QuickAPM - Observability As A Service
A warning: This is a technical product targeted at tech founders and startup developers :)
The Problem -
I've been a DevOps Engineer/SRE for going on 7 years, working in Big Tech and at Series A Startups. A common theme I've noticed with application observability - it's difficult to implement and manage APM solutions (i.e. Datadog, Elastic, etc). These are out-of-reach and WAY too expensive for many indie hackers and small startups. This leads to a lack of transparency and an inability to understand how customers interact with their products. How can we give customers the best possible experience, when we don't understand how they are interacting with our applications? Monitoring endpoint response codes or waiting for poor customer feedback is not enough. This is like a chef only tasting a few of the ingredients of a dish, without understanding how they all come together as a final product being served to customers. It is part of the story, but leaves large gaps in our understanding and could cause customers to not return.
The Solution -
I am kicking around the idea of creating an easy-to-use, opinionated APM platform, built for Indie Hackers and early stage startups. Capturing APM data will be done with libraries implemented as middleware, so there will be no instrumentation outside of the initial setup. The data captured and presented would be based on industry standard metrics. So you don't have to query a large data set, the aggregation will be done for you in the background. Keeping this as cheap as possible is my goal, so I'd like to give users the ability to have as much or as little data as they need. You can have a high level metrics on routes and spans, or keep granular data to dive deeper.
I would love to get some feedback if folks have the time. Is this something you would be interested in? How would you like to see this be implemented? What data do you wish you could have about your product?
TLDR; A cheap and easy-to-use APM tool built for indie hackers and early-stage startups
Didn't newrelic new pricing system have a free generous tier now (after their past crazily unreasonable pricing for the container age), they are now like data size based which IDK how the price comes out at the end but initially sounds like a good offer.
Also there are tons of small SaaS offer that each only does one thing.
And idk if stack driver is usable without gcp nowadays but was interesting offer in the past.
Generally I think opinionated systems are missing in APM, I thought about that being good built on top of a logging system where many constructs can be put in place for different issues. You can see buds of these concepts come in different systems with prebuilds of dashboards/rulesets ect, some in dashboard systems, some in scanners, some as their own systems like in the secops world
I don't think there is much of an actual pain and need for IH most of the time, at best you'd want an uptime+ system. It's the growing startups past 5-10 people that might be the target, but IDK, didn't scan the market recently.
Think from the biz side, what's the pain, how much is the loss, how much of that loss would you save, for what amount of money+effort+sales process.
Newrelic did change their pricing structure lately, there is a relatively generous free tier. But from what I can tell , it's similar to how AWS does it. They rope you in with a free tier only to charge you that much more on the backend. So if you're judicious with the amount of data you're sending/up for setting up new accounts, it is possible to use for free. But you are still responsible for producing dashboards and attempting to make sense of the data.
This would be geared towards teams and IH developers with a reasonable amount of traffic. So you're right, the small startup teams that have generated some MRR, have consistent traffic and need to start thinking about performance and customer retention would be the target.
One of the things that have been holding me back from getting started is the marketing/sales process. It's very niche and hard to validate broadly. I have attempted the, "if you build it they will come" strategy, which obviously resulted in 4 months of development and $5 MRR ha.
Appreciate the feedback, it's easy to get lost in an idea. I can see the pain points clearly, but I deal with this stuff daily. APMs are complicated to implement, the data can be overwhelming and hard to make sense of, they are often expensive and you really have to know what you're doing to extract value of these solutions because they are not opinionated.
Think of after the pain is know vs before, who are you targeting, with what message?
I think I'm too deep in that I'm not sure how people originally get to the need.
I do remember many times it starts with customer complaints and eng. not being able to reproduce... And that usually starts with like exception/error systems...
On ent. Grade it's kinda comes after the system was down multiple times out of working hours and again probably customers complain or someone watching revenue by time.
Yeah I think this pain is typically discovered after you reach a threshold of customers/traffic on your platform. When customers start to have expectations of your product, they begin to notice when things are not working or working poorly. For a startup, at least in my experience, this is when you scramble to find a solution. Startup engineering teams typically try to host it themselves, which is WAY too much overhead. Or just go with a Datadog until they get that $15,000 bill one month.
What projects are you working on? You seem to have knowledge in the space, just curious.
I think we just got a similar background, but I'm not working on anything related. To me these types of projects are too big and hard to sell as a side project. And if you do it as a startups it's going to be pretty ruthless as I don't believe any of this is sticky or can contain any barriers unless you can also work the legal front well somehow.
If you can do the time and sales and keep your headcount small, you can make money here unless the market consolidates with the infrastructure cloud providers.
I think the most money is in the security side but also the most investments and biggest barriers. The most competitive and most potential is the logging. And the common APM metrics is the easiest sell as it's easiest to visualise properly.
Another angle especially opinionated is taking something specific like WordPress and specialise there on improvements recommendations or automatic actions. What's after WordPress maybe just generic static sites nowadays seems to have a renewed life with the newer frameworks, but can also be cost adverse group...
Your point about the size and scope of a project needed to get into the APM market is solid insight. It's crowded, finding a differentiation and competitive barriers will be difficult.
Security is tough, there are not as many common patterns you can work with like in the APM space. And logging is interesting, just a lot of data that would require a bunch of compute on the backend.
Anyways, thank you for taking the time to comment. Would be happy to connect outside of this post, interested to learn more about the side projects you've done and your approach.