Limitations of Zilliqa’s Sharding approach

It is evident that if you don’t have sharding from day 1, your blockchain has no chance of scaling with adoption. Building sharding after the fact is extremely hard. For instance, Ethereum has one of the strongest engineering teams in the blockchain space, and yet their sharding release was pushed back yet again. In their situation, integrating sharding into the system is similar to changing a car engine while the car is driving.

I was the engineer #1 at a database company called MemSQL. MemSQL builds a distributed analytics platform that has large clusters deployed at Goldman Sachs, Uber, Comcast, Akamai and many other enterprise companies. While at MemSQL, I was responsible for its sharding implementation. Since then, I’ve co-founded Near Protocol, which has two other early MemSQL engineers who were responsible for cross-shard transactions and complex distributed joins, as well as four ex-Google engineers.

In the past, we have built sharding that powers large clusters in production and processes millions of transactions per second per aggregator node. From our experiences, we know well how to implement sharding on complex systems and what practical issues it will have.

Leave a Comment