NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE...

8
Prabhat Data & Analytics Services Group January 25, 2019 NERSC Data Ecosystem Overview

Transcript of NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE...

Page 1: NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE COMPUTING DEPARTMENT RICHARD GERBER Department Head ADVANCED TECHNOLOGIES NICHOLAS

PrabhatData & Analytics Services Group

January 25, 2019

NERSC Data Ecosystem Overview

Page 2: NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE COMPUTING DEPARTMENT RICHARD GERBER Department Head ADVANCED TECHNOLOGIES NICHOLAS

Afternoon Schedule1:40 pm Data Transfer Shreyas Cholia

2:10 pm File Systems + Burst Buffer Wahid Bhimji

2:30 pm I/O Best Practices Quincey Koziol

2:50 pm Break

3:10 pm Python and Jupyter Rollin Thomas

3:30 pm Machine Learning Mustafa Mustafa

3:50 pm Shifter Shane Canon

4:10 pm End

Page 3: NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE COMPUTING DEPARTMENT RICHARD GERBER Department Head ADVANCED TECHNOLOGIES NICHOLAS
Page 4: NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE COMPUTING DEPARTMENT RICHARD GERBER Department Head ADVANCED TECHNOLOGIES NICHOLAS

Phase I: 2388 x 32-core Intel Xeon “Haswell” 128 GB DDR4Phase II: 9688 x 68-core Intel Xeon Phi “KNL” 96 GB DDR4 + 16 GB MCDRAM

Gerty Cori: Biochemist and first American woman to win a Nobel Prize in science

Cori Brings HPC & Data Together

Page 5: NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE COMPUTING DEPARTMENT RICHARD GERBER Department Head ADVANCED TECHNOLOGIES NICHOLAS

Production Data StackCapabilities Technologies

Data Transfer + Access

Workflows

Data Management

Data Analytics

Data Visualization

TaskFarmer

Page 6: NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE COMPUTING DEPARTMENT RICHARD GERBER Department Head ADVANCED TECHNOLOGIES NICHOLAS

Cori’s Data-Friendly Features

Cray DataWarp:Burst Buffer forI/O acceleration

12 32-core Haswell500 GB Login Nodes

768 GB “bigmem”Haswell Nodes

Pipeline/WorkflowManagement Nodes

JupyterNotebook

Node

Serial QueueShared-node Queue

Transfer Queue

Real-Time Queues for Co-Scheduling

w/Experiments

Interactive Queue:64 Nodes x 4 Hours

ContainerizedEnvironments

External Network Access to/fromCompute Nodes

Streaming Data to Compute Nodes

Page 7: NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE COMPUTING DEPARTMENT RICHARD GERBER Department Head ADVANCED TECHNOLOGIES NICHOLAS

Asks

● Please engage with NERSC staff members○ Provide feedback, critique○ Tell us about interesting science problems!

Page 8: NERSC Data Ecosystem Overview Data & Analytics Services Group · 1/25/2019  · HIGH PERFORMANCE COMPUTING DEPARTMENT RICHARD GERBER Department Head ADVANCED TECHNOLOGIES NICHOLAS

Questions? Comments?