Building of Cloud Infrastructure with a Custom Auto-Scalable Solution for the Blockchain "Visa-level" Product

zkSync (by matter-labs.io ) is a trustless L2 scaling solution to enable mass adoption of public blockchains. The solution is a remedy for the growing problem of unreliable and expensive transactions on the Ethereum blockchain due to sky-high gas fees. zkSync can process more than 200 transactions per second, which is comparable to transaction volumes routinely processed by PayPal. The product has an upgradable nature and will be able to increase processing speed to be comparable with Visa over the next few months.

Business Task

We were assigned to promptly develop a server infrastructure with the highest level of security, custom, and cost-effective auto-scaling for one of the products—zkSync—that was successfully released in June 2020. The main task was exactly to prepare the product for scaling. We had to do load tests because it was very important to determine the burst throughput close to the day of the launch of the product beta-version. The Dysnix team developed a custom and fault-tolerant mechanism of scaling based on unique product metrics.

The Dysnix team's task was:

  • To build 2 environments (Staging & Production)
  • Create and implement a list of recommendations to achieve the highest level of fault-tolerance and security
  • To describe the infrastructure as code
  • To develop custom auto-scaling solution based on business' metrics
  • To do load tests
  • TCO optimization (it was ten times more expensive on-premise compared to the Cloud)
  • To ensure early warning of possible technical problems
  • Solutions

    • Building a high-available, secure, multizone, and scalable Kubernetes-based server Infrastructure in GCP with private topology
    • Security—a list of requirements and instructions for using security best practices.
    • Building a cost-effective infrastructure
    • Development of the custom auto-scaling solution
    • Full implementation of Infrastructure as code using Terraform
    • Implementation of the monitoring and alerting systems with custom triggers
  • Value delivered by Dysnix

    • Prompt consulting when the team decided to change the strategy and launch the product quickly. At the stage of consultation, we proposed the architecture that met all the requirements, and finally, it was implemented in the production environment successfully. The Dysnix team developed the solution within 2 weeks. And it was tested and implemented in just 24 hours.
    • Full integration and permanent communication of the Dysnix team into the project as if it was a part of the in-house team, which is essential at the stage of implementation of the solutions.
    • Consultation of the in-house development team. A part of the solution development was taken by Dysnix because it was complicated for the in-house engineers and they didn't have the expertise.
    • Dysnix went all the way from the testing to production environments: they provided the support at all stages of the load tests, and were involved in the in-depth debugging processes.
    • The Dysnix team continues supporting the project.