Monday, November 24, 2014

Ubiquity of data

I've been thinking a lot lately about ubiquity of data, or really the lack of it in today's modern technology.

In the 90's and 2000's most of our information lived on our primary machines, whether that was a desktop or a laptop. As an industry we spent a lot of time and resources trying to make that information portable. In the 80's it was the Floppy drive. In the early/mid 90's it was the Iomega Zip Drive. In the late 90's/early 2000's it was the recordable compact disk. In the mid to late 2000's it was flash memory in the form of a USB stick. All of these technologies focused on one thing, making it easier to move data from one place to another.

In the late 2000's/early 201x's we started to talk about shifting our data to the cloud. The thought was that if we put our data in services like Amazon S3, Amazon Cloud Drive, Dropbox, Box, Microsoft One Drive, and etc. that our information would be ubiquitous. In a way we were right in that we can now access that information in the cloud from anywhere. But fundamentally we're still thinking about and interacting with data as something that we move from place to place.

I think as an industry we need to stop thinking about data as a thing that we move from place to place and instead solve the problems that prevent us from accessing our data from anywhere. So what are the problems that we need to solve to make this a reality? This list is by no means exhaustive, but it's where I think we need to start.

  • Federated Identity Management.
  • Data Access Standards.
  • Networks that do not model their business on whether you're a data consumer or provider.

Federated Identity Management

In the world we live in today, each service (i.e. company) owns authenticating who we are. I.e. they keep a proprietary set of information about us that they use to test us with. If we pass the test they considered us authenticated. Most of these tests come in the form of two questions, what's your name and what's your password.

The problem with this is that it takes identity authentication out of the hands of those being identified and puts that into the hands of those wanting to authenticate. There's nothing inherently wrong with wanting/needing third party validation. The problem comes when we have hundreds of places we need to authenticate with, each with it's own proprietary method of authentication. Not to mention that it passes the buck to the user to remember how each one of these services authenticates them.

Tim Bray has a good discussion on federation that you should read if you're interested in the deeper discussion of the problems of identity federation.

Data Access Standards

We need data access standards that any group (for-profit or not) or individual can implement on top of their data that allows any other system (using the federated identity management) to interact with it. These standards would define CRUD operations (create, retrieve, update, and delete) in such a way that any other system and interact with the data on that system on the users behalf.

We have a good start to this with standards like OPML, RSS, WebDAV, CalDAV, CardDAV, and etc but these standards aren't cohesive. On top of that we don't have a real way to query a service to see what type of CRUD operations it supports. If we had the ability for the service to state what it serves then the clients could more intelligently interact with that service. Currently we put the onus on the user to know what a service offers.

Networks that do not model their business on whether you're a data consumer or provider

Right now the people who provide us access to the internet think about us in two categories. The first category I'll call "data consumers" and the second category I'll call "data providers".

Data consumers have the ability to get things from the internet and put things somewhere else on the internet. But data consumers don't have the ability to provide things to the internet without putting it somewhere else.  A good example of this is email. A customer with a standard "data consumer" internet connection cannot run a mail server for two reasons.

First, they get a dynamic IP address from their their ISP (internet service provider). This means that the address from which they connect to the internet is always changing. Think about this analogy to a dynamic IP address. What if your home address was constantly changing either daily, weekly, or monthly. It would be impossible for anyone to contact you via the mail reliably because anytime your address changed mail sent to the previous address would be delivered to the wrong house. It's the same way on the internet. If you want people to be able to talk to you you need to have a static address for them to contact you.

Second, ISPs block the ports necessary for others to talk to you. Even if you had a static address, often your ISP blocks standard email ports (25, 993, 143, 587, and 465) because they're trying to stop spammers from easily distributing their spam. But as anyone with an email address knows, the spammers are doing just fine even with the ISPs not allowing incoming connections. So I don't buy this as a valid reason to block these ports.

Data providers have all the same access as data consumers except they pay more to have static IP addresses and to not have the ports blocked. Notice anything wrong with this situation? The ability to fully participate in the internet is based on how much you pay your ISP. ISPs hide behind the fallacy that they're trying to protect you in order to be able to charge you more for the ability to truly participate on the internet. Does that extra money you pay actually protect you or anyone else on the internet better? No. Most ISPs will probably tell you that your also paying for more reliability. But you're running on the same system as the data consumers, so I don't buy that argument either.

I truly believe that we're not quite moving in the right direction when it comes to solving these problems. Until we do, you will constantly be battling moving your data from one place to the next when any new interesting service comes into existence.

No comments:

Post a Comment