arXiv subjects

Srikanth Sundaresan

Publications and source records attributed to Srikanth Sundaresan.

6 recordsLinked to original sources

arXiv2026

Collective Communication for 100k+ GPUs

The increasing scale of large language models (LLMs) necessitates highly efficient collective communication frameworks, particularly as training workloads extend to hundreds of thousands of GPUs. Traditional communication methods face significant throughput and latency limitations at this scale, hindering both the development and deployment of state-of-the-art models. This paper presents the NCCLX collective communication framework, developed at Meta, engineered to optimize performance across the full LLM lifecycle, from the synchronous demands of large-scale training to the low-latency requirements of inference. The framework is designed to support complex workloads on clusters exceeding 100,000 GPUs, ensuring reliable, high-throughput, and low-latency data exchange. Empirical evaluation on the Llama4 model demonstrates substantial improvements in communication efficiency. This research contributes a robust solution for enabling the next generation of LLMs to operate at unprecedented scales.

cs.DC↗

arXiv2023

Analyzing DCTCP and Cubic Buffer Sharing under Diverse Router Configurations

In this work, we look at the impact of router configurations on DCTCP and Cubic traffic when both algorithms share router buffers in the data center. Modern data centers host traffic with mixed congestion controls, including DCTCP and Cubic traffic. Both DCTCP and Cubic in the data center can compete with each other and potentially starve and/or be unfair to each other when sharing buffer space in the data center. This happens since both algorithms are at odds with each other in terms of buffer utilization paradigms where DCTCP attempts to limit buffer utilization while Cubic generally fills buffers to obtain high throughput. As a result, we propose methods for a measurement-driven analysis of DCTCP and Cubic performance when sharing buffers in data center routers via simulation. We run around 10000 simulation experiments with unique router configurations and network conditions. Afterwards, we present a generalizable ML model to capture the effect that different buffer settings have on DCTCP and Cubic streaming traffic in the data center. Finally, we suggest that this model can be used to tune buffer settings in the data center.

cs.NI↗

arXiv2018

On the Impact of Guest Traffic in Open-Access Domestic Broadband Sharing Schemes

Open-access domestic broadband connection sharing constitutes a voluntary practice that is associated with societal, economic and public-safety benefits. Despite this fact, broadband subscribers are usually hesitant to freely share their broadband connection with guests for a multitude of reasons; one of them being sharing their network might hinder their own broadband quality of experience. In this paper, we investigate experimentally the impact of uplink guest traffic on the sharer's broadband quality of experience under both generic and broadband-sharing-specific packet scheduling policies. Both guest-user traffic and access point profiles employed in our study are developed by analyzing real-world traffic traces and measurements, captured from actual broadband sharing networking environments. Our results validate the suitability of hybrid packet scheduling policies for broadband sharing schemes and show that only a few dozen kilobytes per second of uplink guest traffic can be tolerated by sharers without hampering their broadband quality of experience. In this context, we show that the selection of the most appropriate packet scheduling policy for broadband sharing, as well as its respective configuration, depend largely on the capacity of the broadband connection and the policy's packet-dropping behavior on guest traffic.

cs.NI↗

arXiv2017

Spying on the Smart Home: Privacy Attacks and Defenses on Encrypted IoT Traffic

The growing market for smart home IoT devices promises new conveniences for consumers while presenting new challenges for preserving privacy within the home. Many smart home devices have always-on sensors that capture users' offline activities in their living spaces and transmit information about these activities on the Internet. In this paper, we demonstrate that an ISP or other network observer can infer privacy sensitive in-home activities by analyzing Internet traffic from smart homes containing commercially-available IoT devices even when the devices use encryption. We evaluate several strategies for mitigating the privacy risks associated with smart home device traffic, including blocking, tunneling, and rate-shaping. Our experiments show that traffic shaping can effectively and practically mitigate many privacy risks associated with smart home IoT devices. We find that 40KB/s extra bandwidth usage is enough to protect user activities from a passive network adversary. This bandwidth cost is well within the Internet speed limits and data caps for many smart homes.

cs.CR↗

arXiv2016

Haystack: A Multi-Purpose Mobile Vantage Point in User Space

Despite our growing reliance on mobile phones for a wide range of daily tasks, their operation remains largely opaque. A number of previous studies have addressed elements of this problem in a partial fashion, trading off analytic comprehensiveness and deployment scale. We overcome the barriers to large-scale deployment (e.g., requiring rooted devices) and comprehensiveness of previous efforts by taking a novel approach that leverages the VPN API on mobile devices to design Haystack, an in-situ mobile measurement platform that operates exclusively on the device, providing full access to the device's network traffic and local context without requiring root access. We present the design of Haystack and its implementation in an Android app that we deploy via standard distribution channels. Using data collected from 450 users of the app, we exemplify the advantages of Haystack over the state of the art and demonstrate its seamless experience even under demanding conditions. We also demonstrate its utility to users and researchers in characterizing mobile traffic and privacy risks.

cs.NI↗

arXiv2016

Tracking the Trackers: Towards Understanding the Mobile Advertising and Tracking Ecosystem

Third-party services form an integral part of the mobile ecosystem: they allow app developers to add features such as performance analytics and social network integration, and to monetize their apps by enabling user tracking and targeted ad delivery. At present users, researchers, and regulators all have at best limited understanding of this third-party ecosystem. In this paper we seek to shrink this gap. Using data from users of our ICSI Haystack app we gain a rich view of the mobile ecosystem: we identify and characterize domains associated with mobile advertising and user tracking, thereby taking an important step towards greater transparency. We furthermore outline our steps towards a public catalog and census of analytics services, their behavior, their personal data collection processes, and their use across mobile apps.

cs.CY↗