Monitoring A La Borg

Imagine there’s no console. It’s easy if you try. No central server, no admin tty.

Imagine all the endpoints, living all the same …

I have this vision for the future of computing: where endpoints provide different views, perhaps based on their form factors, but they otherwise access the same data, so they’re interchangeable. Today, your mobile device is still Your Device. It’s customized the way you like it, with all of the settings and applications you prefer, and with Your Data on it.

Lending it out to someone else feels kind of icky, like lending your toothbrush. And losing it is a calamity, even if you have a backup somewhere, because once you buy a factory-new device, you have to spend time breaking it in again and putting everything back on it the way it was before.

I imagine a future where a family just has a pile of devices on the coffee table, and everyone grabs one as they need it. Or maybe there’s one in the kitchen, one in each bedroom, and a couple by the front door to grab on your way out. It wouldn’t matter which one you picked up; once you authenticated to it, the device would show you your last session and be able to access all your own data. Remember when families all shared one or two TVs and corded phones? Wasn’t that quaint?

The cloud promises this sort of interchangeability and dynamic provisioning, but only to a certain level. Vendors are working on providing roaming profiles for handheld tablets (for example, for health-care workers), but that’s for a limited set of software. Desktop virtualization is for, well, desktops. And the endpoint experience can still be completely different for the same functionality, depending on what you’re using: Why do I need an app on my mobile device just to consume the same website content that I can view in a browser on my laptop?

The final frontier is administration: We still use a hierarchy for most infrastructure. This has traditionally meant that we needed central servers with which to manage things, or something on the perimeter in a network choke point that controlled some aspect of all traffic. Well, we all know the P-word is going away; that means, in theory, that no one vantage point over the infrastructure is any better than another. It also means that an administrative console – that trusted entity which makes such a juicy target for attackers – can end up in a pretty arbitrary location.

I’ve been seeing the beginnings of peer-to-peer monitoring and administration, where you can query data that has been passed among all the components of a networked system, rather than waiting for a central entity to poll each one. This functionality is going to be vital as our use of cloud computing expands to the scale that is possible, and as we become more mobile in terms of how and where we make contact with our resources. What if you could query any given endpoint and ask, "Hey, how are we all doing today?"

For the record, I always thought it was cheating to have a Borg Queen: talk about your literal deus ex machina. It provided a central weakness that you could use to defeat what was otherwise a resilient, self-healing collective. If you can spin up one VM to replace another, then couldn’t you have VMs that knew everything that all the other ones did? Couldn’t you have VMs that watch one another for changes and alert on them?

We may not be at the point yet where all the components of a scalable infrastructure are interchangeable -- any more than our smartphones are interchangeable today. But making sure that they all have access to the same data, without having a single point of control and therefore a single point of failure, is a powerful start. Collective monitoring could be a way forward.

Wendy Nather is Research Director of the Enterprise Security Practice at the independent analyst firm 451 Research. You can find her on Twitter as @451wendy.