Alternative to Difficult Stored Procedures in Big Data Computation
Alternative to Difficult Stored Procedures in Big Data Computation
Join the DZone community and get the full member experience.Join For Free
The Architect’s Guide to Big Data Application Performance. Get the Guide.
As we know, the stored procedure is designed to handle computations involving complex business logics.
In the past, the data structure and business logics were so simple that one SQL statement was enough to achieve the user's computational goal. With the rapid growing of the information industry, users frequently find that they need to achieve increasingly complex computational goals to out-perform their competitors. To address such computations, SQL alone is far from enough. Database programmers have the additional demands regarding the judge and loop statements, branches at multiple levels, or more accurate data transverse operations, as well as decomposing an obscure goal into several clear and actionable steps correlated with complex logics. It was for all these demands that the stored procedure was introduced.
The stored procedure has so far become the preferred tools for complex data computations, playing an important role. However, stored procedures still cause various inconveniences. For example, lots of their functions are inconvenient - hard to debug or migrate; and there are some databases with rather weak stored procedures. These problems sometimes affect the efficiency of database developers seriously.
The inconvenient functions of stored procedure include the incomplete step-by-step computation, weak support for set-lized data computing, no sequence number can be set for the data set, and no object-reference mechanism. Let's check it out with a simple example - find the "top n best-selling products in whatsoever State" by analyzing the regional sales table. In this scenario, those inconveniences makes it quite complex to write a stored procedure.
01 create or replace package salesPkg 02 as 03 type salesCur is ref cursor; 04 end; 05 CREATE OR REPLACE PROCEDURE topPro(io_cursor OUT salesPkg.salesCur) 06 is 07 varSql varchar2(2000); 08 tb_count integer; 09 BEGIN 10 select count(*) into tb_count from dba_tables where table_name='TOPPROTMP'; 11 if tb_count=0 then 12 strCreate:='CREATE GLOBAL TEMPORARY TABLE TOPPROTMP ( stateTmp NUMBER not null, productTmp varchar2(10) not null, amountTmp NUMBER not null ) ON COMMIT PRESERVE ROWS'; 13 execute immediate strCreate; 14 end if; 15 execute immediate 'truncate table TOPPROTMP'; 16 insert into TOPPROTMP(stateTmp,productTmp,amountTmp) select state,product,amount from stateSales a where not( (a.state,a.product) in ( select state,product from stateSales group by state,product having count(*) > 1 ) and rowid not in ( select min(rowid) from stateSales group by state,product having count(*)>1 ) ) order by state,product; 17 OPEN io_cursor for 18 SELECT productTmp FROM ( SELECT stateTmp,productTmp,amountTmp,rankorder FROM (SELECT stateTmp,productTmp,amountTmp,RANK() OVER(PARTITION BY stateTmp ORDER BY amountTmp DESC) rankorder FROM TOPPROTMP ) WHERE rankorder<=10 order by stateTmp ) GROUP BY productTmp HAVING COUNT(*)=(SELECT COUNT(DISTINCT stateTmp ) FROM TOPPROTMP); END;
In which, the code at line 16 is to filter the duplicates, and write the filtered data to the ”temporary table." Since it is difficult to retrieve the distinct data directly, try this tip: find the duplicate data, then use "not" to reverse the condition, and the remaining is the distinct data. This function can be implemented by embedding two sub-queries.
For another example, the line 18 is to find the products ranking among the top 10 in whatsoever State. Firstly, use the window function to get the product rankings rankorder of each State; Secondly, filter out the top 10 products in each State; Lastly, get the products ranking among the top 10 in whatsoever State. SQL does not provide any functions to seek the intersection sets. So, to address this weak point, here is another tip: group by product so as to check if the number of a same products is equal to the number of States; if they are equal, it indicates that the product is ranking among the top 10 in every State.
Besides the inconvenient functions, the rather weak support for debug function is another inconvenience of stored procedures.
Although there are Oracle, DB2, and other databases offering debug functions for their respective stored procedures in the market, their debug functions are incomplete. In running the stored procedure, no matter if a SQL statement is long or short, no matter how many nested loops or computational steps it contains, programmers can only view the result of one statement, and the intermediate procedure is completely transparent to them. It rather defeats the purpose of step-by-step debug, and compromising the benefits of running to cursor or the next step. The programmers are actually only allowed to view the cursor and the simple variables. Such variables are useful, but by no means of the same importance as the intermediate procedure of SQL. Another trouble is that lots of setting and preparation workloads are required to launch the debugging tools. The beginners can hardly handle it without guides.
The third inconvenience is that the stored procedure is hard to migrate. Generally speaking, SQL can be migrated with a few simple modifications. Despite the slight difference in syntax details, the SQLs from various vendors are all the supersets of the ANSI standard. However, the stored procedure is quite another thing. Migrating a stored procedure is much more complex than rewriting one because the relevant standards of various vendors differ greatly. In this situation, users have no choice but stick to one database vendor rigidly. There is not any room left for users to beat down the price if database vendors overcharge them on upgrading their servers, storages, and user license agreements.
SQL is an essential function of any database, while the stored procedure is not. Some databases only offer the relatively poor stored procedures, and the others do not provide it at all. Take the stored procedures of MySQL for example. Its functions and performances are worse than that of the MS SQL, Oracle and some other databases, and MySQL may throw many exceptions on intensive concurrency. MSSQL Compact, SQLITE, Hive, Access, and other databases do not support the stored procedures.
Obviously, the inconveniences of stored procedures have compromised the computational performance of databases, and given much trouble to programmers – considerable development difficulty, inefficient development, and awkward maintenances. In addition, these inconveniences also affect the result of implementing business logics, achieving complex computational goals, and making smart business decisions. Then how do you empower the stored procedure?
esProc is a database computing script specially built for addressing the complex computational goal. With the computational performance equally good and even better than that of stored procedures, esProc additionally provides the advantageous intuitive grid style, step-by-step computation, professional debug features, agile syntax, complete computational system, and seamless support for the interactive computations between various databases.
esProc is a scripting tool with a grid style. With esProc, the computational logics can be laid out in a 2D space conveniently. The business algorithm can thus be interpreted in the computer language more easily. esProc advocates the step-by-step computation. To put it concretely, esProc enable users to decompose a complex goal into several simple steps in its grid, and ultimately achieve a complex goal by accomplishing those simple objectives. Designed with the "step-by-step" thoughts, a really practical debug function is introduced with esProc, including various functions like the break point, stepping, run to cursor, start, and end. Unlike the fake debugging script as SQL/SP, esProc can debug the basic steps directly and straightforwardly, needless building a specific intermediate table. The break point can be set in any position without altering the code.
esProc supports the true data type of a set. A member of a set can be the data of any simple data types, records, and/or other sets. esProc supports the ordered set, which means that users can access to the set member and perform the sequence-number-related computation, for example, ranking, sorting, year-over-year comparison, and link relative ratio comparison. The set-of-set can be used to represent the equal grouping, align grouping, and enum grouping. In addition, users can operate on the individual records in the same way as they used to operate on an object. esProc can represent the complex computation more easily with its agile syntax, for example, computing the relative positions in multi-level groupings, and grouping and summarizing by a specified set.
esProc can empower the stored procedures, and ultimately boost the computational capability of a database, reduce the development difficulty for programmers, improve the development efficiency, and facilitate the code maintenance and migration. esProc can implement the complex data algorithms and business logics easily.
Finally, for the above case we've discussed to conclude the inconveniences of stored procedures, let's check out the esProc solution shown below:
Opinions expressed by DZone contributors are their own.