SSLboard
TLS Health and Hygiene
After more than a year trying to sell a subscription model, I switched to a transaction model. People can pay for a one-off audit, instead of subscribing at a monthly cost. People are done with SaaS. It's not that AI killed it, rather it's the ubiquitous subscription model that did.
Now at least I'm getting much more action, there's no registration, no friction.
I'm still not making money, but at least I put my product in front of people.
We all grew tired of paying a subscription for everything. My gym app needs a subscription, seriously?
And now that AI Agents are the future of service shopping, I think it's wise to bank on a transactional model instead of a subscription model.
For a full year, I've made a nice product of SSLBoard. It's geared towards companies who want to implement good TLS Hygiene by auditing and monitoring their TLS certificates, how they're deployed, how good or bad are their configuration.
I kind of think my product is irresistible yet customers don't seem to agree. I know, you've heard stories like that forever, nobody needs it. The problem is that by testing SSLBoard for various organisations, it's obvious that they need this, they're serving obsolete TLS, expired certificates, it's a mess in most companies. So they objectively need this!
I'm trying to figure it out...
She's my partner in crime, but she had never seen the product before. She gave me invaluable comments and support!
But I also discovered a big showstopper on the foundation of SSLBoard. We talked about it for hours this morning, and she helped me a great deal to discover the solution.
What I mean is "don't stay alone". Before today, I would have probably just bought some chocolate or beer and procrastinated for 6 months before dropping the project altogether.
Sometimes you just need to talk about your problems. Really.
Coming from me, believe me, it's an epiphany.
v1.1.9
It may not sound like an achievement but I was stuck.
When I started working on sslboard, I decided to just use plain node.js and nunjucks templates for the web app. No Single Page application, no React, no build step, KISS.
Except it broke my desire to move forward. I was reluctant to code new screens just because it's so painful and boring with this basic stack.
I knew React would make the whole thing a lot heavier, with a build step, JSX code, components, etc...
I wanted to keep things simple yet more powerful than plain HTML. I didn't want to spaghetti-code JS event handlers.
-- Enter Vue.js --
I think it's perfect for my use case, because it allows a progressive approach: start small, one page at a time, just bring life to those HTML pages.
That's what I needed. I'm super motivated, I'm sure it will bring a lot of motivation.
I always find it hard to say "okay, it's version 1.0.0". As long as it's less than 1, the version number isn't a problem, but there's a time when you must tell yourself "let's do this".
So yeah, I've labelled v1.0.0 one service of sslboard today after much refactoring, cleaning up, etc! I've deployed it on my production k8s cluster so it can do its continuous book keeping, and I'll forget it a little.
It's the core of sslboard, so I've invested quite a lot of time on it.
Do you also struggle with version numbers?
There's actually a "database" of all the certificates in existence, called Certificate Transparency. Except it's distributed as 25+ databases, each huge and quite slow, and you have to scrape them.
I did manage to get all of that content, across a week of scraping with 5 servers running day and night!
Now I keep my database up to date with the same system, only scaled down.
I used Kubernetes (DigitalOcean) for the scaling aspect. Database only contains currently valid certificates, deduplicated: that's about 260 million certificates, 60GiB worth of database (mongodb).
I've also stored the 260 million certificates on DO's version of S3 (and I'm fighting an overcharge for it!)
The resulting database is the starting point for sslboard: it's going to be a data company!
Comment