Building a fair cloud-hosted financial exchange

Gupta, Eashan

Permalink

https://hdl.handle.net/2142/120585 Copy

Description

Title

Building a fair cloud-hosted financial exchange

Author(s)

Gupta, Eashan

Issue Date

2023-05-03

Director of Research (if dissertation) or Advisor (if thesis)

Mittal, Radhika

Department of Study

Computer Science

Discipline

Computer Science

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Financial exchange
• logical clock
• cloud
• fairness

Abstract

We consider the problem of hosting financial exchanges in the cloud and build a real system for it. Financial exchanges require predictable, equal latency to all market participants to ensure fairness for various tasks, such as high speed trading. However, it is extremely difficult to ensure equal latency to all market participants in existing cloud deployments, because of various reasons, such as congestion, and unequal network paths. Delivery Based Ordering (DBO) theory offers solution to address the unfairness that stems from lack of determinism in cloud networks. DBO argues that predictable or bounded latency is not necessary to achieve fairness. It is inspired by the use of logical clocks in distributed systems, a new approach, that ensures fairness by instead correcting for differences in latency to the participants. The variable latencies in a cloud deployment affects fairness and DBO is designed to resolve these issues in-theory. In this thesis, we build a system to show the feasibility of DBO theory on a public cloud using standard virtual machines. While building the system, we find that these network challenges not only affect the issues of fairness, but it also affects our implementation on multiple levels like handling buffer buildups, system artifacts, ensuring performance and low-latency, scaling such a system, enforcing fairness, collecting traces etc. All of these implementation challenges require careful design and experimentation, including finding and resolving bottlenecks, platform specific optimizations etc. We further evaluate DBO in our public cloud deployment. We use our deployment to study the impact of various parameters of DBO and find the limitations and advantages of our system. We also discuss the possible future directions and how the DBO system would look like in the future. Finally, we demonstrate that it is feasible to achieve guaranteed fairness and sub-100μs while operating at high transaction rates.

Graduation Semester

2023-05

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/120585

Copyright and License Information

Copyright 2023 Eashan Gupta

Owning Collections