PARALLEL DATA LAB 

25th Annual Workshop & Retreat
October 23 - 25, 2017

INFO & Agenda

Retreat Agenda


Monday, October 23: CMU/Bedford Springs Resort
8:30 am - 9:00 am Registration & Continental Breakfast
-- SINGLETON ROOM, ROBERTS HALL
9:00 am - 9:30 am Welcome -- SINGLETON ROOM

Greg Ganger - Parallel Data Lab 2016 Retreat Welcome & Overview

9:30 am - 10:30 am Speaker -- SINGLETON ROOM

David Andersen – Indices, Indices, Indices

10:30 am - 11:30 am Industry Poster Session -- ROBERTS HALL ATRIUM
11:30 am - 1:45 pm Bus Ride to Bedford Springs & Lunch -- FREW STREET
2:00 pm - 4:00 pm Session I: Making Database Systems Awesome
-- COLONNADE BALLROOM

Joy Arulraj - What Non-Volatile Memory Means For Database Systems

Dana Van Aken - Automatic Database Management System Tuning Through Large-scale Machine Learning

Ben Blum - Testing Concurrent Transactional Programs

Jinliang Wei - Automatic Parallelization for Distributed Machine Learning Training

4:00 pm - 4:15 pm Break -- CONSTITUTION HALL
4:15 pm - 6:15 pm Session II: Specializing Systems (to RDMA, NVM, ...)
-- COLONNADE BALLROOM

Anuj Kalia - eRPC: High-performance and Usable RPCs for Lossless Kernel-bypass Networks

Rajat Kateja -
Storage management for NVM systems

Andrew Chung -
Tributary: spot-dancing for elastic services with latency SLOs

Kevin Hsieh -
On Machine Learning over Non-I.I.D. Data Partitions

Rashmi Vinayak - Smart redundancy for big data systems: Theory and Practice

6:15 pm - 7:15 pm Dinner -- EISENHOWER BALLROOM
7:15 pm - 8:30 pm Session III: Short Work-in-Progress Talks
-- COLONNADE BALLROOM

Aaron Harlap - Pipe Dream: Pipeline Parallelism for DNN Training

Prashanth Menon -
Relaxed Operator Fusion for In-Memory Databases

Abutalib Aghayev -
Effective Memory Usage for Multitenant Machine Learning Training with Memory-Mapped I/O

Lin Ma -
Query-based Workload Forecasting for Self-Driving Database Management Systems

Rajat Kateja -
Viyojit: Decoupling Battery and DRAM Capacities for Battery-Backed DRAM

8:30 pm - 10:30 pm Reception/Poster Session I -- COLONNADE BALLROOM

Tuesday, October 24: Bedford Springs Resort

8:15 am - 9:00 am Full Breakfast -- CRYSTAL DINING ROOM
Continental Breakfast
-- CONSTITUTION HALL
9:00 am - 10:30 am Session IV: Orchestration of Video Stream Processing
-- COLONNADE BALLROOM

Chris Canel - FilterForward: Enabling Publish-Subscribe for Streaming Video Analytics​

Angela Jiang -
Mainstream: Adaptive compute sharing for video analysis

10:30 am - 10:45 am Break -- CONSTITUTION HALL
10:45 am - 12:15 pm Session V: Toward NVM and Flash Efficiency
-- COLONNADE BALLROOM

Yixin Luo - Mitigating Data Retention and Process Variation Errors in 3D NAND Based SSDs

Charlie McGuffey -
Caching Policies to Save Energy and Bandwidth on NVM

12:15 pm - 1:15 pm Lunch -- EISENHOWER BALLROOM
1:15 pm - 2:30 pm Socializing and Chat-filled Walk -- HOTEL GROUNDS
Wii games room available 1:00 pm to midnight --
REAGAN-B
2:30 pm - 4:00 pm Session VI: Cluster Workloads and Scheduling
-- COLONNADE BALLROOM

George Amvrosiadis - Bigger, Longer, Fewer: What do cluster jobs look like outside Google?

Junwoo Park -
3Sigma: Distribution-Based Cluster Scheduler for Runtime Uncertainty

4:00 pm - 4:15 pm Break -- CONSTITUTION HALL
4:15 pm - 6:15 pm Session VII: File Systems in the Large
-- COLONNADE BALLROOM

Saurabh Kadekodi - A Case for Packing and Indexing in Cloud File Systems

Qing Zheng -
DeltaFS PLDB: Light-weight in-situ analysis for faster HPC queries

Greg Ganger, et al. -
PDL Turns 25

6:15 pm - 7:30 pm Dinner -- EISENHOWER BALLROOM
7:30 pm - 8:30 pm Session VIII: Short Work-In-Progress Talks
-- COLONNADE BALLROOM

Saugata Ghose - Vulnerabilities and Security Exploits in MLC NAND Flash Memory Programming

Daniel Berger -
AdaptSize: Orchestrating the Hot Object Memory Cache in a CDN

Gauri Joshi -
Redundancy Techniques in Cloud and ML Infrastructure

Nathan Beckmann -
LHD: Improving Cache Hit Rate by Maximizing Hit Density

8:30 pm - 10:00 pm Reception/Poster Session II -- COLONNADE BALLROOM

Wednesday, October 25: Bedford Springs Resort
Morning Workshop Talks
8:45am - 9:15am Full Breakfast -- CRYSTAL DINING ROOM
Continental Breakfast
-- CONSTITUTION HALL
9:15am - 11:30am Session IX: Caching and Automating Efficient Parallelization
-- COLONNADE BALLROOM

Conglong Li - Workload Analysis and Caching Strategies for Search Advertising Systems

Jin Kyu Kim -
STRADS+: Simplifying Parallel ML Programming

Chris Fallin -
Finding Hidden Parallelism: a Data-Structure-Aware Analysis for Auto-Parallelization

Afternoon Feedback, Departure
11:30am - 1:15pm Lunch and Industry Feedback
1:30pm Buses depart for Pittsburgh Airport & CMU (one in each direction)
4:00pm Arrive at Airport
4:00pm
Arrive at CMU


CONTACTS:


    Director, Parallel Data Lab
    VOICE: (412) 268-1297


    Executive Director, Parallel Data Lab
    VOICE: (412) 268-5485


    PDL Administrative Manager
    VOICE: (412) 268-6716