to any serverless solution. But why?

Available solutions #

The main reason we have logs in any cloud provider is believement in this feature as essential part of any system. In current environment we just expect to have it and the most of us will be surprised to find out that it’s not available.

But what about metrics?

Suprisingly, the most providers provide system metrics along with logs. We expect to have some high-level system metrics along with logs for every component of our system. But, there is no such case for custom metrics. The most of the providers do not provide a buildin solution for custom metrics in a service.

We can try to explain it as an expensive solution, or not popular request from customers. However, in my opinion, it’s because there is no the first provider who started to provide it.

It’s only matter of time after the first big player who will provide it, other would copy and standardize the solution. I’m waiting for the first serverless provider (Supabase? Firebase? Vercel?) to introduce it as a free (or cheap) additional service and it will be spread to all small and big cloud proviers at no time.

Conclusion #

Why have I decided to write down my thoughts?

I hope to engage someone to add this small, but really important feature to their platform. I believe it will be a killer feature for the platform (at least for short period of time while it’s uniq their feature) as well as a great investment to our bright future of cloud services.