Transcript
Page 1: THE REPLICA LOCATION SERVICE The Globus Project and the DataGrid Project

THE REPLICA LOCATION SERVICEThe Globus Project and the DataGrid Project

Consistent state maintained in Local Replica Catalogs (LRCs)• Mappings between logical names (LFNs) and target names

Collective state with relaxed consistency in Replica Location Indices (RLIs)• Each RLI contains mappings from LFNs to LRCs• Variety of index structures can be created by varying number of RLIs, redundancy and partitioning

Soft state maintenance of RLI state• LRCs send state information to RLIs using soft state protocols• Information in RLIs times out and must be periodically refreshed

Optional compression of soft state updates• Prototype implements Bloom filter compression

Membership service • Keeps track of LRCs and RLIs and their soft state update patterns• Prototype implements static configuration for RLS

Five Components of the RLS Architecture Framework

Front-End Server • Multi-threaded• Supports GSI Authentication• Common implementation for LRC and RLI

Back-end Server• mySQL Relational Database • Holds logical name to target name mappings

Client APIs: C and Java

RLS Components

The RLS Implementation

LRC LRC LRC LRC

RLI RLI RLI

Replica Location Index Nodes

Local Replica Catalogs

The RLS SC2002 Testbed

The Replica Location Service Architecture

Replication in Data Grids

RLS Designers and Developers RLS Sponsors and Testbed ParticipantsGlobus Project: www.globus.org/rls

Shishir Bharathi, Ann Chervenak, Ewa Deelman, Ian Foster, Carl Kesselman, Matei Ripeanu, Bob Schwartzkopf, Mei-Hui Su

DataGrid Project: http://cern.ch/grid-data-management

Leanne Guy, Peter Kunszt, Heinz Stockinger, Kurt Stockinger

Two types of soft state updates from LRCs to RLIs• Complete list of logical names registered in LRC• Bloom filter summaries of LRC

Bloom filter advantages • Reduce size of soft state updates• Reduce associated memory and network requirements• Sending updates is faster and scales better with size of LRC

User-defined attributes • May be associated with logical or target names

Partitioning• Divide LRC soft state updates among RLI index nodes using pattern matching of logical names

Redundancy• Send soft state updates to multiple RLIs

Implementation Features

Large data sets are frequently replicated to:• Reduce access latency• Maintain local control over necessary data• Improve reliability and load balancing

A Replica Location Service is a distributed registry service that records the locations of data copies and allows discovery of replicas

possum

emu

wombat

koala

Melbourne

RLIs

LRCs

n16

n19

n17

n18

dc-n1

dc-n4

dc-n2

dc-n3

a33

a36

a34

a35

rls01

rls02

rls02

rls01

ANL(Chicago)

ISI(Los Angeles)

SC2002(Baltimore)

SLAC(Palo Alto)

Replica Location Index Nodes

Local Replica Catalogs

0342

pcr25

0343

pcr24

0344

grid03

0345

grid01

0346 grid8 grid6

grid7.mi

grid1

grid7.pi

CERN(Geneva)

Glasgow INFN(Pisa)

INFN(Milan)

Replica Location Index Nodes

Local Replica Catalogs

Australia Sites Unites States Sites Europe Sites

DB

LRC/RLI Server

ODBC (libiodbc)

myodbc

mySQL Server

clientclient

Top Related