Fast partial access to objects from very large files in the SDSC Storage Resource Broker (SRB5) can be extremely challenging, even when those objects are small. The HDF-SRB project integrates the SRB and NCSA Hierarchical Data Format (HDF5), to create an access mechanism within the SRB that is can be orders of magnitude more efficient than current methods for accessing object-based file formats. The project provides interactive and efficient access to datasets or subsets of datasets in large files without bringing entire files into local machines. A new set of data structures and APIs have been implemented to the SRB support such object-level data access. A working prototype of the HDF5-SRB data system has been developed and tested. The SRB support is implemented in HDFView as a client application.

1. Integrating HDF5 with SRB
Peter Cao
The HDF Group
xcao@hdfgroup.org
November 29, 2006
Sponsored by CIP/NLADR, NFS PACI Project
in Support of NCSA-SDSC Collaboration
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
1

2. Imagine
100 Frames x 10 GB
10 GB
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
2

3. Outline
•
•
•
•
Introduction to the project
HDF5-SRB model
Server and client application
Demo if time permits
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
3

4. Motivation
SRB
Distributed data system
Indexing and searching
Access control, etc
HDF5
Large and diverse data
High performance I/O
Subsetting, etc
High performance distributed data system
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
4

5. Goal
A client/server system for object-level
access to HDF5 files stored at SRB
server
• Use SRB as middleware to transfer data between the
server and client
• Use Object-level access for interactive and efficient access
to parts of the file
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
5

6. HDF5 Data Access on SRB
• Normal ways to access SRB:
• Get the whole file: large files (100GB)
• Use POSIX low level calls: low performance
• New way:
• Implement proxy operations to access objects or parts of
objects in one request
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
6

7. Example: Whole File Access
client
I need to see the eye of
Hurricane Bob!
server
Get the file
HDF5
Transfer large file – slow!
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
7

8. Example: Object Level Access
client
I need to see the eye of
Hurricane Bob!
Get me th
e
eye of hu
rric
server
ane Bob
HDF5
1 request, small transfer – fast!
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
8

9. HDF5/SRB Model
client
server
HDF5 Library
HDF5
HDF5 Object
(File, Group, Dataset,
Subset, Attribute)
HDF5 Object
(File, Group, Dataset,
Subset, Attribute)
SRB Server
HDF5-SRB Module
(pack/unpack messages)
02/18/14
HDF5-SRB Module
(pack/unpack messages)
MCAT
HDF and HDF-EOS Workshop X, Landover, MD
9

10. New Data Objects and 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
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
10

11. Running Server/Client
• A SRB server that supports HDF5
• HDF5 library and other external libraries (SZIP, ZLIB)
• A SRB version 3.4.0 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://www.hdfgroup.org/hdf-srb-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
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
11

12. Example
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
12

13. HDFView property file
.hdfview2.3
srbaccount0.domain=demo
srbaccount0.home=/A/home/srbAdmin.demo
srbaccount0.host=kagiso.hdfgroup.uiuc.edu
srbaccount0.password=srbDemo05
srbaccount0.port=5544
srbaccount0.resource=demoResc
srbaccount0.user=srbAdmin
srbaccount1.domain=demo
srbaccount1.home=/A/home/srbAdmin.demo
srbaccount1.host=stanley.ncsa.uiuc.edu
srbaccount1.password=srbdemo05
srbaccount1.port=5544
srbaccount1.resource=demoResc
srbaccount1.user=srbAdmin
02/18/14
HDF and HDF-EOS Workshop X, Landover, MD
13