Exploring Scalable Storage: DSpace

1
ExploringScalable Storage DSpace SRB

• Chris Frymann
• University of California San Diego Libraries
• DSpace User Group Meeting
• March 10, 2004

2
Extending DSpaceStorage Capabilities

• Much of the value and success of the Web is a
  result of its enormous size, which has been
  achieved through a distributed storage model.
• The current DSpace model assumes local storage.
• What if DSpace collections could be of virutally
  unlimited size, stored, replicated and accessible
  via federated grid technologies?

3
This Presentation Will

• Report on a proposed project to extend DSpace
  data management capabilites through integration
  with the San Diego Supercomputer Centers Storage
  Resource Broker (SRB)
• Provide an overview of SRB
• Review how both DSpace and SRB users will benefit

4
NARA Proposal Participants

• San Diego Super Computer Center (SDSC)
• Member of National Partnership for Advanced
  Computational Infrastructure (NPACI) an NSF
  sponsored program
• MIT Libraries
• UC San Diego Libraries (UCSD)
• Hewlett Packard Laboratories (HP)
• National Archives and Records Administration
  (NARA)

5
Proposal Views DSpace As

• Simple user-friendly front end providing
• Digital content ingestion
• Search and discovery
• Content management
• Dissemination services
• Preservation

6
What is SRB?

• Storage Resource Broker
• Data management infrastructure
• Developed at San Diego Supercomputer Center
• Utilizes data grid and federation technologies

7
Levels of Possible Integration

• Replace DSpace file system calls with SRB access
  calls
• Utilize SRB metadata management capabilities
• Provide support for federation of DSpace systems

8
Storage Resource BrokerTechnical Overview
9
Definition of SRB

• Middleware which allows other applications to
  treat a diverse collection of physical storage
  devices as a single logical resource
• A distributed file system (Data Grid), based on a
  client-server architecture.

10
What SRB Does

• Replicates, syncs, archives, and connects
  heterogeneous resources in a logical manner using
  abstraction mechanisms.
• Also provides a way to access files and computers
  based on their attributes rather than just their
  names or physical locations.

11
A Sample Storage Environment
12
Commodity IDE Disk Drive
200 GB - 200
13
Grid Brick
2 Terabytes
14
Grid Bricks Details

• Hardware components
• Intel Celeron 1.7 GHz CPU
• SuperMicro P4SGA PCI Local bus ATX mainboard
• 1 GB memory (266 MHz DDR DRAM)
• 3Ware Escalade 7500-12 port PCI bus IDE RAID
• 10 Western Digital Caviar 200-GB IDE disk drives
• 3Com Etherlink 3C996B-T PCI bus 1000Base-T
• Redstone RMC-4F2-7 4U ten bay ATX chassis
• Linux operating system
• Cost is 2,200 per Tbyte plus tax
• Gig-E network switch costs 500 per brick
• Effective cost is about 2,700 per TByte

15
Rack of Grid Bricks 12 TB
16
Data Grid 50 TB
Room of Racks
17
Grid Bricks at SDSC

• Used to implement picking environments for
  10-TB collections
• Web-based access
• Web services (WSDL/SOAP) for data subsetting
• Implemented 15-TBs of storage
• Astronomy sky surveys, NARA prototype persistent
  archive, NSDL web crawls
• Must still apply Linux security patches to each
  Grid Brick

18
SDSC Production Data Grid

• SDSC Storage Resource Broker
• Federated client-server system, managing
• Over 90 TBs of data at SDSC
• Over 16 million files
• Manages data collections stored in
• Archives (HPSS, UniTree, ADSM, DMF)
• Hierarchical Resource Managers
• Tapes, tape robots
• File systems (Unix, Linux, Mac OS X, Windows)
• FTP sites
• Databases (Oracle, DB2, Postgres, SQLserver,
  Sybase, Informix)
• Virtual Object Ring Buffers

19
SRB Collections at SDSC
20
Data Grids

• Distributed data sources
• Inter-realm authentication and authorization
• Heterogeneity
• Storage repository abstraction
• Scalability
• Differentiation between context and content
  management
• Preservation
• Support for automated processing (migration,
  archival processes)

21
Data Grid Components

• Federated client-server architecture
• Servers can talk to each other independently of
  the client
• Infrastructure independent naming
• Logical names for users, resources, files,
  applications
• Collective ownership of data
• Collection-owned data, with infrastructure
  independent access control lists
• Context management
• Record state information in a metadata catalog
  from data grid services such as replication
• Abstractions for dealing with heterogeneity

22
Data Grid Abstractions

• Logical name space for files
• Global persistent identifier
• Storage repository abstraction
• Standard operations supported on storage systems
• Information repository abstraction
• Standard operations to manage collections in
  databases
• Access abstraction
• Standard interface to support alternate APIs
• Latency management mechanisms
• Aggregation, parallel I/O, replication, caching
• Security interoperability
• GSSAPI, inter-realm authentication,
  collection-based authorization

23
Data Grid Federation

• Data grids provide the ability to name, organize,
  and manage data on distributed storage resources
• Federation provides a way to name, organize, and
  manage data on multiple data grids.

24
Distributed Data Grids 200 TB
SRB Federation
25
SRB APIs
26
SDSC Storage Resource Broker Meta-data Catalog
Application
Linux I/O
OAI WSDL
Access APIs
DLL / Python
Java, NT Browsers
GridFTP
Federation
Consistency Management /
Authorization-Authentication
SRB Server
Logical Name Space
Latency Management
Data Transport
Metadata Transport
Storage Abstraction
Catalog Abstraction
Databases DB2, Oracle, Sybase, SQLServer
Drivers
HRM
27
Federated SRB server model
Peer-to-peer Brokering
Read Application
Parallel Data Access
Logical Name Or Attribute Condition
1
6
5/6
SRB server
SRB server
3
4
5
SRB agent
SRB agent
2
Server(s) Spawning
R1
MCAT
1.Logical-to-Physical mapping 2.Identification of
Replicas 3.Access Audit Control
R2
Data Access
28
Peer-to-Peer Federation

• Occasional Interchange - for specified users
• Replicated Catalogs - entire state information
  replication
• Resource Interaction - data replication
• Replicated Data Zones - no user interactions
  between zones
• Master-Slave Zones - slaves replicate data from
  master zone
• Snow-Flake Zones - hierarchy of data
  replication zones
• User / Data Replica Zones - user access from
  remote to host zone
• Nomadic Zones - synchronize local zone to
  parent zone
• Free-floating myZone - synchronize without a
  parent zone
• Archival BackUp Zone - synchronize to an archive

29
Principle peer-to-peer federation approaches
30
SRB Availability

• SRB source distributed to academic and research
  institutions
• Commercial use access through UCSD Technology
  Transfer Office
• William Decker WJDecker_at_ucsd.edu
• Commercial version from
• http//www.nirvanastorage.com

31
SRB Info Resources

• SRB Homepage
• http//www.npaci.edu/DICE/SRB/
• Grid Port Toolkit
• https//gridport.npaci.edu/
• Data Intensive Computing Environment (DICE)
• http//www.npaci.edu/DICE
• mySRB - Web-based Browser and Query Tool for
  Storage Resource Broker
• http//www.npaci.edu/dice/srb/mySRB/mySRB.html

32
NARA Proposal - Two Goals

• Demonstrate possibility of federating two
  different preservation architectures
• Support exchange of documents between both
  systems
• Demonstrate DSpace/SRB integration leads to
  improved life-cycle support

33
NARA Proposal - Plan of Work

• Evaluation of life-cycle management
• Use SRB as filestore for DSpace bitstreams
• Identify METS storage profile
• Enable exchange of data and metadata between
  independent DSpace and SRB systems

34
Schedule Deliverables

• Year 1 Develop Research Prototype
• Develop functional requirements
• Specify standard interfaces METS profiles
• Prototype implementation of specified design
• Ingest data, evaluate functionality and
  performance

35
Schedule Deliverables

• Year 2
• Federation with additional systems, possibly
• CDL
• OCLC
• Fedora
• Scalability testing
• Ingestion of more content types

36
UCSD LibrariesTo ProvideTest Collections

• Slide Collection
• Over 200,000 Digitized Slide Images
• Over 1 million files (counting derivatives)
• Approximately 5 Terabytes
• Moving Image Collection
• California movie newsreel footage
• Size of collection to be determined

37
Testing Will Explore

• Management of terabyte scale collections
• Handling of compound documents
• Automating aspects of archival workflow
• Integration of METS
• Feasibility of deriving descriptive metadata from
  the material during ingest
• Automated verification and validation checking

38
Hewlett Packard Labs

• Will provide a second storage utility which will
  be used to test
• Federation
• Zone level access and management controls
• Data validation and authenticity

39
Data Model

• Paired Content and Metadata Files
• Metadata encoded in standard METS profiles
• Stand-alone METS files used to describe arbitrary
  levels of aggregation of lower level objects

40
(No Transcript)
41
Integration of DSpace / SRB EnablesMultiple
Modes of Control

• From DSpace
• Via SRB APIs
• Specification of Federated SRB Zone configuration
  and interoperability

42
ConcusionExpected Results / Benefits

• DSpace users
• Federation collection management through
  distributed grid technology
• Exchange of METS encoded collections
• SRB users
• User friendly ingest mechanism
• Extended life-cycle management
• Exchange of METS encode collections

43
Q A

• Presentation will be available at
• http//libnet.ucsd.edu/dspace/user_group_2004.03.
  ppt