Celestia, a crypto project that builds modular blockchain networks, has announced the completion of its first upgrade, The Shwap upgrade. The upgrade comes with a new set of features that makes data availability (DA) network sampling 12 times faster than before.
The update is now active on the Arabica and Mocha testnets using the celestial-node v0.18.12. Shwap allows leather blocks and smaller nodes by reducing the storage requirements by sixteen times, further advancing towards the main goal of the Celestia community roadmap, which is to scale to 1 GB blocks.
Celestia Improvement Proposal, or CIP-19, specifies Shwap as a new messaging protocol and storage system for the DA network. Shwap keeps the need for light nodes minimal while enabling huge scaling of the DA throughput. This allows existing crypto apps to speed up and makes new applications readily accessible.
Apps with lower prerequisites allow anyone to operate a light node either in a browser or a wallet, opening the door for truly verifiable web apps. After some more testing, Shwap should drop its Mainnet Beta sometime in November.
The DA Network Now
The DA Network currently runs on two main protocols. The first one handles circulating chain headers and employs a go-header system. In contrast, the second protocol focuses on managing block data, which is organized within a data square consisting of small shares or samples.
Light nodes are responsible for sampling these shares, and the current protocol, which is rooted in the IPLG system from the Devnet era, centers around this sampling strategy. Although it is operational, this protocol faces several scalability challenges, especially when accommodating a one-gigabyte block.
Challenges:
One of the common problems in Blockchain Engineering is hash addressability, which is the ability to identify and locate data stored as hashes across the various nodes in a blockchain network. Fortunately, hash addressability has some readily apparent benefits, such as a lack of duplicates and a nice amount of flexibility.
For instance, one can store the hash-based data structures’ hashes (such as data square storage). When light nodes carry out verifications, the partial data and the Merkle proof of each partial data are downloaded by retrieving each node individually.
With a 528×528 data square, the four-step process becomes a laggard. While a full node has the global index of the Merkle tree, the process is rather slow. As a result, hash addressability pointed to two major issues: one, storing Merkle proofs instead of recalculating them on the fly; two, banking on a slower data access pattern, i.e., O(log2n) while O1 can be utilized to increase the efficiency for large data volumes.
Shwap Upgrade:
The Shwap upgrade, which comes from the words share and swap, solves the scalability issues that the DA network continues to experience by changing how data is stored and accessed within the protocol.
Further, in Schwab, the previous inefficient and experimental version of block reconstruction is also eliminated to favor a more production-ready version in the future upgrade. In pre-Shwap, the put time for 2 MB was two seconds; however, in post-Shwap, the set time for 2 MB is reduced to 10 milliseconds.
