You are here:Home1/Blog2/Decentralized Computing & Storage: Building a Global Supercomputer...
Decentralized Computing & Storage: Building a Global Supercomputer For All
The early days of blockchain technology focused on securing and transferring digital assets, like tokens, between users. However, blockchain wasn’t the first peer-to-peer network. Before blockchain there were torrents and other methods for transferring information in a distributed way. Torrents are already capable of storing and sending large files in small pieces on hundreds of computers. To download a file via a torrent, you just send a request and dozens or even hundreds of “seed” connections will send you tiny parts of the file. Your torrent client can then stitch together the parts of the file to reassemble the whole file.
The addition of cryptography makes it so that blockchain-secured ledgers can’t have their assets copied. For example, even though you can see all the Bitcoins on the network, you can’t make a copy of a Bitcoin for yourself. The network would reject it as invalid.
If this security applies to decentralized currencies, maybe we can apply it to other arenas. For instance what if you could establish a secure connection for renting idle processing power or unused storage space on your home computer? You could assemble the processing power and storage of thousands of computer on one network to accomplish previously challenging tasks.
This article explores decentralized computing and storage and the ways they could change the ways we think about our devices and what’s possible with computers.
When we talk about computing, we generally mean processing power to accomplish computations. Building a decentralized network of hundreds or thousands of nodes is like building a supercomputer that has access to many parallel processors. The basic idea is the network coordinates the assignment and reassembly of various chunks of computation. Each individual processor is only doing a fragment of the overall job, but in concert the network is faster and cheaper than building a processing farm of your own.
As a user, you would submit a job to the network for some computationally expensive task like rendering or big data analysis. You’d send the requisite information and data to the network where it would get divided, processed, and reassembled. Ordinary people could be the ones providing processing power, leaving their home computers on while they’re asleep or even running high-power processors for profit.
There are several companies exploring this supercomputing on the blockchain trend. The farthest along is Golem who launched their mainnet recently. At the Ethereal Summit in New York in May of 2018, Golem demonstrated the mainnet’s capabilities in rendering 3D CGI images. Renderings ran all day long at the Golem booth, using the network capacity to complete the task. Two other competitors, SOMN and iExec are less far along, but they’re also showing promise in similar arenas. SOMN’s roadmap is more ambitious than Golem or iExec at the moment with more planned features and a different architecture. iExec on the other hand has been consistently early on its development roadmap. All three of these competitors will push each other, and decentralized computing, forward.
2. What is Decentralized Storage?
Whereas computing was about processing power, storage is about hard drive space for saving files. All of us have unused hard drive space on our machines at a given time. For modern computers with hard drives reaching up to 1TB, it’s unlikely an ordinary user will ever run out of space on a system that large. Imagine if you could sell that excess space on the network and act as a tiny part of a cloud storage provider, like Dropbox or Google Drive.
Decentralized storage works by encrypting the entire file and then breaking the encrypted nonsense into tiny shards. The shards get sent to many different computers and only the user’s private key can tell the network how to reassemble and decrypt the file. This means that ordinary computers on the network don’t have access to sensitive information, only encrypted nonsense.
The benefit of decentralized storage for the end user is similar to the benefit of Dropbox or Google Drive. However, decentralized storage is much more secure since your files live on hundreds of machines around the world and nobody knows how to reassemble them except you. Additionally, this type of storage could prove cheaper than centralized solutions. In some systems, file owners pay hard drive operators directly for their services.
Filecoin, Sia, and Storj are the leading companies working on blockchain-based storage. Of the three, Storj is the farthest along, with an operating network that dwarfs Sia’s and an established user base. That said, Sia is also doing well and moving along with their development plans on their operational network. Filecoin has the most financial backing. Its 2017 ICO raised $257 million and received backing from major venture capitalists. So far, however, Filecoin doesn’t have a working product.
3. Open to Anyone
The cool thing about these solutions is they’re open to anyone. The decentralized nature of the blockchain means anyone can set up and operate processors or hard drives. Anyone can make money from their unused computing resources.
On the flip side, it also means anyone can make use of the services. You can store files on Storj or Sia today, and if you’re into 3D rendering, you can use Golem to do that. As the capabilities of these platforms grow, anyone will be able to develop high-powered applications from anywhere in the world and run them on the network.
4. Potential Applications
The potential applications of these platforms are limitless, but they’re most useful in areas where large amounts of processing or storage power are needed. Big data analysis, complex simulations, CGI rendering, machine learning and neural network training, AR/VR world building, and biosystems modeling are some potential areas where this type of computing gets useful.
These platforms get even more interesting when bridges between computing and storage get built or hybrid solutions come along. Imagine storing a huge dataset on a decentralized network and then feeding that data to an algorithm running on a decentralized processing network. Play the scenario out, and it’s possible computers could have less local storage and processing. Instead, they’d increasingly rely on decentralized storage and processing online.
Decentralized storage and processing have the potential to radically alter the way we think about computers and the tasks they can accomplish. If these blockchain-based platforms are successful, they could power the next generation of AI, VR, and other complex programs.