Data Replication Service

1
Data Replication Service

• Sandeep Chandra
• GEON Systems Group
• San Diego Supercomputer Center

2
Outline

• Motivation
• Data Replication Service (DRS)
• Components for DRS
• RLS, GridFTP, RFT
• DRS Deployment
• DRS setup on GEON
• Next Steps

3
Motivation

• Science domains spend considerable effort
  collecting and managing large amounts of data
• Science domains develop customized data
  management services that vary with the type of
  application
• Common data management requirements
• Publish and replicate large datasets
• Register data replicas in catalogs and discover
  them
• Perform metadata-based discovery of datasets
• May require ability to validate correctness of
  replicas

4
Motivation (cont.)

• These systems demand considerable resources to
  design, implement maintain
• Typically cannot be re-used by other applications
• Need for a long-term solution
• Generalize functionality provided by these data
  management systems
• Provide suite of application-independent services
• Design and build on lower-level grid services
• Globus Reliable File Transfer (RFT) service
• Replica Location Service (RLS)
• GridFTP

5
A possible solutionData Replication System (DRS)

• Higher level data management service based on low
  level data management components like RLS and RFT
• The primary functionality is to
• Allow users to identify a set of desired files
  existing in their grid environment
• Make local replicas of those data files by
  transferring files from one or more source
  locations
• Register the new replicas in a Replica Location
  Service

6
Replica Location Service (RLS)

• A simple registry that keeps track of where
  replicas exist on physical storage systems.
• Users or services register files in RLS when the
  files are created.
• Query RLS servers to find these replicas.
• RLS can be a distributed registry, consisting of
  multiple servers at different sites.
• Distributed RLS increases the overall scale and
  store more mappings than would be possible in a
  single, centralized catalog.

7
RLS (cont.)

• A logical file name is a unique identifier for
  the contents of a file.
• A physical file name is the location of a copy of
  the file on a storage system.
• RLS maintains mappings between logical file names
  and one or more physical file names of replicas.
• Users can provide a logical file name to an RLS
  server and ask for all the registered physical
  file names of replicas.
• Users can also query an RLS server to find the
  logical file name associated with a particular
  physical file location.

Logical File Name XYZ
XYZ replica 1
XYZ replica 2
XYZ replica 3
Site 3
Site 1
Site 2
8
RLS (cont.)

• Two servers LRI, LRC
• LRC stores mappings between logical names for
  data items and the physical locations of
  replicas.
• Query the LRC to discover replicas associated
  with a logical name.
• RLI server collects information about the logical
  name mappings stored in one or more LRCs.
• RLI returns a list of all the LRCs it is aware of
  that contain mappings for the logical name
  contained in a query.
• The client then queries these LRCs to find the
  physical locations of replicas.

Replica Location Index (RLI) Nodes
RLI
RLI
RLI
LRC
LRC
LRC
LRC
Local Replica Catalogs (LRC)
9
RLS in Context

• The RLS is one component in a layered data
  management architecture
• Consistency management provided by higher-level
  services

Replica Consistency Management Services
Metadata Service
Reliable Replication Service
Replica Location Service
Reliable Data Transfer Service
GridFTP
10
GridFTP

• The GridFTP protocol provides for the secure,
  robust, fast and efficient transfer of
  (especially bulk) data.
• Globus Toolkit provides the most commonly used
  implementation of the protocol, though others
  exist.
• The Globus Toolkit provides
• server implementation called globus-gridftp-server
  
• scriptable command line client called
  globus-url-copy
• a set of development libraries for custom clients

11
Reliable File Transfer (RFT)

• A WSRF compliant web service that provides job
  scheduler like functionality for data movement.
• You provide a list of source and destination URLs
  (including directories or files), then the
  service writes your job description into a
  database and moves the files on your behalf.

12
RFT (cont.)

• Accepts SOAP description of a desired transfer
• Service methods are provided for querying the
  transfer status
• WSRF tools to subscribe for notifications of
  state change events
• Supports all the same options as globus-url-copy
  (buffer size, etc)
• Increased reliability because state is stored in
  a database
• Supports concurrency, multiple files transferred
  for better performance

13
Globus Services

• WSRF Services
• Data Replication Service
• Delegation Service
• Reliable File Transfer Service
• Pre WSRF Components
• Replica Location Service (Local Replica Catalog,
  Replica Location Index)
• GridFTP Server

Local Site
Reliable
Data
File
Delegation
Replication
Transfer
Service
Service
Service
Replicator
RFT
Delegated
Resource
Resource
Credential
Web Service Container
Local
Replica
GridFTP
Replica
Location
Server
Catalog
Index
14
DRS Deployment

• Local storage system
• GridFTP server for file transfer
• Replica Location Service
• LRCs stores mappings from logical names to
  storage locations
• RLI collects state summaries from LRCs
• RFT WSRF service to perform data transfer
• DRS The master replication service

Create a Transfer request
RFT Service
DRS Service
Replica Location Index
Location Replica Catalog
GridFTP
Server
Database
Site Storage System

15
Local Site
Client
1
3
Reliable
Data
File
Delegation
2
Replication
Transfer
Service
9
Service
Service
Request File
5
4
Replicator
RFT
Delegated
8
Resource
Resource
Credential
12
6
10
Web Service Container
Replica
Local
GridFTP
Location
Replica
13
Server
Index
Catalog
7
Remote Sites 1N
11
Reliable
Data
File
Delegation
Replication
Transfer
Service
Service
Service
Replicator
RFT
Delegated
Resource
Resource
Credential
Web Service Container
Local
Replica
GridFTP
Replica
Location
Server
Catalog
Index
16
DRS Functionality

• Initiate a DRS Request
• Create a delegated credential (Delegate
  Authority)
• Create a Replicator resource (Replication
  Service)
• Monitor Replicator resource (Status)
• Discover replicas of files in RLS, select among
  replicas
• Start data transfer to local site with RFT
  service
• Check status
• Register new replicas in RLS catalogs
• Allow client inspection of DRS results
• Destroy Replicator resource

17
Geon DRS Test Setup
ASU
SDSC
Globus Container
Globus Container
Create a Transfer request
Create a Transfer request
RFT Service
RFT Service
DRS Service
DRS Service
Replica Location Index
Replica Location Catalog
Replica Location Index
Replica Location Catalog
GridFTP
GridFTP
Server
Server
Database
Site Storage System
Database
Site Storage System


Data Transfer
18
Next Tasks

• Transfer LIDAR data from ASU to SDSC resource.
  (HPSS, etc)
• Extend the testbed to include more nodes.
• Benchmarking data movement.
• Package DRS and components with GEON software
  stack version 2.0

19
Acknowledgement

• Ann Chervenak Robert Schuler (ISI)
• www.globus.org (slides)

20
Questions?