Paper 2018/269

Vault: Fast Bootstrapping for the Algorand Cryptocurrency

Derek Leung, Adam Suhl, Yossi Gilad, and Nickolai Zeldovich


Decentralized cryptocurrencies rely on participants to keep track of the state of the system in order to verify new transactions. As the number of users and transactions grows, this requirement becomes a significant burden, requiring users to download, verify, and store a large amount of data to participate. Vault is a new cryptocurrency design based on Algorand that minimizes these storage and bootstrapping costs for participants. Vault’s design is based on Algorand’s proof-of-stake consensus protocol and uses several techniques to achieve its goals. First, Vault decouples the storage of recent transactions from the storage of account balances, which enables Vault to delete old account state. Second, Vault allows sharding state across participants in a way that preserves strong security guarantees. Finally, Vault introduces the notion of stamping certificates, which allow a new client to catch up securely and efficiently in a proof-of-stake system without having to verify every single block. Experiments with a prototype implementation of Vault’s data structures show that Vault’s design reduces the bandwidth cost of joining the network as a full client by 99.7% compared to Bitcoin and 90.5% compared to Ethereum when downloading a ledger containing 500 million transactions.

Available format(s)
Publication info
Published elsewhere. MINOR revision.Network and Distributed Systems Security (NDSS) Symposium 2019
Contact author(s)
dtl @ mit edu
2018-12-19: revised
2018-03-13: received
See all versions
Short URL
Creative Commons Attribution


      author = {Derek Leung and Adam Suhl and Yossi Gilad and Nickolai Zeldovich},
      title = {Vault: Fast Bootstrapping for the Algorand Cryptocurrency},
      howpublished = {Cryptology ePrint Archive, Paper 2018/269},
      year = {2018},
      doi = {10.14722/ndss.2019.23313},
      note = {\url{}},
      url = {}
Note: In order to protect the privacy of readers, does not use cookies or embedded third party content.