Jump to Content
Rajagopal Ananthanarayanan (ananth)

Rajagopal Ananthanarayanan (ananth)

Authored Publications
Google Publications
Other Publications
Sort By
  • Title
  • Title, desc
  • Year
  • Year, desc
    Opportunities and Challenges Of Machine Learning Accelerators In Production
    Mahesh Sathiamoorthy
    Manasi N Joshi
    Peter Brandt
    OpML (2019) (to appear)
    Preview abstract The rise of deep learning has resulted in tremendous demand for compute power, with the FLOPS required for leading machine learning (ML) research doubling roughly every 3.5 months since 2012 [1]. This increase in demand for compute has coincided with the end of Moore’s Law [2]. As a result, major industry players such as NVIDIA, Intel, and Google have invested in ML accelerators that are purpose built for deep learning workloads. ML accelerators present many opportunities and challenges in production environments. Workloads span a diverse portfolio including but not limited to visual search, fast text transcoding and translations, recommendations, and scientific computing such as climate prediction. This paper discusses some high level observations from experience internally at Google. View details
    Photon: Fault-tolerant and Scalable Joining of Continuous Data Streams
    Venkatesh Basker
    Sumit Das
    Haifeng Jiang
    Tianhao Qiu
    Alexey Reznichenko
    Deomid Ryabkov
    Shivakumar Venkataraman
    SIGMOD '13: Proceedings of the 2013 international conference on Management of data, ACM, New York, NY, USA, pp. 577-588
    Preview abstract Web-based enterprises process events generated by millions of users interacting with their websites. Rich statistical data distilled from combining such interactions in near real-time generates enormous business value. In this paper, we describe the architecture of Photon, a geographically distributed system for joining multiple continuously flowing streams of data in real-time with high scalability and low latency, where the streams may be unordered or delayed. The system fully tolerates infrastructure degradation and datacenter-level outages without any manual intervention. Photon guarantees that there will be no duplicates in the joined output (at-most-once semantics) at any point in time, that most joinable events will be present in the output in real-time (near-exact semantics), and exactly-once semantics eventually. Photon is deployed within Google Advertising System to join data streams such as web search queries and user clicks on advertisements. It produces joined logs that are used to derive key business metrics, including billing for advertisers. Our production deployment processes millions of events per minute at peak with an average end-to-end latency of less than 10 seconds. We also present challenges and solutions in maintaining large persistent state across geographically distant locations, and highlight the design principles that emerged from our experience. View details
    No Results Found