Monolith: EXPOSED

It bears mentioning that the purpose of this blog is actually to archive information about projects I do. You like it? FAB! But seriously, I tend to forget everything I do as soon as I'm done. This way I don't have to try to remember things on the spot.

And now, the long story of the Monolith, WFYI's Broadcast Archive Portal.

Chapter One. Nosy Beginnings.

Several years ago, a thing started bugging me (which is how most of my adventures start): We've got these CRAZY big disk arrays all arranged to take video and send it out over the air. Everyone was on and on about convergence and file-based whatever. These buzzy terms were as popular and meaningless as synergy was for general business.

The bugging was from this: if you wanted to see any of the programming stored in these servers, you had to wait and watch it on television when it aired. It wasn't quite that bad, but it was still bad. There were two locations in the building where you really could dial something up, and it was a lot like working a VCR.

I wanted to see what was in those things, and a mild obsession began.

There were some fantastic false starts and total failures, but the fun part began when I finally discovered that the things supported FTP. They even had their own weird implementation built around moving video. I was eventually able to pick a show, pull a copy from the servers, and then watch it at my leisure. No need to go anywhere special, and no big console buttons to push.

And I was happy.

The obsession didn't end, however. With that initial success, I wanted to be able to let everyone watch whatever they wanted with a minimum of screwing around.

First up was the file format: GrassValley's proprietary GXF. Great when used inside GrassValley's K2 system, but awful anywhere else. Really awful.

The K2 was built wishing only for speed. It had to guarantee the ability to take a video signal, turn it into an MPEG2 file, and get it written to disk without losing a single frame. You kids today don't know how impossible that used to be. So it was good and fast, but not even remotely efficient. Doing all that work in real time was only possible by using very fast storage to collect stupid amounts of data.