A Tale of Two Generals

Distributed systems depend upon reliable communications. But just how reliable can they be?

Nowhere in computing is immutability more valuable than in sharing data among machines. But before we can truly understand why, we must first understand the scope of the problem. And there is no better way to do that than with the parable of the two generals.

Client-Generated ID

Database IDs are a popular choice for object identifiers. But is it the best choice? Does this simple technology decision lead to architectural complications and poor user experience?

Creating a new object is just the first step of a user interaction. The next step will be doing something with it. That second step always requires knowing the identity of the object the user just created. An important metric of any application is therefore how long the user will have to wait before performing that second step. What is the latency between creating an object and generating its ID?

How Many Things can a Computer Do at the Same Time?

Every program is made up of individual operations. How can you arrange those operations so that they can slide in next to each other?

In every laptop or cell phone that you can buy today, multiple processors are at work running your programs. So you might think that a 4 core computer can do 4 things at once. But, like so many things, the answer is more complicated than that.

What Blockchain Got Wrong

Blockchain technology has moved from novelty, to notoriety, to utility. What fatal flaw does it carry with it?

I’m a sucker for a good math paper. And Bitcoin: A Peer-to-Peer Electronic Cash System by Satoshi Nakamoto is a really good paper. It is extremely well written and understandable. What it created is undeniably one of the most profound data structures in use today.

And yet, the many blockchains that have emerged based on its principles got one thing horribly wrong. They inherited the worst bug of Bitcoin without benefit of its best feature.

What Blockchain Got Right

Blockchain isn’t all bad. If we do away with proof-of-work, what benefits remain?

As much as current implementations of the blockchain data structure propagate a flawed solution to a non-existent problem, they do offer some benefits. They demonstrate some potentially world-shifting concepts that — if they were extracted from the crippling inefficiencies of proof-of-work — could change the way that organizations interact in the future. We could be witnessing the biggest change in business collaboration since the World Wide Web. There are many concepts that blockchain got right.