270 likes | 393 Views
JuxMem: An Adaptive Supportive Platform for Data Sharing on the Grid. Gabriel Antoniu, Luc Bougé, Mathieu Jan IRISA / INRIA & ENS Cachan , France. Grid Data Service (GDS) meeting, Rennes, September 22 th 2003. Context: Data Management on the Grid.
E N D
JuxMem: An Adaptive Supportive Platform for Data Sharing on the Grid Gabriel Antoniu, Luc Bougé, Mathieu Jan IRISA / INRIA & ENS Cachan, France Grid Data Service (GDS) meeting, Rennes, September 22th 2003
Context: Data Management on the Grid • Distributed numerical simulations (code coupling) • Needs data sharing • No many functional systems Designing a satellite Solid mechanics Optics Dynamics Thermodynamics
Existing Data Management Systems • Non-transparent large scale data management • GridFTP (Globus) and MPI-IO • Internet Backplane Protocol (IBP) • Explicit transfert • No consistency ensured • Transparent small-scale data management • Distributed shared memory (DSM) • Consistency models and protocols • Transparent access • Small-scale, static and homogeneous architecture TreadMarksTM
Another approache: peer-to-peer systems • Peer-to-peer systems (P2P) • Distributed (large-scale) • Volatile peers • Peers have the same capacities and responsabilities • Sharing of immutable data • Centralized (Napster) • Flooding (Gnutella, KaZaA) • Distributed hash table (CFS, PAST) • Sharing of mutable data • One writer per data + static architecture (OceanStore) • Conflicts have to be manually resolved (Ivy)
Design principles • Proposition: data sharing service for the grid • DSM systems: consistency and transparent access • P2P systems: scalability and high dynamicity
A Data Sharing Service for the Grid Persistent storage Transparency of localization Internet ? Data transfert
A Data Sharing Service for the Grid Optimization of access Data consistency Internet Internet Data transfert
A Data Sharing Service for the Grid Internet Internet Scalability of the architecture
A Data Sharing Service for the Grid Internet Internet Handling volatility
JXTA: a framework for P2P • Open-source platform for programming P2P applications • Specify a set of protocols • A peer • Uniquely identified (ID) • Address independent of the physical location • Several network access point (TCP, HTTP, etc) Peer ID Peer ID Peer ID Peer ID Peer ID Peer ID Peer ID Peer ID Peer Peer Peer TCP/IP Peer Peer Peer Peer Peer Peer Firewall Firewall Peer Peer Peer Peer Peer HTTP
JXTA: peer groups • Set of peers that share a common set of interests • Scope of communications • Specific policy of management • Peer group services NetPeerGroup PeerGroupA Peer ID Peer ID Peer ID Peer ID Peer ID Peer ID Peer ID Peer ID PeerGroupB
JXTA: Advertisements • Every resource is described by an advertisement • Peers • Peer groups • Communication channels • Services • … PeerGroup Advertisement: <?xml version="1.0"?> <!DOCTYPE jxta:PGA> <jxta:PGA> <GID> urn:jxta: uuid- BCBCDEABDBBBABEABBBABA000000 </GID> <MSID> urn:jxta:uuid- BFEFDEDFBABAFRUDBACE00000001 </MSID> <Name> My Group </Name> <Desc> This group is to be used for my own testing </Desc> </jxta:PGA>
JuxMem: a prototype Peer group juxmem Peer group data Peer group cluster A Peer group cluster C Peer group cluster B Logical architecture Physical architecture
API of JuxMem • Alloc (size, attribs) • Map (id, attribs) • Put (id, value) • Get (id) • Lock (id) • Unlock (id)
Managing Memory Resources • Advertisement of type provider: peer group cluster • Advertisement of type cluster: peer group juxmem Peer group cluster Memory provided Peer group juxmem Size 10 Size 10
Managing Shared Data Blocks • Allocation of a memory area = creation of a peer group data • Data blocks identified by the ID of the peer group • Advertisement published in the peer group juxmem • Shared access for clients by knowing the ID • Consistency • Data blocks are replicated on providers • Updated simultaneously (logical multicast) • Clients are not notified of updates • Synchronization • Lock for each data block
Handling the Volatility of Peers • A manager by peer group (cluster and data) • Dynamic monitoring of available peers • Data blocks are automatically replicated (data) • Updates advertisement of type cluster (cluster) • Volatility of managers • Periodic exchange of hearbeats • Dynamic replication of managers if needed on other peers
Implementation of JuxMem • JuxMem • JXTA service • + 5000 lines • Graphical tool
Preliminary Evaluations • Cluster • PentiumII: 450 Mhz and 256 Mb of RAM • Network used: FastEthernet 100 Mb/s • Number of nodes used: 20 • Experiments • Overhead memory consumption • Low: @ 6 % with respect to the underlying JXTA peers used • Study of the volatility of providers
Study of the Volatility of Providers (1) Data of one byte Degree of redundancy = 3 Data manager is not killed 1 client: 100 iterations lock-put-unlock 16 providers Peer group juxmem Peer group data Peer group cluster
Study of the Volatility of Providers (1) Data of one byte Degree of redundancy = 3 Data manager is not killed 1 client: 100 iterations lock-put-unlock 16 providers Peer group juxmem Peer group data Peer group cluster
Study of the Volatility of Providers (2) • Internal lock when replicating • Ensure the consistency of the new replica • Client is blocked Peer group juxmem Peer group data Peer group cluster
Study of the Volatility of Providers (3) • JXTA and Java • Time-out detection adapted for the WAN • Independent of the data size Reconfiguration time @ 11 seconds Targeted volatility is weaker ( >> 80 seconds)
Conclusion • Defines an hierarchical architecture for a data sharing service for the grid • Specific policy for each cluster • Scalability • Storage and transparent access to data blocks • Application level: data block ID • Persistency • Shared consistency • Handling volatility of peers • Actively taken into account
Future Work • Transparent mobility of data blocks • Unavaibility of nodes or even clusters • Affinity data – computations • Affinity data – data • Parameterizable consistency • Specific for each client • Hierarchical synchronization
Allocation Request 6 4 3b 1 2 5 3a 3b 3a