Such is the hype around Bitcoin and blockchain technologies that numerous companies have seen their share prices rise simply by changing their names to include one of those words. In their rush to invest in the hot new thing, some investors don't check whether the companies actually have anything to do with blockchain.
The technology is driving a degree of interest in the data center world, too, with the big question being how much power blockchains will end up consuming. As Peter Kelly-Detwiler wrote in a recent post, it's unlikely that power needs will reach the alarming heights predicted by some pundits. But, nobody really knows, yet, how much power we will need.
But what about blockchain's other effects on the data center industry? I spoke to Tate Cantrell, Verne Global's CTO, about what's happening.
Blockchain was designed to tackle the problem of trust in a decentralised network - specifically a digital currency. The problem with digital currencies has always been the issue of double spending - without a centralised authority managing payments, someone can theoretically copy their digital currency and spend it twice.
The blockchain solves that by creating a decentralised ledger. Instead of a central source keeping track of transactions, every node on the network keeps a copy of the ledger. Adding a transaction to the ledger requires a cryptographic hash - effectively a kind of digital fingerprint - that is partly derived from the previous block. Computers on the network race to be the first to verify a block of transactions by finding a hash that satisfies the requirements, a process known as mining.
Once found, the block is added to the ledger, extending the chain. The winning computer shares its 'proof of work' with the network and is rewarded with a token - usually some new portion of the cryptocurrency.
Blockchain's power needs are driven by mining, Tate says. The costs involved in the blockchain are split between renting the server space, paying for the server compute used and energy costs. Most of the variable costs are taken up by the latter.
"The early days of blockchain were a lot like the search engine wars of the late 1990s and early 2000s," says Tate. "Blockchain companies just needed a lot of server space. The margins they were making were such that they weren't too concerned about cost."
As competition rose, however, cost became more of an issue. "It forced us to be really innovative," Tate says. He points out that when the modern data center was developed 40 years ago, a key requirement was to avoid power failures. So, data centers were equipped with uninterruptible power supplies (UPS), backup batteries and other redundancies.
However, the networked nature of the blockchain means that power interruptions are not as critical. Tate says Verne Global came up with a solution for blockchain companies that cut the service back to basics and allowed them to extend as necessary with additional features like extra security or power protection on a modular basis. This was facilitated by the data center’s Iceland location, which means Verne Global can trust in a massively reliable power supply and an infrastructure that is less dependent upon UPSs and backup batteries.
He says: "It's a major change to how data centers are run. When we take people on tours I tell them that they are going to cover four decades of changes in 40 steps.” In those 40 steps a guest of the Verne Global campus will walk from a conventional data hall with fully redundant electrical and mechanical systems, through a secured vault and into an overwhelming feast for the senses.
The guest enters the next generation of computing which looks, sounds, and feels like a cross between a high-tech clean room and a high-density, industrial-grade factory floor. This new design not only applies to supporting the latest in Blockchain technology, but also is being applied to the latest applications in high performance computing.
This groundbreaking approach keeps data center costs under control but there is still the issue of mining and power demand. Private blockchains, which are becoming more common, will help because they are designed for networks where the nodes are trusted, such as within an organisation. In these cases, it should be possible to devise ways to clear transactions without using as much power. There’s also potential to switch to ‘proof of stake’ verification, which is more energy efficient than 'proof of work' and is being explored by some cryptocurrencies.
The Googles and Facebooks of the blockchain era are already out there and, as they approach scale, want predictable costs and predictable partners. Finding ways to do that has been an exciting challenge.