Embedded SQL, Database APIs, and PL/SQL Juliana Freire Some slides adapted from J. Ullman, L. Delcambre, R. Ramakrishnan, G. Lindstrom and Silberschatz, Korth and Sudarshan SQL: Benefits • Declarative languages: program is a prescription for what data is to be retrieved, rather than a procedure describing how to retrieve the data • When we write an SQL select query, we do not make any assumptions about the order of evaluation • Can be automatically optimized! – Decision about order and evaluation plan is left to the optimized – Optimizer has the resources to make sophisticated decisions SQL: Limitations • Not flexible enough for some applications – Some queries cannot be expressed in SQL – Non-declarative actions can’t be done from SQL, e.g., printing a report, interacting with user/GUI – SQL queries may be just one small component of complex applications Trade-off: automatic optimization of queries expressed in powerful languages is hard Solution • SQL commands called from within a powerful host language program • Three main integration approaches: – Embed SQL in the host language (Embedded SQL) – Create special API to call SQL commands (JDBC,ODBC) – Allow ‘external’ code to be executed from within SQL • Query processing performed by the database – Leverage database capabilities Three-Tier Architecture • A common environment for using a database has three tiers of processors: 1. Web servers --- talk to the user. 2. Application servers --- execute the business logic. 3. Database servers --- get what the app servers need from the database. Three-Tier Architecture Example: Amazon • Database holds the information about products, customers, etc. • Business logic includes things like “what do I do after someone clicks ‘checkout’?” – Answer: Show the “how will you pay for this?” screen. Environments, Connections, Queries • The database is, in many DB-access languages, an environment • Database servers maintain some number of connections, so app servers can ask queries or perform modifications • The app server issues statements : queries and modifications, usually Client and Servers in SQL Environment CONNECT TO <server name> AS <connection name> AUTHORIZATION <name and password> - A client may maintain several connections to the server, but only one can be active at any point in time - Operations performed while a connection is active form a session SQL Connection SQL Client Session Server Three Kinds of ‘Modules’ • Module = application program • Embedded SQL • Generic SQL Interface • True modules – stored functions and procedures Embedded SQL • Embed SQL in the host language. – A preprocessor converts the SQL statements into special API calls – optimized access plans that are stored in the database – Then a regular compiler is used to compile the code. • SQL statements can refer to host variables (including special variables used to return status). – Input: variable given a value by the host program to be used in a query – Output: variable used to retrieve value from the database • Language constructs: – Connecting to a database: EXEC SQL CONNECT – Declaring variables: EXEC SQL BEGIN (END) DECLARE SECTION – Statements: EXEC SQL Statement; Declaring Variables • All variables used in SQL statements must be declared EXEC SQL BEGIN DECLARE SECTION char c_sname[20]; long c_sid; short c_rating; float c_age; EXEC SQL END DECLARE SECTION be careful with type mismatches: check basic type correspondences, e.g. in C, Varchar(n) = char[n+1] Executing Statements EXEC SQL INSERT INTO SAILORS(sname,sid,rating,age) VALUES (:c_sname, :c_sid, :c_rating, :c_age) EXEC SQL UPDATE SAILORS SET rating = :c_new_rating WHERE sid = :c_sid These queries could be used to populate and update the Sailors table through a Web interface Error/Status Reporting • Each SQL statement executed returns a status code + additional information – SQL communication area (SQLCA) • Two special “error” variables: ‒ SQLCODE: product specific • zero indicates normal execution • negative indicates an error has occurred • positive indicates warnings or special conditions, e.g., data not found ‒ SQLSTATE: defined by ANSI/ISO SQL92 standard • Error class, e.g., syntax error • Error subclass – implementation dependent Impedance Mismatch • Updates are simple to embed: they execute, possibly modify the contents of the database, and return SQLCA structure indicating what happened • Queries return data! • SQL relations are (multi-) sets of records, with no a priori bound on the number of records. No such data structure exist traditionally in procedural programming languages such as C • SQL supports a mechanism called a cursor to handle this. Cursors • Can declare a cursor on a relation or query statement (which generates a relation) – Name associated with a query • Can open a cursor, and repeatedly fetch a tuple then move the cursor, until all tuples have been retrieved • Open causes the query to be evaluated EXEC SQL open c END-EXEC • Fetch causes values of one tuple in the result to be placed on host variables EXEC SQL fetch c into :cn, :cc END-EXEC Repeated calls to fetch get successive tuples in the query result Cursors (cont.) • SQLSTATE gets set to ‘02000’ to indicate no more data is available • The close statement causes the database system to delete the temporary relation that holds the result of the query EXEC SQL close c END-EXEC Note: above details vary with language. • Can use ORDER BY to control the order in which tuples are returned Cursor: Example Define SQL query that gets names of sailors who’ve reserved a boat of color desired_color, in alphabetical order, and declare a cursor for it EXEC SQL DECLARE sinfo CURSOR FOR SELECT S.sname FROM Sailors S, Boats B, Reserves R WHERE S.sid=R.sid AND R.bid=B.bid AND B.color=:desired_color input variable ORDER BY S.sname When are input host variables bound? When open is executed! Embedding SQL in C: An Example char SQLSTATE[6]; EXEC SQL BEGIN DECLARE SECTION char c_sname[20]; short c_minrating; float c_age; char user_name[10]; char user_pwd[10]; EXEC SQL END DECLARE SECTION c_minrating = 5; strcpy(user_name,”juliana”); strcpy(user_pwd,”xxxxx”); EXEC SQL CONNECT :user_name IDENTIFIED BY :user_pwd EXEC SQL DECLARE sinfo CURSOR FOR SELECT S.sname, S.age FROM Sailors S WHERE S.rating > :c_minrating input variable ORDER BY S.sname; EXEC SQL OPEN sinfo do { EXEC SQL FETCH sinfo INTO :c_sname, :c_age; output printf(“%s is %d years old\n”, c_sname, c_age); } while (SQLSTATE != ‘02000’); variables EXEC SQL CLOSE sinfo; FETCH Statement • What happens if updates are applied to table(s) referenced in a cursor after a result is fetched from a cursor? • Important note [Chamberlin, A Complete Guide to DB2]: The result of a query associated with a cursor may be – Completely materialized when the 1st row is fetched, or – Materialized one row at a time • Choice depends on the access plan chose by the optimizer Updates Through Cursors Can update tuples fetched by cursor by declaring that the cursor is for update" declare c cursor for select *# from account where branch-name = $Perryridge%# for update" ! To update tuple at the current location of cursor" update account set balance = balance + 100# where current of c# ! Apart from their method of finding the row to be updated/deleted, positioned updates behave exactly the same as UPDATE and DELETE statements that contain a search condition# Dynamic SQL •! Sometimes, we can’t predict in advance what SQL statements will be needed •! Dynamic SQL allows construction of SQL statements on-the-fly –! Need to PREPARE/EXECUTE –! Example: char c_sqlstring[]= {“DELETE FROM Sailors WHERE raiting>5”}; EXEC SQL PREPARE readytogo FROM :c_sqlstring; EXEC SQL EXECUTE readytogo; •! Run-time overhead – use only when essential –! Incurs the cost of access path selection (optimization) at run time –! Static SQL prepares statements in advance – well-suited for applications that perform repetitive queries/transactions Database APIs: Generic Interface Rather than modify compiler, add library with database calls (API) •! Special standardized interface: procedures/objects •! Pass SQL strings from language, presents result sets in a language-friendly way •! Embedded vs. Generic API: the difference is more a matter of look and feel than of substance •! Sun’s JDBC: Java API •! Supposedly DBMS-neutral –! a “driver” traps the calls and translates them into DBMS- specific code –! database can be across a network Database APIs: Generic Interface See textbook JDBC: Architecture for details! •! Four architectural components: –! Application (initiates and terminates connections, submits SQL statements) –! Driver manager (load JDBC driver) –! Driver (connects to data source, transmits requests and returns/translates results and error codes) –! Data source (processes SQL statements) Java Driver Oracle applet API Java JDBC JDBC application Driver DB2 Driver Manager XSB-Oracle Interface •! Allow calls to Oracle from Prolog •! XSB-specific calls: db_open(oracle(Name, Pass)) write('% Connected to Oracle as '), write(Name), writeln(' ...'),nl, db_sql('drop table DEPT'), db_create_table('DEPT', 'DEPTNO NUMBER(2),DNAME VARCHAR2(14)'), db_import(‘DEPT'(‘DEPTNO',‘DNAME'), deptall), db_insert(dept_put(A1,A2),(deptall(A1,A2))), dept_put(1,’computer science’), dept_put(1,’biology’),… Stored Procedures/Functions •! Issues with accessing DB from remote application: –! JDBC incurs data transfer overheads –! DB resources are tied up, e.g., open cursors •! Advantageous to execute some of the application logic inside the database –! Minimize data transfer