The Sia blockchain enables a marketplace for decentralised data storage, which allows private data storage services to be competitively priced and safely replicated, and aims to be more secure than existing cloud storage.
Sia is the result of work from a a venture backed company founded in Boston in 2014 called Nebulous Inc., which operates a network of cloud storage and develops the Sia software. Sia has had a live product since 2016 and the roadmap moving forward hopes to see the network reach speeds similar to Amazon S3 and become a true competitor with regards to functionality and user experience, with the associated security and privacy benefits of a decentralized network of appropriately incentivized rational actors. Sia’s core and founding team is comparatively small and development focused; including a front-end designer with extensive experience in product management and experience as a software engineer for MyEtherWallet. The Sia team decided to hard-fork the networks codebase in late 2018 as a means to mitigate the centralizing influence of those ASICs specifically built by Bitmain and Innosilicon though the Obelisk (a subsidiary of Nebulous) ASIC’s will continue to function post-fork.
The Sia blockchain is heavily influenced by Bitcoin, with a modified notion of transactions designed for storage contracts. A Sia transaction includes a ‘FIle Contract’ field that determines: the duration of the contract (default 90 days), the required frequency of proofs of correct storage, the reward per correct proof, the fee per failure to submit proof of storage and the maximum number of such failed proofs before the contract is breached. When a user decides to commit data to the Sia network, they interface with software that breaks the data up into 30 shards before encryption and distribution across the network. This is achieved via erasure coding that allows for the full original file to be retrieved on the basis of any 10 of the 30 shards, thereby introducing redundancy capable of tolerating up to 2/3rds of the shards becoming corrupted or otherwise unavailable. A user enters into a contract with a data storage provider and then the network is responsible for ensuring the data’s ongoing availability by periodically requiring proofs of storage. The storage provider will be compensated for correct proofs of adequately stored data and incur costs to themselves should proofs not be submitted. A network of payment channels is utilized, analogous to Bitcoin’s Lightning Network, to provide low-cost small value transactions between storage providers and network users. To compliment Sia, the team launched a second layer protocol in February 2020, named Skynet, that offers storage for a variety of files as well as hosting decentralized applications (Apps), meaning that Web3 developers can store their applications on a truly decentralized network rather than centralized alternatives as is the common scenario. Beyond this it is envisioned that Skynet will serve as a content delivery network (CDN), offering publishers and content creators a censorship-resistant service.
Those providing data storage capacity to the network are required to stake Sia as collateral for the opportunity to provide services. This allows storage providers to be fined should they renege on their commitments to both store data, for the agreed time, and sufficiently frequently prove this to be the case. User’s renting out distributed storage capacity on the network will make payments in Siacoin. Sia hope to provide order of magnitude cost reductions per unit data stored per unit time, although this price comparison will only be truly meaningful once the latency and user experience match that of existing centralized cloud storage providers.
Siacoin’s PoW consensus mechanism uses the Blake2b hash algorithm, with two variations in its history. Sia mining has a complex and controversial history. In mid-2017, David Vorick announced that ASIC development could not be prevented and that the core team, under a different corporate entity, would develop its own ASICs so as to better control the security of the network. Pre-orders on these “Obelisk” devices would also help fund core development. Delivery was promised for mid-2018, but two ASICs beat the Obelisk to the market, first the Bitmain A3 and then the Innosilicon S11, which was significantly more efficient than either the A3 or the Obelisk specs. It was revealed that the Obelisk was compatible with both the original Blake2b and a slight variation that would render other ASICs optimized for the original Blake2b incompatible. In response, the SIA community proposed a fork to implement the variation so as to preserve both the profitability of the Obelisk and appease a community unhappy with a core development team that was late on shipping now-unprofitable miners. After initially opposing the fork, the core team, citing concerns about the complete monopoly the S11 had on the network, ended up releasing a fork to make the Obelisk the only compatible ASIC. This also gave rise to a variety of other forks. Additional details and links about this history is detailed well here. The delay and underperformance of the Obelisk resulted in a class action lawsuit against the team.