Anatomy of a Datafortress

You're connected to a datafortress, searching for the financial archives of a small corporation hoping to find something to extort them for. But there's nothing here, only terminal, door and camera links. "Whats wrong?" you ask yourself as you patrol the virtual space within the fortress.

A new data wall rises in front of you, it's security clearance higher than before. As you crack it, the virtual sound of an alarm resounds in your ears. "My program didn't trigger it, I'm sure" you wonder as you hurry to pass the wall an reach the other side.

"One LDL?" is the only think you're have time to think while you jump into it, you don't have time to wonder while the bloodhounds run toward you.

The basics

The typical thing a netrunner do is stealing data from a datafortress, but which kind of data and how do they find it?

Information is stored on databases, folders and files in the system. Those are typically represented as memory blocks which represent local data. But some data is saved on the "cloud", basically an external system usually maintained by another corp.

To access external data you have to enter a protected LDL icon (with a standard Code Gate opener), which will send you to another datafortress with new security measures. Take into account that cloud services boast about their security measures, so expect extreme brute force if they detect your illicit access to their system.

What's in a typical Cloud Memory Unit (CMU)? Boring stuff mainly. Work databases, operations documentation, personal info, provider/client databases, personnel calendars, etc.

What's in a local memory Unit (MU)? The same stuff as a CMU (if the datafortress doesn't use cloud services) and meaningless stuff like power point presentations, internal messages, local info or images ... and the good stuff, the one that's so black that you wouldn't entrust external corporations to manage it.

Bigger is better

The previous schematic is for a single shard server. The one you would find on small enterprises, lone buildings or similar.

Single shard? You must be asking yourself. Well ... when things grow up, a single server isn't enough to control everything. Nor is it even desirable. If your factory machines aren't working as expected, you don't want to reboot security cameras, doors or international operations.

So you connect multiple servers (called shards) and divide the workload between them. Now, if your factory is having problems with the machinery, you don't need to unplug the security systems.

And now every shard has it's own defenses.

Yes, that's more work for you. Aren't you happy?

What does this mean? It basically means that now you have to defeat various data walls to reach the destination you want. You first break into the login/net server which only contains minimal operational MUs and then search for a shard index and try to breach it's security.

You want access to cameras and doors? That's a shard.
You want access to machinery? That's another shard.
You want access to terminals and accounting MUs? Guess it. Yes, another shard.

Every one of them has it's own data walls, code gates and probably, a sysop.

Additionally, some shard are only accessible once you have connected to a specific shard. For example, if you want access to the black ops file, you will need to hack into the initial login shard, then the management shard and then to the internal security shard.

Who runs it?

A lot of datafortress have enough CPU to be managed by AI.

But you won't find a lot of them managing sites anyway. Why? Because AI are predictable, reprogrammable and exploitable. They do not learn or react as a humans (you don't really want an AGI wandering freely) and they do not spawn spontaneously ... which mean you must pay someone to create them.

And when someone find a way to exploit one, the news spread fast between the netrunners hubs.

It's easier, cheaper and more reliable to use human sysops to monitor the system. True, those can be bargained, extorted or even bribed ... but don't expect it. And if the sysop is fried, you don't have to pay his salary.

If a sysop is working locally, they will use the mainframe as their terminal to administrate the datafortress. Which means multiple actions and amazingly fast programs. Also, it means you know exactly where they are on the building and someone can "unplug" them in the real world.

If not, they will use their own cyberdecks and work as a netrunner would.

And that's all.

Try not to die the next time you assault a multiple shard datafortress with multiple cloud services and a mix of black ice ;)