February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS...

26
February 2-3, 2006 SRB Workshop, San Diego Peter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level Access to Remote Files Integrating HDF5 with SRB

Transcript of February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS...

Page 1: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego

Peter Cao, NCSA

Mike Wan, SDSC

Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration

Object-level Access to Remote Files

Integrating HDF5 with SRB

Page 2: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 2/26

Outline

Introduction to HDF5

The HDF-SRB model

SRB Support in HDFView

Page 3: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 3/26

Overview of HDF5 Answering big

questions …

Matter & universe

Weather & climate

August 24, 2001 August 24, 2002Total Column Ozone (Dobson)

60 385 610

Life & nature

Page 4: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 4/26

Overview of HDF5 Involves big

data …

Page 5: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 5/26

Overview of HDF5 On big

computers …

Page 6: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 6/26

Overview of HDF5 HDF solution

Software & tools

open source & multiple platform

Commonmodels

extensions Standard

APIsconventions & easy use

File formatfor all kinds of data

Efficiencystorage & IO

Page 7: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 7/26

Overview of HDF5 Exmaple HDF5

Page 8: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 8/26

Overview of HDF5 HDF Software

HDF I/O Library

Tools & ApplicationsTools & Applications

HDF FileHDF File

Page 9: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 9/26

Overview of HDF5 Object

model

Primary ObjectsGroupsDatasets

Additional ways to organize dataAttributesSharable objectsStorage and access properties

Page 10: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 10/26

Overview of HDF5 Groups

“/”

tom dick harry

temp

A mechanism for collections of related objectsEvery file starts with a root groupSimilar to UNIXdirectoriesCan have attributes

Page 11: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 11/26

Overview of HDF5

Datasets

DataMetadataDataspace

3

RankRank

Dim_2 = 5Dim_1 = 4

DimensionsDimensions

time = 32.4

pressure = 987

temp = 56

AttributesAttributes

Chunked

compressed

Dim_3 = 7

Storage infoStorage info

IEEE 32-bit float

DatatypeDatatype

Page 12: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 12/26

Overview of HDF5 Data

subsetting

(c) A sequence of points from a 2D array to a sequence of points in a 3D array.

(d) Union of hyperslabs in file to union of hyperslabs in memory.

(b) Regular series of blocks from a 2D array to a contiguous sequence at a certain offset in a 1D array

(a) Hyperslab from a 2D array to the corner of a smaller 2D array

Page 13: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 13/26

Project Description

Motivation

SRB HDF5

Indexing and searching Distributed data system Access control

Large and diverse data High performance access Interactive and subsetting

High performance distributed data system

Page 14: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 14/26

Project Description Goals

Working prototype of client/server system forobject-level access to HDF5 stored in the SRB

Use SRB as middleware to transfer data between the server and client Use Object-level access for interactive and efficient access to part of the file

Page 15: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 15/26

Remote Data Access on SRB

Methods

Normal ways to access SRB:Get the whole file: large files (100TB

SCEC)Use POSIX low level calls: low

performance

New way: Implement proxy operations to access

objects or parts of objects in one request

Page 16: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 16/26

Normal SRB File Access Architecture

SRB ServerSRB Server

HDF5HDF5

MCAT

clientclient

HDF5 File(whole file or a

sequence of bytes)

Page 17: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 17/26

Object-level File Access Architecture

SRB ServerSRB Server

MCAT

HDF5 LibraryHDF5 Library

HDF5-SRB Module(pack/unpack messages)

HDF5 Object(File, Group, Dataset,

Subset, Attribute)

HDF5-SRB Module(pack/unpack messages)

ClientClient ServerServer

HDF5 Object(File, Group, Dataset,

Subset, Attribute)

Client Application HDF5 file

Page 18: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 18/26

Examples of File Access

HDF5HDF5

I need to see the eye of Hurricane Bob!

Page 19: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 19/26

Examples of File Access Whole file

transfer

clientclient

Get the file

Transfer large image – slow!

HDF5HDF5

Page 20: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 20/26

Examples of File Access SRB POSIX

API

HDF5HDF5

clientclient

image foundimage open

open image

find imagefile’s open

Open file

Many small messages – slow and complex!

Page 21: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 21/26

Examples of File Access Object

level

clientclient

HDF5HDF5

Get me the eye of hurricane Bob

1 request, small transfer – fast!

Page 22: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 22/26

HDF5-SRB Model New

objects/APIs

A new set data objects H5File, H5Group, H5Dataset, H5Datatype, etc Encapsulated client requests and server results

Enhanced SRB APIs Pack/Unpack routines (exchange data between

byte stream and structure) to handle complicated struct – string, pointers, pointers to arrays, arrays of pointers, etc

New srbGenProxyFunct (general Proxy Function) handles other types of request besides HDF5

Page 23: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 23/26

HDF5-SRB Model Data Flow

Client APIsrbObjRequest(void *obj, int objID)

Server APIsrbObjProcess(void *obj, int objID)

srbGenProxyFunct

1. packMsg()

2. u

npac

kMsg

()

HDF5 Library

HDF5 file

3. H5Obj::op()

4. Access file

5. H5Object

6. p

ackM

sg()

7. unpackMsg()

SRB Server

Page 24: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 24/26

Running Server/Client

A SRB server that supports HDF5 HDF5 library and other external libraries (SZIP, ZLIB) A SRB version 3.4 or later from

http://www.sdsc.edu/srb/ Follow instruction on how to run SRB server from UG

packed with SRB source release or online at http://hdf.ncsa.uiuc.edu/hdf-srb-html/HDF-SRB-UG.html

Any client application that implements HDF5-SRB Objects No HDF5 library is required on the client Example client application: HDFView 2.3 or above

Page 25: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 25/26

Short Demo HDFViewSupport Windows and Linux

Page 26: February 2-3, 2006SRB Workshop, San Diego P eter Cao, NCSA Mike Wan, SDSC Sponsored by NLADR, NFS PACI Project in Support of NCSA-SDSC Collaboration Object-level.

February 2-3, 2006 SRB Workshop, San Diego 26/26

Question / Comments?