Would you store your data with a cloud storage provider that crowdsources its storage infrastructure but promises fast, secure service at extremely low prices?
Storj Labs is counting on it.
The seven-year-old Atlanta-based startup provides decentralized object-based cloud storage by renting excess storage capacity from companies and individuals.
Because Storj doesn’t have the expense of purchasing or maintaining its own data center infrastructure, it can provide much cheaper service than public cloud providers like AWS, Azure, or Google Cloud, said Storj CEO Ben Golub.
Storj’s pricing is one-sixth or one-seventh of what public cloud providers charge for static data storage, and it’s even more affordable if you factor in bandwidth costs, Golub said in a recent interview with DCK. Storj ensures that data stays secure and private by encrypting the data and using erasure coding, which breaks up the data into different pieces and stores them in different locations, he said.
“Think of us like an AirBnB for disk drives,” he told us. “Instead of storing your data on drives that we own, purchase, and run in large data centers, we store your data across a big network of drives run by individuals and companies all over the planet. It lets us be really competitive.”
Storj is among half a dozen storage companies in the emerging decentralized cloud storage market, which has the potential to disrupt the cloud-based enterprise storage market within the next two to three years, IDC analysts say.
Storj and the other players still need to prove that their technologies are viable for enterprises, but until then it’s a market for enterprises to keep an eye on, said IDC analyst Eric Burgener, who estimates that Storj’s object storage service is up to 80 percent cheaper than the large cloud providers.
“It’s an economic play,” Burgener told DCK. “The reason you’d buy this object service from one of those six or seven vendors instead of Amazon and the other public cloud providers is purely economics.”
Other startups in the space –all enabled in one way or another by blockchain – include BloqChain, Cryptyk, Filecoin, Sia, ScPrime, and 0Chain, with Storj and Sia being the most mature, said IDC analyst Lucas Mearian.
“They are all companies that use a peer-to-peer network in order to aggregate data from unused sources from around the world,” Mearian told DCK.
Storj Targets Developers as a Path Toward the Enterprise
The analysts said they were impressed with Storj’s technology. Erasure coding makes the service more highly available than most enterprise-class storage arrays, Burgener said.
Storage’s current infrastructure is made up of more than 13,000 node operators spread across more than 90 countries, he said. Storj Decentralized Cloud Storage (DCS) software first encrypts files that are to be uploaded. Then it cuts each file into 80 segments.
The erasure coding algorithm uses 29 data segments and 51 parity segments with all the segments stored in different storage nodes. That means when customers want to retrieve a file, they only need 29 of the 80 segments to reconstitute the file. That also means 52 of the storage nodes have to be lost before data integrity is compromised, Burgener said.
“It makes it extremely resilient,” he said.
Storj isn’t targeting enterprises yet. Today, most of its 30,000 customers are developers building new cloud-native applications, services, and open-source projects, Golub said.
Developers are more inclined to try new technologies like decentralized cloud object storage, he said. “This has been a proven path to mainstream adoption for many new technologies that make their way to the enterprise, including centralized cloud computing.”
The company is also marketing its service to managed service providers, data scientists, research institutions, and startups with data-oriented applications, Golub said. Besides software development, use cases for its service include software distribution and video storage and distribution, data storage and sharing for artificial intelligence and machine learning workloads, and hot backups, he said.
The company has designed its product to deliver enterprise-grade performance and durability. In fact, Storj offers enterprise-level service level agreements with 99.95% availability and eleven 9s of durability, Golub said. But Storj currently doesn’t yet have some ancillary items that many larger enterprises would require, such as customized support, systems integration partners and certain certifications, he added.
Once its Storj DCS software matures a bit more, the company will offer the product to enterprises, Golub said.
“Once we have the validation from developers and a few more quarters of operating experience we can move aggressively into the enterprise,” he said. “We do have regular conversations with enterprises to ensure we are aligning new Storj DCS features and capabilities with their needs for long-term adoption.”
For example, based on feedback from customers (including potential enterprise customers), the company this April added new features including multi-region and multipart upload support and an S3-compatible gateway, which provides customers a familiar interface, he said.
Dennis Hahn, Omdia’s principal analyst of data center storage, is skeptical that enterprises will want to adopt Storj’s technology.
They may be willing to use it for non-sensitive and non-mission critical data, but the issue is cultural, because enterprises want to know where their data is, he said in an interview with DCK. With other cloud providers, the data is stored off-premises, but at least they know it’s within AWS, Azure, or Google Cloud’s data centers, he said.
“It’s difficult for me to see an enterprise store data that way. It’s the cultural aspect of where is my data?” Hahn said. “But maybe time will prove it.”
The IDC analysts agree that Storj and its competitors have to prove themselves. To do that, they need an initial set of enterprises to test the service and then adopt it.
“Right now they have to prove the mettle of the service to ensure that performance is reliable and does meet availability requirements,” Mearian said.
Burgener reiterated that the crowdsourced model for cloud-based storage could be disruptive if Storj and its startup competitors prove that their technology can be used for enterprise workloads.
“It’s not clear it will prove out,” he said. “But if it did, the economics of it are just going to blow the other options out of the water.”
While Storj wants data center operators as customers in the future, they can currently rent their spare capacity to Storj and be a supplier of storage space, Golub said. They can make money off their excess storage, and if they ever need the storage back, they can easily reclaim it, he said.