Skeptically, it seems to me...
1) From the SF.com point of view - they are dependant on the Oracle DB on the back end, and it has served them well in buidling their offerings. They wanted to get a great, cheap, and predictable-over-time price for the DB as they keep expanding their user base. They also likely wanted to avoid the potential to have their price increased dramatically on a renewal of their license and support agreement at any point.
So I see this as mainly SF.com cementing a predictable cost structure for their infrastructure technologies. They made noises and motions about shifting to PostgreSQL, as part of the dance of negotiation with Oracle (hiring key Postres engineers for example). It seems that SF.com had a pretty good upper hand in the discussions, but also was pretty trapped into the Oracle DB as the back end for all their solutions. So the reality of switching to Postgres would be ugly and costly for them.
I am sure SF.com does not really want to run Exa-systems or Oracle Linux, and get themselves further locked into more Oracle technologies. Since Exa-systems are the opposite of being able to take advantage of low-cost scale out computing with x86 systems and infrastructure software (mainly open source). But to get the best price, they agreed to tell the world they will use these other Oracle technologies that are lackluster at best.
2) From the Oracle view - this, and the Microsoft cloud move, are about realizing that they are slipping on many fronts in the modern, cloud computing era. With hardware, middleware, and Oracle apps sales all being very challenged, and some new challengers to their database technologies (Hadoop, NoSQL, NewSQL), they had to do something since people are not buying the Exa-system message or story. So I think they gave the deal-of-a-lifetime on the cost side of the DB to SF.com, and they will be able to keep saying SF.com uses their database. I doubt SF.com will use many of their Exa-systems, or Oracle Linux, or Oracle Java, as the employee antibodies at SF.com against doing so, will be hard to change. While Marc B is off trotting around the globe pitching SF.com I doubt he will be checking on how the Exa evaluations are going.
So, I see this as a good win for SF.com to get a cheap, key piece of their infrastructure for many years, and Oracle gets to keep saying a hot company uses their DB, and the rest as "much ado about nothing" or "frivolous noise".
Biting the hand that feeds IT
