Leveraging ML to Detect Application HotSpots [@scale, of Course!]

Note: Presentation times are in Coordinated Universal Time (UTC).

Website Maintenance Alert

Due to scheduled maintenance, the USENIX website may not be available on Monday, March 17, from 10:00 am–6:00 pm Pacific Daylight Time (UTC -7). We apologize for the inconvenience and thank you for your patience.

If you would like to register for NSDI '25, SREcon25 Americas, or PEPR '25, please complete your registration before or after this time period.

Wednesday, 13 October, 2021 - 16:45–17:00

Sanket Patel, LinkedIn

Abstract: 

This talk will explore various analyses done on service latency metrics and their correlation, while LinkedIn's data-center is under a stress test.

Note: you do not need to be a machine learning expert to make sense of this talk. We will not be diving deeper into the mathematics part of it but would rather focus on the approach.

Sanket Patel, LinkedIn

Sanket is Site Reliability Engineer at LinkedIn where he is working in infrastructure space with the capacity engineering team. He is also into cycling and blogging [superuser.blog].

SREcon21 Open Access Sponsored by Indeed

BibTeX
@conference {276679,
author = {Sanket Patel},
title = {Leveraging {ML} to Detect Application {HotSpots} {[@scale}, of {Course!]}},
year = {2021},
publisher = {USENIX Association},
month = oct
}
Download
View the slides

Presentation Video