Large-scale cluster deployments are common in today's cloud-hosted application environments. Online service providers such as Amazon, Facebook and Google often employ clusters of thousands of nodes for serving web requests.
These online services often handle thousands of customer requests per second.
However, developing efficient load balancers for large, distributed
clusters is challenging for several reasons: (i) large clusters
require numerous scheduling decisions per second, (ii) such
clusters typically consist of heterogeneous servers that widely
differ in their computing power, and (iii) such clusters often
experience significant changes in load.
The goal of this project is to develop scalable, heterogeneity-aware load balancers. We will employ queueing-theoretic ideas to design simple yet adaptive load balancers that provide near-optimal performance.
Collaborators:
Publications:
- EASY: Efficient segment assignment strategy for reducing tail latencies in Pinot.
Seyyed Ahmad Javadi, Harsh Gupta, Robin Manhas, Shweta Sahu, Anshul Gandhi
ICDCS 2018 [pdf]
- Modeling and Analysis of Performance under Interference in the Cloud.
Scott Votke, Seyyed Ahmad Javadi, Anshul Gandhi
MASCOTS 2017 [pdf]
- DIAL: Reducing Tail Latencies for Cloud Applications via Dynamic Interference-aware Load Balancing.
Seyyed Ahmad Javadi, Anshul Gandhi
ICAC 2017 [pdf]
- Minimizing Electricity Cost for Geo-Distributed Interactive Services with Tail Latency Constraint.
Mohammad Islam, Anshul Gandhi, Shaolei Ren
IGSC 2016 [pdf]
- HALO: Heterogeneity-Aware Load Balancing
Anshul Gandhi, Xi Zhang, Naman Mittal
MASCOTS 2015 [pdf]
- Optimal Load-Balancing for Heterogeneous Clusters
Anshul Gandhi, Naman Mittal, Xi Zhang
Poster @ DCC 2015 [pdf]
- Analyzing the Network for AWS Distributed Cloud Computing
Anshul Gandhi, Justin Chan
DCC 2015 [pdf]
Students:
Sponsors: