Papers of the day   All papers

Serverless in the Wild: Characterizing and Optimizing the Serverless Workload at a Large Cloud Provider

Comments

Mark Russinovich: Working with Microsoft Research, we've published insights into the characteristics of serverless functions in Azure: "Serverless in the Wild: Characterizing and Optimizing the Serverless Workload at a Large Cloud Provider" https://arxiv.org/pdf/2003.03423.pdf

3 replies, 233 likes


Jonathan Mace: Very interesting and revealing paper from Microsoft Research about Serverless Workloads in Azure just popped up on arXiv https://arxiv.org/pdf/2003.03423.pdf cc @adriancolyer @MShahrad @rodrigo_fonseca and others

2 replies, 60 likes


ahmet alp balkan: Cool paper by @AzureFunctions folks on analysis of FaaS workload characteristics and patterns on managed infra. https://arxiv.org/abs/2003.03423

1 replies, 35 likes


Colin Breck: A very interesting empirical, longitudinal study of FaaS execution patterns from Microsoft Research: https://arxiv.org/pdf/2003.03423.pdf #Serverless #FaaS

1 replies, 11 likes


Ruichuan Chen: Serverless workloads from Azure -- Interesting! There is another work from former MPI folks which takes a very different approach to strike the balance between startup latency and resource efficiency in serverless: https://www.usenix.org/system/files/conference/atc18/atc18-akkus.pdf worth a read @mpi_jcmace @adriancolyer

0 replies, 9 likes


Eduardo Laureano: This is great work by Microsoft Research indeed, they are excellent partners to the PG Cc: @paulbatum

0 replies, 9 likes


Peter Kalambet: Moreover, for most applications the average execution time is at least two orders of magnitude smaller than the average inter-arrival times. Great recap of arXive paper (https://arxiv.org/abs/2003.03423) on serverless from @adriancolyer https://blog.acolyer.org/2020/03/20/serverless-in-the-wild/

0 replies, 4 likes


マリウス: Serverless in the Wild: Characterizing and Optimizing the #Serverless Workload at a Large #Cloud Provider https://arxiv.org/abs/2003.03423

0 replies, 2 likes


Daniel Bilar: Characterizing & optimizing production FaaS workloads [Azure 2017-19; 75% of funcs exec time max 10s -> critical to reduce cold starts, 'hybrid histogram policy'; sanitized traces https://github.com/Azure/AzurePublicDataset ; see measuring coldstarts https://twitter.com/daniel_bilar/status/1040534924334383104 ] https://arxiv.org/pdf/2003.03423.pdf https://t.co/KyrduNBxW2

0 replies, 2 likes


beardify: "Serverless in the Wild" (https://arxiv.org/abs/2003.03423) —Trade-offs to provide optimal conditions for #Serverless applications, yet maximing resource usage on the provider side— Referred by https://blog.acolyer.org/2020/03/20/serverless-in-the-wild/ (which an awesome post itself btw) #Cloud

0 replies, 2 likes


Josh Carlisle: Some light reading on #Serverless for the weekend! Seriously though this looks super interesting!

0 replies, 2 likes


arxiv: Serverless in the Wild: Characterizing and Optimizing the Serverless Workload at a Large... http://arxiv.org/abs/2003.03423 https://t.co/1o9uvbl3L1

0 replies, 1 likes


Content

Found on Mar 14 2020 at https://arxiv.org/pdf/2003.03423.pdf

PDF content of a computer science paper: Serverless in the Wild: Characterizing and Optimizing the Serverless Workload at a Large Cloud Provider