Key takeaways
- Storj provides S3-compatible decentralized cloud storage for developers, sourcing capacity from over 11,000 independent nodes across 96 countries.
- The platform targets developers rather than consumers to avoid building consumer-facing features while reaching end users indirectly through applications.
- Storj launched new branding as Storj DCS with an S3-compatible gateway, multi-region availability, and pricing at $4 per terabyte for storage.
- The decentralized model eliminates data center capital expenditure and maintenance costs, allowing Storj to undercut traditional providers by 80 percent.
- Client-side encryption and edge-delegated authorization enable developers to build secure applications where Storj never accesses encryption keys or user information.
A cloud storage network built for developers
Ashton Addison: For viewers who don't know Storj, what does it do in the blockchain space?▶ 00:19
John Gleeson: We're a cloud object storage service for developers. We make open-source software that anybody can run, whether that's data centers, prosumers or businesses. Many of those businesses have spare hard drive capacity, so they run our software and share it with us. We aggregate it and make it available to developers as an S3-compatible storage service. It's your alternative to Amazon in the decentralized space. Today our capacity comes from over 11,000 independent people around the world running our software, in I think 96 countries now.
Ashton: Why focus on developers rather than consumer storage?▶ 01:18
John: If we competed with the Dropboxes and Boxes of the world, we'd have to market to a far larger number of people. We'd also have to build a lot of things that aren't core to what we offer. By targeting developers, we reach the same customer base indirectly. We provide a service that's 80% cheaper and lets them build more private and more secure applications. We get into those applications and reach all of those end users without having to break through in the consumer space or deal with consumer support issues like passwords. So far it's working out great.
We think of it like the Airbnb for hard drives.
New brand, new pricing, and how it's cheaper
Ashton: What did you announce this past week?▶ 02:36
John: Developers told us they loved the distributed storage, the security and the privacy, but they wanted to interact with us the way they do with every other cloud storage provider. That meant an S3-compatible gateway, which is a big challenge when your storage is spread over 11,000 nodes. We spent about eight months on it. We also launched a new brand, Storj DCS, for decentralized cloud storage, because developers found some of our earlier brand decisions confusing. Alongside the gateway we've introduced multi-region high availability for every component of the network. We've also launched new pricing: $4 per terabyte for storage and $7 per terabyte for egress bandwidth. And there's a free tier of 150 gigs of storage and bandwidth. The reception has been excellent.
Ashton: Decentralized storage on Ethereum doesn't seem cheap. How can you be cheaper than a centralized provider?▶ 04:16
John: The traditional way is to build a data center, fill it with racks and people, bring in internet pipes and sell cloud storage. You have to be at absolute hyperscale to even enter that market. Decentralized cloud storage lets you enter in a completely different way, with opex instead of a ton of upfront capex. We've crowdsourced the same storage by tapping latent capacity around the world, so we avoid the upfront and ongoing maintenance costs and can pass the 80% savings on to customers. It's also economically rewarding for the people who run storage nodes: about 60 cents of every dollar we take in goes back out to the network. We think of it like the Airbnb for hard drives. It delivers the same service, with an enterprise SLA and high availability, at a cost point that can't be touched.
Privacy by architecture
Ashton: Why is Storj more private and secure for developers?▶ 06:07
John: A lot of the advantages are a byproduct of what you have to build to run a decentralized platform at all. When you run on hardware belonging to people you don't know, you need strong encryption and built-in redundancy through erasure coding. That lets you encrypt data client-side, encrypt the metadata, break it into tiny pieces and spread it across many nodes. You can lose a bunch of nodes and still keep the data available. We then expose that as features. With end-to-end encryption we never have the keys, and we don't want to know what's stored on our platform. We also delegate authorization to the edge. Developers can build file-sharing applications where no storage component knows any usernames or users. All we know is that a request is valid or invalid. Because of the fundamental architecture, it's impossible for us to get at that data.
Ashton: Do developers have to run nodes themselves, and can anyone run one?▶ 08:56
John: Early on we made a design decision driven by our audience. With most decentralized projects, a single node handles every function. We found most users either had hard drive space to share or needed to store data for an application, but rarely both. So we split the network into three peer classes. One handles metadata and intermediation, including deciding which storage nodes hold data. The second is the storage node: if you have spare drive space and bandwidth, you set it and forget it, and it stores data and lets people retrieve it. The third is the developer client, which comes as an S3-compatible gateway, a developer library and a command-line interface. There are also tools that plug into things like FileZilla and rclone. If a tool works with Amazon S3, it'll work with Storj: generate credentials and away you go. Developers don't need to run anything.
it's impossible for us to get at that data.
The Storj token and what drives growth
Ashton: How is the Storj token used in the ecosystem?▶ 11:19
John: It's a utility token that facilitates transfers of value for storing and retrieving data. Developers can pay by credit card or with the token, and there's an incentive to choose the token. We pay storage nodes in the token, and it's available on a couple dozen exchanges, so it's easy to convert. Some node operators make a couple of dollars a month when they first join, and others make a thousand dollars a month. Doing those payouts through a standard payment processor would eliminate a lot of the value. The recent increase in fees has been painful in some cases, I'll fully admit that. But we're working with Matter Labs and zkSync on an L2 payout solution, which is turning out to be an excellent alternative.
Ashton: What will be the key factors in Storj's long-term success?▶ 13:55
John: The biggest one is channel partners: finding applications that have end users and scale very broadly, and establishing traction through those partners. The free tier is also critical, because it lets people who find decentralized cloud storage intriguing try it before they invest. Then there are the incentives we can offer for using the Storj token to buy storage, which make it even more economical for end users. So far, knock on wood, we're seeing great traction and a huge burst already with the new features, branding and pricing model.
This interview has been edited for length and clarity. Recorded April 2021.
