Money, Hackers And Lawyers

How far can you afford to be from your data, and what does that have to do with power and performance?


When magnetic tape backup was introduced in the early 1950s, it was stored in vaults. Access was slow. Someone actually had to go get the tape, load it onto a mainframe computer, and start the spool rolling. But at least you could be sure that, minus a well-planned break-in or cloak-and-dagger theft, it would remain private.

The latest push into cloud-based storage has altered the risk equation substantially, and along with it power and performance guarantees. By how much, in any of these respective areas, no one really knows. While data mining tools are effective at identifying patterns in coding or key words, phrases—spoken, typed, or in documents or videos—there is much more that is left behind. And nearly all of it is vulnerable to attack through one channel or another, given the spaghetti-like virtual access routes that have cropped up to guarantee access and improve performance.

Even the term cloud is hard to define. There are private servers, shared servers, fog servers (low-lying cloud servers that are closer to the network), on-site backup, and more options ahead for on-chip storage so that less has to be sent to the cloud or local servers. All of it is linked together somehow, but who or what has access to what, and through what channel, may vary greatly by activity surrounding that data.

The basic rule of thumb is that if you’re willing to pay more in terms of energy, battery, processor speed and real estate costs (on-chip or on-site), you will have faster access to your data. Security in that case would be a function of various levels of software, Internet access, and how good the locks are on the doors—physical and virtual. Nothing is guaranteed, of course. Given the steady stream of data breaches, even on-site storage isn’t that secure. It’s gotten to the point, in fact, that mainstream media doesn’t even report most breaches anymore.

But the balance tips even further with the Internet of Things, wearables, ingestibles, and the quest for faster access to vast amounts of data. Better speed and connectivity is convenient, but it carries a price. If the data is useless to anyone else, that really doesn’t matter. If it provides a connection into more sensitive data, it does. And securing that sensitive data carries a price, too—once again in terms of energy, processor speed and real estate—and the effort it takes to lock down that data in the first place and keep it locked down later on.

Storage typically has been a separate part of the computing world, aside from cache and memory, which are built into chips. But increasingly, it is an integral part. It affects speed, cost, and it provides a trail that can be used by lawyers at any point in the future for all sorts of purposes. And the more data that is available, the greater the risk on all of these fronts and the higher the cost that people are willing to pay to access their data quickly and safely and to secure it when it’s not in use.

This may sound ominous, but every threat has a silver lining. For the semiconductor industry, this presents a whole new market. Locks don’t prevent break-ins, but there are an lot of them sold each year. Architecting chips that control access to data efficiently and quickly, wherever it is stored, is a huge opportunity, and with the IoT it’s about to get a lot bigger.