SYS202 Test 3
Terms
undefined, object
copy deck
- Shared Locking
- Shared locking allows multiple transactions to"get" objects but no "update" until an exclusive lock is obtained. Deadlocking occurs more frequently than with exclusive locking.
- Data cube
- a multidimensional data model that allows enterprise data to be modeled as a set of facts organized along a number of characteristic dimensions.
- Concurrency Issues
- The "lost update" problem, The "uncommitted dependency" problem, The "inconsistent analysis" problem
- Data Warehouse
- A historical and unchanging data storage system
- Exclusive Locking
- Exclusive locks typically holds for an entire transaction
- Using XPath
- /node1/node2/.../finalnode/ returns nodes below final node
- XML is not
- Not a replacement for relational databases, not a programming language
- Data mining
- refers to the automated discovery of implicit patterns and unstructured knowledge contained in large data stores.
- Query Manager
- the data warehouse software component that fields requests for data warehouse content and manages the search, retrieval,and presentation of retrieved warehouse data
- Concurrency Control
- Dealing with multiple users accessing and changing database
- Operational Data sources
- Operational, transaction databases of the enterprise plus external data sources
- Cube
- The organization of facts and dimensions
- ETL (Extract, Transform, Load) tools
- Extraction: selecting required data (tables, tuples)from operational data sources for transfer to the warehouse and access to database schemata to properly locate the required data (i.e., SQL SELECTstatements)Transformation: manipulating retrieved source data and "cleaning it up" (e.g., dealing with missing data)or changing thedata (converting from one base to another, summarizing fields, etc.)Loading: then storing the extracted and transformed data in the appropriate warehouse structure
- Transaction Process Monitor
- Software that handles transactions, two approaches 1) make changes as occur 2) accumulate changes in transaction log save until transaction completed
- Support (Association Rule)
- =P[A∪B] / total # of transactions , Percentage s which supports A => B from A∪B
- Durability
- Once a transaction commits, its updates survive (even a system crash).
- HOLAP
- Hybrid mix of Relational and Multidimensional data warehouse
- dimension
- like attribute for a data warehouse
- ROLAP
- Relational OLAP server use Relational Database front end for data warehouse
- Uncommitted dependency
- User B updates User A retrieves, User B's update is rolled back. Can happen in reverse order
- Inconsistent Analysis
- User A is summing three values, collects one value Z, User B updates value A,
- MOLAP
- Multi dimensional model of data warehouse
- Step to data mining
- Determine task-relevant data, select general approach, define measure of interestingness
- Transaction Results (2)
- 1) commit - work finished, changes made 2) Rollback/Restart - work failed, undo work and transaction restarted
- SOA Services
- There is a separation of functionality based on the idea of a service. Clients consume services that are provided by one or more servers.A service can support multiple clients and may itself be a client of some other services.
- XML
- Semistructured Data Model, X-Markup Language, software and hardware independent means of information exchange
- Subject-Orientated
- data warehouses are organized around "subjects" - things or objects of interest to the enterprise for decision-making purposes in contrast to an enterprise process view point.
- SOAP
- Simple Object Access Protocol
- Types of data mining
- Visualization, Prediction, Cluster Analysis, Decision Trees, Association Rule
- OLAP operations
- roll-up, drill down, slice and dice
- Non-volatile
- no insertions,deletions, and update, just growth in size
- Association Rule
- form A => B which is interpreted as "tuples that satisfy condition (attribute/value pair) A are likely to satisfy attribute/value pair B Such rules are derived as empirical probability estimates, i.e, by counting the number of times A and B occur together in a sample of some (large) size
- Four Properties of Transactions
- Atomicity, Consistency, Isolation, Durability
- Facts
- are numerical measures; they are the fundamental data items pertaining to a particular subject, such as students, products,patients, etc.
- OLAP
- Online Analytical Processing, data summarization and aggregation
- Web service
- a function library residing on a server that can be called by remote applications
- Consistency
- Transactions preserve database consistency
- Atomicity
- Transactions are all or nothing
- Time Stamping
- the preferred method of concurrency control in a distributed environment.
- Metadata repository
- A database about data including a data dictionary. Operational database schemata
- Drill-Down
- this operator is the reverse of roll-up where we either step down a concept hierarchy, e.g., drilling down in the "US" cuboid or add a dimension to a cuboid, e.g.,adding Gender back into the "all_cube".
- Star schema
- A commonly used structure for representing a data warehouse.This is centered upon a fact table containing the raw data and a set of dimension tables.
- Encapsulation of Services
- The details of accomplishing a task are internal to a service and are not exposed to clients Service behavior can be modified without affecting clients as long as interfaces are not changed.
- Slice and Dice
- the slice operator works on one dimension of a data cube resulting in a sub-cube; the dice operator defines a new sub-cube by selecting on two or more dimensions.
- Lost update
- User A retrieves and User B retrieves, User A updates before User B updates (User A's update is lost)
- SOAP Request
- The SOAP request is just an HTTP POST request. The body of the POST request is an XML document. The root node is the SOAP envelope then a child node for the "body" then a child node for the function, children for the arguments
- Service Oriented Architecture
- An application can be "factored" into its functional components These functional components could be run on different machines Any component can use the services of another by sending messages and receiving results
- Remote Procedure Calls (RPC)
- Allow remote calls to the server without resending all the information
- SOA ideas
- Independent of hardware and software capabilities, using servers
- Business Intelligence
- Using data warehouse to gather data to make business decisions
- XML Rules
- There must be exactly one pair of tags for the root, and all other text is contained within these tags. Every start tag must have a matching end tag. Elements may not overlap
- Data Warehouse DBMS
- a storage organization to hold the data warehouse content in multiple levels of detail to accommodate different reporting and analysis requirements
- User Tools
- Reporting and query tools, OLAP, Executive Support System, Decision Support System, Data mining tools
- Confidence (Association Rule)
- =P[A∩B] / P[A] , Percentage c which supports A => B from A∩B
- How time stamping works?
- Every transaction is assigned a unique(system-wide) identifier at start time.Time stamping does not employ locks.
- XPath
- Query language that parses through XML
- Uses of data warehousing
- Capturing and archiving data, Analyzing data
- Wildcards in XPath (/*/)
- /*/*/finalnode/ all paths till final node (All nodes with finalnode as their 3rd node)
- Serializability of transactions
- A given interleaved execution of the transactions is serializable iff it produces the same result as some serial execution of the transactions.
- XML + HTTP
- Using XML to encode function calls and using an HTTP server to send and receive them
- Time-invariant
- data are archived to the data warehouse from operational databases at discrete points in time.
- Wildcards in XPath (//)
- //finalnode (All nodes that have final node as their final node)
- Data Warehouse Architecture
- Data warehouse DBMS, Operational data sources, Metadata repository, Query manager, ETL tools, Data marts, User tools
- Deadlock
- Each user is dependent on each other in a never-ending circle
- Isolation
- Transactions are isolated from one another (consistency and non-interference in the face of concurrent execution).
- Locking
- is the preferred method of concurrency control in a centralized, shared database environment.
- fact table
- contains the names of the facts, and their measures, as well as the key to each related dimension table.
- Deadlock Breaking
- Abort (rollback) any transaction waiting more than d time units or Choose a "victim" in the wait-for graph and abort (rollback)
- The semistructured model
- Data can be represented by any tree, highly flexible, not as structured as relational
- Data Marts
- a subset of a data warehouse for a particular enterprise unit (like a "view" in the relational model of data)
- Typical Transactions
- Retrieval, insertion, deletion, update
- Transaction
- A logical unit of work between Client and Server application
- Approaches to concurrency control
- Allow only one change at a time or interleaving actions that leave the database in consistent state
- Integrated
- an enterprise will likely have many databases which are designed,developed, built, and used operationally as distinct databases - billing, inventory control,sales, financial accounting, manufacturing, etc.
- Condition Selection
- //finalnode[//Title = 'Kyle']
- Roll-up
- aggregation on a data cube by climbing up a concept hierarchy, e.g. rolling up Home into a "US" group or by dimension reduction, e.g., eliminating an entire dimension such as Gender.