Decentralized finance (DeFi) has quickly gained favor among cryptocurrency proponents as a multitude of platforms continue to emerge — from decentralized lending services like MakerDAO to security token issuance platforms like Polymath.
In particular, the impressive growth of Maker’s Dai stablecoin is indicative of the broad popularity of DeFi applications and products. However, the potential scaling issues of Maker Dai has led to several needs to raise its stability fee to retain parity with its USD price peg — revealing a problem with DeFi products: scalability and performance in line with existing financial systems.
Katallassos is a blockchain designed for deploying and running DeFi applications with the type of interoperability and performance that financial professionals have become accustomed to over the years. The platform relies heavily on its optimized consensus algorithm, known as Albatross, which is a proof-of-stake (PoS) algorithm with near-theoretical performance capacity that retains PBFT security assumptions.
So, what exactly is Albatross?
Albatross for Katallossos
Built by Trinkler Software AG in collaboration with Nimiq, Albatross’s conception and design were led by Reto Trinkler, Bruno Franca, Marvin Wissfield, Pascal Berrang, and Philipp von Styp-Rekowsky.
Notably, Trinkler is also the co-founder of Melonport, the crypto asset management protocol.
Albatross draws inspiration from several ‘speculative byzantine fault-tolerant’ algorithms where on-chain consensus is designed to rival centralized models with a ‘trust but verify’ maxim instead of the ‘never trust’ position taken by pessimistic BFT algorithms. Pessimistic variants are more conservative in their security approach but have slower processing speeds as a result.
As a speculative BFT algorithm, Albatross takes a comprehensive approach with influences from Tendermint, Ouroboros-BFT, and other algorithms deployed in permissionless networks.
Speculative BFT algorithms are designed as advances over similar algorithms, like Tendermint’s standard BFT, that place limits on the number of Byzantine actors. At a high level, this means that, in the best or ‘optimistic’ cases, the performance capacity of the algorithm can surge to near-theoretical levels for single-chain PoS systems. Conversely, when nodes are Byzantine, performance is akin to standard PBFT algorithms in the ‘pessimistic’ state.
Like all PoS consensus algorithms, Albatross relies on a validator set, which is composed of participants who stake the native token as a representation of risk (i.e., skin in the game) to act honestly. Should the participants act maliciously, then their stake is slashed. However, Albatross has some quirks that are tailored for high performance ‘optimistic’ cases where the performance can climb when tampering of the protocol is not occurring.
Albatross utilizes two kinds of blocks: macro and micro blocks. Macro blocks determine the active validator list for each cycle that is comprised of 4 micro blocks. Each micro block contains the transactions and is randomly selected from the active validator set. Macro blocks are produced with PBFT while micro blocks only need to be signed by the validator.
By leveraging the power of randomness with the validator selection — Albatross uses a VRF — participants can correctly discern who the next random validator for a micro block will be. The VRF is initiated with BLS signatures — similar to DFINITY. However, any delays, invalid blocks, or forking of macro blocks will trigger the pessimistic, conservative-driven performance of the algorithm.
Macro blocks do not contain transactions and are used as the measure for the main chain by simply measuring the longest chain of macro blocks. Macro block proposals are voted on in two rounds by the current set of validators. If there are no cases of forking of macro blocks, invalid block proposals, or delays caused by Byzantine validators, then the algorithm can function in its ‘optimal’ state. However, requirements for slashing will slow the production of blocks down to the conservative state with traditional PBFT assumptions.
One of the prominent trade-offs between an ‘optimistic’ consensus algorithm and a more conservative approach — such as Nakamoto Consensus in Bitcoin — is the notion of availability vs. consistency. According to the Albatross whitepaper:
“From the CAP theorem [11], we know that when suffering a network partition a blockchain can only maintain either consistency or availability. PBFT favors consistency over availability and will stop in the presence of a network partition. Albatross also favors consistency, but can still produce a few micro blocks before stopping.”
Consistency refers to all of the network participants converging on the same, most recent, value while availability refers to whether or not they can freely access the most recent value without interference. In Bitcoin, forks can continue parallel to the root chain for extended periods before converging on the main chain — meaning that the algorithm favors availability. The opposite is true with Albatross, where an affinity towards consistency means that forks of the most recent state are mutually exclusive in their operation, only one or none can function.
Overall, Albatross highlights three primary optimizations that allow it to differentiate between an ‘optimized’ and conservative (i.e., pessimistic) performance states:
- Proportional validator staking makes the network permissionless
- Selection of block producers via the previous block using the VRF increases adversary resistance
- Increasing performance by relying on speculative execution of the blocks
Speculative execution of the blocks concerns the concept that there is no block time target in the optimal state. For example, in Bitcoin, the difficulty adjustment algorithm for mining is designed to ensure an average block time of 10 minutes. In Albatross’ optimal state, blocks can be produced nearly as fast as the network allows — enabling peak performance.
The results of Albatross’ design yield some impressive theoretical performance metrics including low latency and up to 10k transactions per second (TPS) on-chain. More technical details on Albatross are available in its white paper.
Implications of Scaling On-Chain for DeFi
High-performance on-chain processing is a marked advantage for a blockchain underlying numerous DeFi products that require quick settlement and trade execution. Many DeFi products built on Ethereum, particularly DEXs, often rely on hybrid on-chain/off-chain systems because the on-chain capacity cannot adequately support trade matching or high volume settlements.
Facilitating a permissionless ecosystem of DeFi that supports a range of financial tools and products — from crypto-collateralized loans to futures contracts — needs a powerful base layer that is capable of adapting its performance based on security requirements. Albatross strives for this and works in combination with other building blocks of the Katallassos platform such as the ACTUS financial standard, Substrate development framework, and Enso virtual machine.
The experience of Katallassos’ team in financial tools (i.e., Melonport) along with their partnerships among ChainLink, Polkadot, and Nimiq position them well for steering innovation in the field of DeFi.
DeFi promises a future of interoperable digital assets and open, modular frameworks for creating financial products. The prospect of a mainstream DeFi ecosystem requires a robust base performance layer, and Albatross is the speculative BFT consensus algorithm designed to supplement financial products on the blockchain.
