FaaSKeeper: Learning from Building Serverless Services with ZooKeeper as an Example

Overview

• FaaS (Function-as-a-Service) has revolutionized cloud computing by replacing persistent virtual machines with dynamically allocated resources.
• This shift trades locality and statefulness for a pay-as-you-go model more suited to variable and infrequent workloads.
• The main challenge is to adapt services to the serverless paradigm while meeting functional, performance, and consistency requirements.

Plain English Explanation

FaaS, or Function-as-a-Service, has transformed cloud computing in a significant way. In the traditional cloud model, applications were run on virtual machines (VMs) that were persistent, meaning they were always available and running. FaaS has replaced these VMs with a more dynamic approach, where resources are allocated on-demand as needed for specific tasks or functions.

This shift has several advantages. The pay-as-you-go model of FaaS is better suited for workloads that are variable or infrequent, as you only pay for the resources you use. This can be more cost-effective than maintaining always-on VMs. However, the tradeoff is that you lose some control over the locality and state of your application, as the underlying infrastructure is more ephemeral and abstracted.

The main challenge with FaaS is adapting existing services and applications to this new serverless paradigm, while still meeting the functional, performance, and consistency requirements that users expect. This paper explores ways to push the boundaries of FaaS computing and design a serverless variant of ZooKeeper, a popular coordination service.

Technical Explanation

In this work, the researchers push the boundaries of FaaS computing by designing a serverless variant of ZooKeeper, a centralized coordination service with a safe and wait-free consensus mechanism. They define synchronization primitives to extend the capabilities of scalable cloud storage and outline a set of requirements for efficient computing with serverless.

The resulting system, called FaaSKeeper, is the first coordination service built on serverless functions and cloud-native services. By exploring the limitations of current serverless offerings, the researchers propose improvements essential for complex and latency-sensitive applications like ZooKeeper.

FaaSKeeper maintains the same consistency guarantees and interface as the original ZooKeeper, but with a serverless price model that can lower costs by up to 110-719x for infrequent workloads. The researchers share their serverless design lessons based on their experiences implementing the ZooKeeper model in a cloud-deployable fashion.

Critical Analysis

The paper does a good job of identifying the key challenges in adapting existing services to the serverless paradigm, such as maintaining functional, performance, and consistency requirements. By designing a serverless variant of ZooKeeper, the researchers have provided a concrete example of how these challenges can be addressed.

However, the paper does not delve deeply into the limitations or potential issues with their FaaSKeeper approach. For example, it would be helpful to understand the trade-offs between the serverless model and the traditional ZooKeeper approach in terms of performance, scalability, and reliability under different workload conditions.

Additionally, the paper could have explored the broader implications of their work beyond the specific ZooKeeper use case. How could the insights and design lessons from FaaSKeeper inform the development of other complex, latency-sensitive serverless applications? What are the potential areas for further research and improvement in the serverless computing landscape?

Conclusion

This paper presents a significant advancement in the field of serverless computing by demonstrating how a complex coordination service like ZooKeeper can be adapted to the FaaS model. The researchers have shown that it is possible to maintain the same consistency guarantees and functionality while leveraging the cost-efficiency and scalability benefits of serverless architectures.

The lessons learned from the FaaSKeeper project can inform the development of other sophisticated serverless applications, potentially paving the way for a new generation of cloud-native services that seamlessly integrate with the dynamic, on-demand nature of FaaS. As the serverless paradigm continues to evolve, this work highlights the importance of pushing the boundaries of what is possible and addressing the technical challenges that arise.