Slide 1AgendaIntroductionIntroduction (cont)Slide 5Slide 6ScopeContextSlide 9Slide 10Database DiscoveryDatabase StatementsDatabase Statements (cont.)Slide 14Slide 15Delivery SystemDelivery System (cont.)Distributed TransactionsDistributed Transactions (cont.)Database MetadataDistributed Query ServiceSlide 22DS in OGSARequirements from OGSARequirements from OGSA (cont.)Slide 26DAIS StandardsOGSA-DAI SystemSlide 29ConclusionDAIS Grid 1Database Access and Integration Services on the Grid** http://www.cs.man.ac.uk/grid-db/papers/dbtf.pdfAuthors:N. Paton, M. Atkinson, V. Dialiani, D. Pearson, T. Storey, P. WatsonFlorida International UniversitySchool of Computing and Information SciencesSummer 2006Presented by: Ariel CaryDAIS Grid 2Agenda Introduction Scope and Context of Proposal Proposed Database Services DS in OGSA Current DAIS Standards and Systems ConclusionDAIS Grid 3Introduction•Grid research generally focus on applications where data is stored in files• DBMS systems have a central role in data organization for numerous applications, e-Science: particle physics (LHC@CERN), earth sciences, bio-informatics• There is a need to interconnect pre-existing and independently operated databasesDAIS Grid 4Introduction (cont)This work seeks to encourage the development of standards that can meet those needs.A (preliminary) proposal is made for the staged development of a collection of Grid Database Services that allow access to existing, autonomous databases within GridFollows a service-based approach within OGSA framework for DBMS integrationDAIS Grid 5Introduction (cont)How functionalities are supported may come to be implemented in different ways (performance characteristics, etc.)Services definitions essentially state what functionality is to be supportedDAIS Grid 6Scope and Context of ProposalDAIS Grid 7ScopeThe proposal has several characteristics–Independent of any specific Grid toolkit (could skew and restrict it)–It does not propose the development of a new DBMS for the Grid, but wrapping existing systems to a consistent interface and developing distributed managers–Independent of any specific data model or access languageDAIS Grid 8ContextRelevant terms related to Databases–Database Service is any service that supports a database interface (WSDL)–Service interfaces are abstract and not prescriptive on how they are supported, or the data model that underpins a DBMS–Specific DBMS services could provide access to relational or object DBMS, XML repositories, specialist storage systems …DAIS Grid 9Context–Grid Database Service (GDS) provides capabilities for querying, updating and evolving a database–The interface also describes:Data delivery: transmitting structured dataTransactions: coordinating collections of operationsDatabase Metadata: accessing information about the data a DB service providesDAIS Grid 10Proposed Database ServicesDAIS Grid 11Database DiscoveryIt is assumed that a registry lookup returns a Grid Service Handle (GSH), globally unique name for a service instanceA service provider publishes description (WSDL) of a service to a service registryLater consulted by a requestor, and binding created that allow calls to the serviceDAIS Grid 12Database StatementsThus, it is a point of tension with the proposal being independent of the data modelStatements allow queries or change operations to be sent to a DBMSThis implies that the underlying DBMS supports a query or command language, different on every database modelDAIS Grid 13Database Statements (cont.)The pairs (queryNotation, query), … are introduced to allow flexibility (like MIME types for e-mail attachments)For example:–queryNotation=“SQL’92”–query=“Select * from EMP Where Salary>1000”DAIS Grid 14Database Statements (cont.)The optional txHandle indicates if the operation is part of a transaction, provided the DBMS supports transactionsThe final results of an operation are managed via:–resultHandle: generated dynamically–expires: an expiry time up for the result to be claimedDAIS Grid 15Database Statements (cont.)The operations on a GDS will be atomic:–Preparation and Validation: consistency check–Application: operation is performed–Result Delivery: results available to the callerUsually involve transfer of large amounts of data which may take long time to execute (prone to interruptions!)The implementation of the DBMS service should handle such failures to achieve atomicityDAIS Grid 16Delivery SystemMeans by which (potentially large amounts of) structured data is moved from one locations to one or more othersShould be considered complementary to protocols such as GridFTP, which could be used as a delivery mechanismDAIS Grid 17Delivery System (cont.)Single data source to be delivered, represented as a URISeveral destinations represented by URI with delivery mechanisms associatedThe deliver operation initiates delivery of the data from the single source to multiple destinationsA more elaborated delivery system would include encryption, progress monitoring, etc.DAIS Grid 18Distributed TransactionsA minimal transaction interface: performs the role of conferring a guaranteed unique identity on the transactionGiven a transaction handle, other operations over a database service can be put explicitly within the context of a transaction, using the txHandle parameterDAIS Grid 19Distributed Transactions (cont.)For a transaction to span multiple DBMS services, they must provide operations for use by the transaction manager that is overseeing the distributed transactionstartTransaction includes an expires param. to limit the consumption of resourcesprepareCommit operation can be used by a two-phase commit protocol to ensure that all participating database services commitDAIS Grid 20Database MetadataMetadata that could be useful to have access to includes:–Content description: DB schema – data model, logical & physical structures, stats (could be obtained from the data dictionary)–Capability description: language (query /update operations supported), transactional capabilities, protocols supportedThe metadata should be described in a standard representation, e.g. XML document given by the data service providerDAIS Grid 21Distributed Query ServiceQuery DS1 (DQS)Parsed & optimizedSub-queries to relevant DB’sResults collected &
View Full Document