Databases: Past (Oracle), Present (MySQL), future ("db4o?)
Recently, both the British government as the German have a firm commitment to invest in Free Software by gradually migrating its proprietary systems.
In my visit to the Accenture Delivery Center in Malaga , assured me that proprietary software licenses were 25 percent more expensive in 2008. Proprietary software costs that were already quite high in 2007.
One of the most widely spread proprietary products, even where it has a firm commitment to free software, is the management system of Oracle databases. Its main strengths are its robustness, security, and support. Its main flaws are its slowness, complexity and the large amount of resources consumed. These defects should be added to its high cost, both licenses and maintenance, requiring specially trained personnel to administer this type of database.
I think the end of Oracle's golden age has arrived. There is a large database management systems offer "free" with no defects in Oracle, and whose virtues are tailored to what is needed in most information systems developed. May not reach the power of Oracle, but neither need, and we also offer other benefits such as speed, or simplicity, involving extra cost savings.
will remain Oracle databases, as there are still plenty of work for Cobol programmers as there are critical applications, whose migration is hard to take, but it sure will be adopted in most new developments.
I think a good solution is to develop with MySQL, especially since it was bought by Sun, and became more mature. It is easy to install, maintain, the system requirements for operation are minimal, fast, stable ... "Is all I need" and is Free Software. What else do you want?
Most current developments for the Andalusian are done with MySQL, though Hank (Framework for the Development of Andalusia) recommend PostgreSQL (we know how is the policy of development of the government ...) .
As MySQL is the present, do not think it's the future, or at least not be the MySQL we know today. Moreover, no sense that we use the paradigm of Object Oriented Programming, then have to make mapping to convert the objects to records, even if this frameworks like Hibernate, which represent a black box to the developer as far as mapping concerns. Hibernate no longer a "trick", but increasingly offer more features, such as second-level cache, session management ... but have a cost, especially in performance. Moreover, these functions must be assumed, for the most part, by the DBMS.
In my company we have taken development using object-oriented DBMS, db4o , which has a licensing system identical to that of MySQL (dual licensed GPL / commercial). The results have been very good. It is a lightweight database, simple and fast, but do not use SQL operations on objects to persist. With db4o got no need to duplicate the data model of the application domain is ie, diagram no E / R (all are objects). We do not have to sacrifice performance as when we use an ORM. We can use a PDA in one device with Android, on Windows, Linux ... Is a fairly new, and still has a long way to go, but, there must be "shots."
post post Note:
do not know how this can affect to what I have.
0 comments:
Post a Comment