Future.porn Problems and How I Solved Them

Or at least how I'm planning on solving them

This past month has been rough-- a mix of progress, setbacks, and financial chaos. I've made big strides on future.porn, but I've also hit some snags. The biggest one being money.

I've been pouring a lot of cash and energy into improving the site while my income has been trending down. My debt is up, interest payments are up, and my checking account has been in the red. I overdrafted three times in the last 30 days.

In some ways, I'm feeling defeated. I'm overbudget, overdrafted, overlevered. In other ways, I'm feeling strong-- I'm making progress daily and I have more to give.

IPFS Storage Dilemma

Futureporn's IPFS cluster has been down for weeks. The HDD on my budget Virtual Private Server (VPS) filled up completely, and the provider doesn't offer storage upgrades. I tried moving to a higher-end VPS with scalable object storage, but that turned out to cost $150 per day. Yeah, I can't afford that 😆.

IPFS cloud storage is a challenge. I tried the cheapest object storage available, but it ended up being too slow (12 hours to pin a single VOD). Premium object storage is too expensive ($150+ per day). Renting that kind of hardware isn't sustainable right now.

So how do we solve this? Ideally, I'd like run my own datacenter. Self-hosting can have cost savings when compared to the cloud. The problem? I live at home with family that I share Starlink internet with, and I think I'd be using more than my fair share if I self hosted IPFS from home. Someday I'd like to move to a place that has fiber, but it's not something I can do right now.

The solution? A middleground between super cheap object storage and super premium NVMe object storage-- HDD blockstorage attached to a budget VPS at a quality VPS provider. I've got a fresh IPFS kubo instance running and it's slowly working on pinning all the content from the network. Once it's synced up, I will shut down the original to save money.

Zombie Business Economics

Right now, Futureporn isn't profitable. It's basically a zombie-- technically alive, but financially dead.

• Operating costs keep going up.

• My income has been trending down.

• Every solution I try costs more money.

Net result-- not enough cash flow, rising debt, and stress.

How do we solve this? Build a better service that people want to pay for. Personally I think visitors are going to love the vibrator integration I built last month. Getting it working end-to-end is a work in progress!

Build System Bottlenecks

future.porn's build system is the automated infrastructure which takes a VOD and gets it ready for playback on the website. It technically works, but it's way too slow.

• A 6-hour VOD takes 33 hours to process.

• VibeUI (lovense vibrator UI detection) runs 4.9x slower than real-time playback.

• Renting faster servers is out of my budget right now.

How do we solve this? One solution is more powerful hardware, but I can't because of the money problem. I can't shell out big money every month and rent the latest and greatest cloud GPUs, so I have to get creative.

Luckily, future.porn's build system is designed to run on multiple distributed workers, which means I could add my more powerful desktop computer to the swarm and have it run intensive transcode and inference tasks.

Right now I have an Intel i7-4790 (8 cores @ 4GHz), and during development I was seeing it transcode about three times as fast as the cloud CPU I'm renting. I don't have a dedicated GPU right now, but I think I could find a hand-me-down from one of my gamer brothers. A GPU would really speed up the VibeUI inference.

Bandwidth Cost Explosion

People started bulk downloading VODs, like, really bulk downloading. I got a low balance alert from my CDN so I did what I normally do-- add $25 to the account. That usually happens a couple times a month so I didn't think anything of it. Next day, another low balance alert-- the $25 was burned through. "Oh, must be a spike in interest?" I bit the bullet and spent another $25. Next day, same thing, and that's when I realized this was a problem, the type of problem that is speedrunning bankruptsy. In the first three days of October I saw traffic hit 5TB, which is like the entire site's worth of VODs and then some.

I had to take action. I limited connections to 3 per IP address to keep things from spiraling. It sucks to throttle access, but it was either that or watch the project implode.

So What's Next?

Keep pushing! I've got a lot of work to do, cutting costs where I can, and optimizing the code I have to get the most out of it. Increasing the VOD processing speed is essential to getting future.porn off the ground, and I'm so eager to see that happen.

Thank you for your support!