![Migration Steps – Extract Queries into Query Files
Iterator<ProductPO> pIterator = pH.getObjectsBySQLWhere("sku=? and domainID=?
and rownum=1", new String[] { sku, aDomain.getUUID() }).iterator();
Iterator<ProductPO> pIterator = null;
try
{
Map<String, Object> params = new HashMap<>();
params.put("SKU", sku);
params.put("DomainUUID", aDomain.getUUID());
productsIterator =
appProvider.get().getQueryExecutor().executePageableQuery("product/GetProductBySKUSimple",
params);
if (productsIterator.hasNext())
{
ProductPO p = productsIterator.next();
aProduct = productViewProvider.create(p.getUUID(), aDomain.getUUID());
}
} …
14](https://image.slidesharecdn.com/ansisql-ashortcuttomicrosoftsqlserverazuresqldatabaseforintershopcommercemanagement-180405104454/85/ANSI-SQL-a-shortcut-to-Microsoft-SQL-Server-Azure-SQL-Database-for-Intershop-Commerce-Management-14-320.jpg)
ANSI SQL - a shortcut to Microsoft SQL Server/Azure SQL Database for Intershop Commerce Management
- 1. © Intershop Communications Jens Kleinschmidt - Technical Product Manager / Architect ANSI SQL - a shortcut to Microsoft SQL Server/ Azure SQL Database for Intershop Commerce Management
- 2. Agenda Introduction Evaluation Why Microsoft SQL Server? Work Ahead MS SQL Support A Story of Epic proportion ANSI SQL to the rescue Migration Steps Outlook Summary Q&A 2
- 3. Introduction Monolithic applications supporting: Oracle Database ICM, ICI - based Intershop ORM Engine tightly coupled with Oracle PostgreSQL IOM - based on Wildfly in theory will run on “any” DB ICI – used for aggregated reporting data, JDBC Microservices supporting: Oracle Database, PostgreSQL, Derby via JPA implementation EclipseLink Consolidation to at least support one DB for all Intershop applications 3
- 4. Evaluation Evaluation PostgreSQL MariaDB MySQL Evaluate effort for PostgreSQL Evaluate EDB and embedded oracle Evaluate effort for Enterprise DB Develop Prototype for MS SQL Server Create concept for MS SQL Server 4
- 5. Why Microsoft SQL Server? PostgreSQL Supported already by IOM and Microservices Open Source, therefore no license fee, only monthly support costs Closest to Oracle, feature and syntax wise Microsoft SQL Server Commercial DB License costs, monthly support costs Cost neutral at worst to Oracle DB 5
- 6. So again, why we decided for Microsoft SQL Server?! We are Microsoft partner and SQL Server is our commitment to Microsoft. Azure SQL Database is the only DBaaS available under Azure. Meanwhile there is a preview available of Azure Database for PostgreSQL and Intershop got an invite for a closed beta test program. 6
- 7. Work Ahead Data Definition (DDL): >750 tables 16 views >2100 indexes Data Manipulation (DML): >150 stored procedures ~100 functions <10 PL/SQL packages (transformed into MS SQL functions) Data Query (DQL): 750+ queries (65 of them are Oracle/Microsoft specific) 7
- 8. MS SQL Support – A Story of Epic proportion Remove Oracle Client from ICM Create Test Framework for SQL Statements Extend SQL logs to dump input/output Create test replaying queries Compare results from Oracle/MS SQL runs Test coverage tool in Intershop Studio Adapt DBInit Create dump import/export tool Divide into structure and data creation Transform PL/SQL packages to MS SQL functions 8
- 9. MS SQL Support – A Story of Epic proportion Adapt ORM for MS SQL Locking Commit Adapt Platform to cater for MS SQL Adapt Storefront Functionality Separate queries from managers into query files Develop Query JUnit test skeleton Transform >750 query files + test cases Use ANSI SQL wherever possible 9
- 10. MS SQL Support – A Story of Epic proportion Adapt Backoffice Functionality Use Microsoft’s full-text index for search Enable Staging for MS SQL Build solution for our staging shadow tables in Microsoft Migration Support from Oracle DB to MS SQL Migration Guides Recipes for JUnit tests Operations Manual 10
- 11. MS SQL Support – ANSI SQL to the rescue We need to support Oracle and MS SQL with mostly small differences in syntax Dismantle to use ANSI SQL to lower maintenance 11
- 12. Migration Steps - Remove Oracle Specifics from Java Code Remove Oracle specifics from Java code 'nvl(' | 'trunc(' | 'asc nulls' | 'desc nulls' | 'to_date(' | 'bitand(' ProductVariationMappingRule.java UPDATE statement: … "SET p.typecode=BITAND(p.typecode, -1 -32) … … "SET p.typecode="+JDBCUtils.sqlFunction("bitand","p.typecode","(-1 -32)")+" 12
- 13. Migration Steps – Convert Functions to Procedures MS SQL functions can’t change DDL or DML Convert all functions to procedures which changes DDL or DML MS SQL doesn’t support PL/SQL packages Convert all PL/SQL packages into functions 13
- 14. Migration Steps – Extract Queries into Query Files Iterator<ProductPO> pIterator = pH.getObjectsBySQLWhere("sku=? and domainID=? and rownum=1", new String[] { sku, aDomain.getUUID() }).iterator(); Iterator<ProductPO> pIterator = null; try { Map<String, Object> params = new HashMap<>(); params.put("SKU", sku); params.put("DomainUUID", aDomain.getUUID()); productsIterator = appProvider.get().getQueryExecutor().executePageableQuery("product/GetProductBySKUSimple", params); if (productsIterator.hasNext()) { ProductPO p = productsIterator.next(); aProduct = productViewProvider.create(p.getUUID(), aDomain.getUUID()); } } … 14
- 15. Migration Steps – Implement Dialect Remove Oracle specifics by using ANSI SQL or implement Microsoft dialect <?xml version="1.0" encoding="UTF-8"?> <query> <processor name="JDBC“/> <template sqlDialect="Oracle"> … </template> <template sqlDialect="Microsoft"> … </template> </query> 15
- 16. Migration Steps – Implement Dialect Think twice if you want to use this! SELECT <sql-dialect name="Oracle"> sysdate FROM DUAL </sql-dialect> <sql-dialect name="Microsoft"> GETDATE() </sql-dialect> 16
- 17. Outlook Beta phase in May/June Restricted access from Partners/Customers Fully functional except Staging Further Evaluation of effort to support PostgreSQL No decision made if and when Depends on amount of new/old on-premise customers demanding a free alternative 17 βeta
- 18. Key Takeaways Intershop Shortcuts like ANSI SQL lowered the effort significantly Maintenance should be less painful ANSI SQL helpful as lowest common denominator This isn’t worth a dime if automated tests aren’t verifying the results! Customers Manageable effort for Migration ANSI SQL is only helpful if Oracle and MS SQL needs to be supported Azure customers should migrate to SQL Database Otherwise a DBaaS can’t be used in Azure 18
- 19. Q&A 19
- 20. The world of commerce is changing. Unlock your potential with the exciting possibilities of Intershop omni-channel commerce. Jena, Germany Hong Kong, China Melbourne, Australia San Francisco, USA Amsterdam, Netherlands Berlin, Germany Frankfurt, Germany Hamburg, Germany London, UK Nuremberg, Germany Paris, France Rio de Janeiro, Brazil Sofia, Bulgaria Stuttgart, Germany intershop.com [email protected] Furthermore Intershop is represented in Austria, Belgium, China, Denmark, Finland, India, Italy, Norway, Russian Federation, Spain, Sweden, Switzerland, and Turkey. For a full overview, as well as for contact details please consult our website: www.intershop.com/offices-and-subsidiaries 20
Editor's Notes
- #2: - What did Intershop do to support MS SQL? - Facts, numbers, performance measurements: Oracle DB vs. MS SQL - Steps to migrate from Oracle DB to MS SQL