So...
I'm a new grad from a basic public uni. Some good experience under my belt (internships). But don't really have a huge network or group of people I can help iterate on my idea. I've built an MVP, showed it to some of the smartest people I know and have gotten a unanimous "This isn't shit, and actually provides value to developers."
But... I'm not sure where to go from here, It's an idea I built out of frustration with work and a way to just simplify my work flow handling distributed systems. I have no clue really what it entails really brining my idea to life. I think it has a good chance of being something that is acquired rather than standing alone. Again, though unsure what to do as it's just me, a fresh new grad, whose basically showing their shiny new toy to everyone I know.
project idea: visual monitoring for distributed systems, combines are your transaction into one place and provides feedback on better traceability for systems.
Looking for some general advice, I honestly just want to see where this goes lol.
On to the project though... I was a developer for a long time, before moving into DevOps/SRE which I've now been doing for even longer so I'm quite aware of the value there is in being able to aggregate data from each part of your system to generate insights and as a way to simplify debugging issues. Over the years though there have been many projects that have tried to provide this, with many differing strategies for their implementation. One of the most popular methods was implementing the ELK stack:
ElasticSearch - Provides the data store for the logs from across the system, along with fulltext searching to query that data
Logstash - Runs alongside each part of your system, ingesting the logs, annotating them if needed, and then pushing them to the ElasticSearch datastore
Kibana - Provides a way to search, sort, and display the data from the store.
More recently, within kubernetes you're more likely to see a setup of Prometheus and grafana.
Then you have tools like sentry, which focuses on centralising any error messages from across your system but then also enriches that data by storing context about the environment the error occurred in, such as the stack trace, the version of the code, the version of the dependencies, etc.
Then there are tools less aimed at developers. If you're an organisation that has an obscene amount of money to splunk up the wall, then you're gonna go with... splunk :P
What I'm trying to get at is, to give your product the best chance of being adopted by users, think about what you're offering that either other tools just don't offer, or that you're able to offer in a better, or more simple, way. This really doens't mean trying to implement some of the features of these over systems, but set yourself apart from them by tailoring your offering to a specific use case, or by making it much more simple to setup and maintain. A good example of this is Dozzle, it doesn't have all the features of these other systems, but still, when I'm developing microservices on my homelab setup it's still one of the first services I set up, and this is because when it comes to installing/configuring it, I only have to type 1 command into my terminal and less than 30 seconds later I have a service that pulls all the logs from every container that gets run on that system and a simple web app to display and filter all this data.
Anyway, this has become a much longer post than I had ever intended it to be. If you ever want someone to bounce ideas off of, or if you're up for showing off your toy and giving me a demo of your MVP, you can find my details at https://devopsbrett.github.io/
reply