They have the financial resources of Google, dwarfing Amazon in that regard. I don't think any of this will be decided by who has large data centers - that's a commodity that can be acquired at will when you're talking about companies with tens of billions in the bank. This will all come down to execution.
And Oracle does dwarf Amazon, significantly, in terms of financial resources. That's not a subjective opinion, it's an easy direct comparison.
Amazon has $7.6 billion in cash, a pile that is contracting rather than expanding, and they have to fund the continued expansion of their retail business, investing billions into new fulfillment and automation. Meanwhile it's highly unlikely Amazon is earning much profit out of AWS itself. Amazon produces some cash flow, but they do not produce net income, and do not compile much cash to their balance sheet from operations.
Oracle has $39 billion in cash, net income of $11 billion, and even more cash flow. Out of $37 billion in revenue, Oracle generates $14.6b in operating income; Amazon generates approximately zero operating income.
To put it all in perspective, Oracle generates more net income per quarter than Amazon has in its entire history. Oracle generates more cash per year to its balance sheet than Amazon has in the last ten years.
Saying they are comparable is saying that when Oracle sells a DB license and Amazon sells a toaster, these are only different in the amount of money the generate. They both have equal experience with software, even though Amazon's experience involves putting an item in a box and putting a label on it.
0. http://www.informationweek.com/cloud/infrastructure-as-a-ser...?
And you're obviously including something in calculating Oracle's size, since Oracle's cloud business is $0 IIRC -- which is much smaller compared to AWS's $3B.
Try $1 billion annually in saas revenue
Amazon gives you access essentially to mostly open source tools or licensing fees for closed source stuff. it seems almost a different market to me.
Iduno, the hosting aspect of the Java world is so weird. My Python/RoR friends try to tell me there is no good Java hosting & then I point out the fact that its right there on the Heroku they love so much. I think cloud-enabling Oracle will just make the massive power of the JVM more apparent with better tools (yes they may cost some money but so do Heroku add-ons)
That said, I do think the Java community should continue developing it's "rapid dev" sortof languages/frameworks (Clojure and maybe Scala/Play?) if they are ever going to attract start-up and small business types.
But yeah Java gets such a bad rap... I've had way less headaches since moving over from RoR. Java is a little more DIY in terms of composing a toolset (if you happen to be starting from scratch -- many like myself are not) but it's all there & way powerful. I deploy a handful of pretty well encapsulated projects to Websphere Application Server and it's pretty much a breeze, plus the IBM products, for example (not Oracle I know...) have massive admin interfaces that a rapid dev team couldn't mimic/maintain even if they wanted to.
The open source tools are great too, but as many like to complain -- Spring and the like "are just overkill for small projects". That may be true but why even put all this thought into your toolset if you are just trying to make small low-traffic CRUD sites...? Iduno, strange world.
They can't afford to cannibalize their existing cash cow.
That's still a very lucrative market, but I would guess that options from both the low and high end are starting to erode new sales opportunities and they might be losing some outright. e.g. it used to be the default in government IT systems to just stamp in "Oracle dB" in high level architecture diagrams and not get fired, but now postgres and other alternatives are being seriously considered and used.
"Oracle dBA" is being replaced with "dBA with experience in X, Y and Z databases" in many job reqs as well.
They still have huge enterprise license agreements, particularly in the fed space, but as budgets get cut those are simply being reconsidered.
This isn't really a move for "today", but for where Oracle will continue to vector if nothing is done over the next 3-5 years.
I grant you my experience with them is dated (they were a huge part of the NetApp push into data base markets) but they were all about the 'added value option' and high touch management / maintenance. Not something cloud compute is about.
To my way of thinking it seems like step one here, would seem be build a PUE < 1.05 data center with 10MW of provisioned power. Have they done that?