SRB Campus Grid Presentation

1
An Introduction to the SDSCStorage Resource
BrokerNIEeS SRB WorkshopJanuary 2006 Peter
Berrisford
2
What is SRB?

• Data Grid middleware developed at theSan Diego
  Supercomputer Center (SDSC)
• A distributed file system, based on a
  client-server architecture.
• Allows users to access files seamlessly across a
  distributed environment, based upon their
  attributes rather than just their names or
  physical locations.
• It replicates, syncs and archives data,
  connecting heterogeneous resources in a logical
  and abstracted manner.

3
Why use SRB?

• Users need to know their data is
• Collected and stored safely
• Easily accessible, from anywhere
• By the right people
• Reasonable response times
• Metadata to assist searching
• Available 24/7, this year and next and next
• Dont need to know how need to be freed from IT
  concerns to focus on their research
• Available for collaboration

4
Data Virtualization with SRB
User Application
Archive at RAL
Database At UCL
File System at Daresbury
5
Data Virtualization with SRB
Logical name space Location independent
identifier Persistent identifier Collection
owned data Access controls Audit trails
Checksums Descriptive metadata Inter-realm
authentication Single sign-on system
User Application
Common naming convention and set of attributes
for describing digital entities
Archive at RAL
Database At UCL
File System at Daresbury
6
The Application of SRB
7
Archival Services
Archival Services across single or distributed
sites
Archival Services operate on economy of scale and
require expert staff to operate them, thus
central services for larger Grids make financial
sense.
CCLRC operate these services both on site and
for external partners.
This example is for a customer with about 16
sites across the UK. They operate on their own
network, with only their main site connected to
Janet. Scheduled archives and restores are
handled via this central site.
8
Archival Process Data Transfer
Remote Institute Site
Request Tracker database used to control data
transfer process
Local Storage
Filer
Central Cache Site
RAL
Firewall
JANET WAN
Site WAN
Central SRB Server
Local machines
Local SRB Server
Tape Traffic
SRB-ADS Server
ADS Tape Resource
disk
ADS SRB Disk Cache Resource
disk
Central cache Vault
disk
Local Vault
Firewall
Note firewalls
Firewall
9
Request Tracker

• Tracks the movement of archive packages using a
  database application
• Monitors archive, restore and extract utilities
• Provides error handling, reporting and request
  priority intelligence
• Interacts with front-end application, server
  scripts and CCLRC Atlas Data Store
• Takes advantage of Oracle features
• Scripts call stored procedures

10
DIAMOND Architecture
Beam line control system and Detector
ADSC
ADSC
ADSC

RAID 2TB
RAID 2TB
RAID 2TB
SRB space
20TB disc store, some fraction allocated as SRB
Vault
20TB disc store, some fraction allocated as SRB
Vault
20TB disc store, some fraction allocated as SRB
Vault
20TB disc store, some fraction allocated as SRB
Vault

160TB SRB Vault
Background transfer of data to ADS
Replicas maintained across resources
Logical view of data can be maintained
ADS Resource
11
Experimental results distribution
ISIS Facility
Hole opened in firewall
SRB Server (no vault)
ADS-SRB Server
Atlas Drivers
ADS
12
DLS Zone
ISIS Zone
13
SRB Implementation
14
How is SRB implemented?

• The logical file system structure is maintained
  in a database
• MCAT Metadata Catalogue
• All metadata (system and application) stored in
  the MCAT
• Files stored in distributed physical resources
  (vaults)
• Logical resources can be created, mapping to
  multiple physical resources

15
SRB The Key Components

• The Metadata Catalogue (MCAT)
• The MCAT-EnabledSRB Server
• The SRB Storage Server
• The SRB Client

16
The MCAT Database

• The MCAT database is a metadata repository that
  provides a mechanism for storing information used
  by the SRB system.
• Includes both
• Internal system data required for running the
  system
• Application (user) metadata regarding data sets
  being brokered by SRB.

17
The MCAT Server

• At least one SRB Server must be installed on the
  node that can access the MCAT database. This is
  known as the MCAT-Enabled Server (MES)
• The MES works directly against the MCAT database
  to provide SRB Services
• All other SRB Servers interact through the MCAT
  Server

18
The SRB Server

• The SRB Server is a middleware application that
  accepts requests from clients and
  obtains/queries/manages the necessary data sets.
• It queries the MCAT SRB Server to gather
  information on datasets and supplies this back to
  the SRB client.

19
SRB Client Tools

• Provide a user interface to send requests to the
  SRB server.
• 4 main interfaces
• Command line (S-Commands)
• MS Windows (InQ)
• Web based (MySRB).
• Java (JARGON)
• Web Services (MATRIX)

20
Logical Resources
SRB Server
instant replication
fast archival
resource pooling
My-Logical-Resource
SRB Server
21
SRB Containers - Archival of Small files

• Performance issues with storing / retrieving a
  large number of small files to / from tape
• Container design
• Physical grouping of small files
• Implemented with a Logical Resource
• A pool of Cache Resource for the front-end
  resource
• An Archival Resource for the backend resource
• Read/Write I/O always done on Cache Resource and
  sync to the Archival Resource
• The entire container is moved between the cache
  and the archival resource and written to tape
• Bulk operation with container - faster

22
Logical Resource for Containers
Effective use of SRB The ADS Interface
ADS-logical-resource
ADS-cache resource
ADS-tape resource
Ssyncont
Copies container from cache to tape
Sput c ltcontainer-namegt ltdata-filegt
23
SRB ADS Server
Port 8600
SRB-ISIS server instance
Port 8610
SRB-BADC server instance
Port 8620
SRB-BBSRC server instance
SRB-ADS architecture
24
SRB Services Architecture
App
App
App
App
App
ADS-SRB Multiple Servers
Web Server
MES
MES
MES
MES
MCAT Server
Oracle Client
DB-Instance-1
DB-Instance-2
Oracle RAC Database Server
MCAT Database
Schema1
Schema2
Schema3
Schema4
25
SRB APIs

• Shell commands (Scommands)
• Provide access to all SRB functions
• mySRB web browser
• Provides hierarchical collection view
• inQ Windows browser
• Provides Windows style directory view
• C library calls
• Provide access to all SRB functions
• Jargon Java API
• Similar to java.io. API
• Matrix WSDL/SOAP Interface
• Aggregate SRB requests into a SOAP request. Has a
  Java API and GUI
• Python, Perl, C, OAI, Windows DLL, Mac DLL,
  Linux I/O redirection, GridFTP

26
Terminology

• Location A physical node running an SRB Server
• Physical Resource A storage area managed by an
  SRB Server
• Logical Resource One or more Physical Resources
  can be distributed
• Collection Data abstraction of resources