Those who cannot remember the past are condemned to repeat it.

This famous quote of George Santayna's came to my mind when I tried to follow the publications on the inevitable cloud, in particular on Oracle's cloud. "Oracle has more cloud applications than any company. And cloud apps are hard work to build."  - refer to Larry Ellison at OpenWold user conference Undoubtedly the ultimate foundation for Oracle cloud is the database Oracle12c - refer to Andy Mendelsohn at OpenWorld As matters now stand, it is time to remember Santayana's dictum and take a look, how the grandmaster of Oracle database consultants, Donald K. Burleson takes up an inconvenient position.

I would like to refer to Burleson "Oracle Tuning - The Definitive Reference", Rampant TechPress 2014.

In Chapter 2 Donald enumerates some inconvenient truths:

• The Oracle marketing people have touted their SQL Tuning Advisor as the "Fully automated SQL tuning". In reality, SQL tuning requires human intuition.
• Reactive Oracle tuning treats the symptoms, not the disease. This is common all over the world, because proactive tuning is not widespread.
• There are problems that cannot be tuned: poor design of schema, application, external environment. Common all over the world.
• Offshore bargain promoted by IT managers: they try to save money by having their databases "off-shored" and designed by untrained third world neophytes.
• Database neutral vendor packages such as SAP are not optimized for Oracle, because they are designed to be "database agnostic", able to run on other databases such as DB2 or SQL Server. Because you cannot change a vendor package, database tuning on neutral databases can be very difficult.
• Any database designed in the 20th century is likely to be over normalized, without introducing the data redundancy that is needed to reduce unnecessary table joins.
• As the shifting economics made large computers attractive again, server consolidations caught on in the early 21th century, and mediocre DBAs were fired by the boatload because corporations no longer needed a large staff. The best DBAs remained in high demand, but DBAs who were used for repetititve tasks like installing upgrades on hundreds of small servers found it hard to stay involved with Oracle.

Notwithstanding the amazing efforts of Oracle developers to overcome performance problems - In-Memory-Database, self-learning optimizer, consolidation in pluggable databases, simplified management, etc. - the abovementioned facts tend to kill the joy, because - as Ellison points out - many customers have been using Oracle Database for many years and will not move everything up to the cloud next week: this transition will happen over the next 10 or twenty years. So the regular customers, people who pay Oracle, will live in a permanent state of transition, where nobody feels like "OK guys, let's put hands on!" And we all know, that transitional solutions tend tenaciously for eternity.

I will continue to write about this.