Five years and two weeks ago, I was herded into a massive conference hall in Oracle's giant Redwood Shores towers with a bunch of reporters, analysts and Oracle customers. It was celebration day for Oracle. After nine painful months of delays thanks to the DoJ and European Commission, its acquisition of Sun Microsystems was approved.
Ellison had wanted a hardware/software integrated stack for a long time, and this gave him what he wanted. Now, his Exadata and Exalogic servers would be produced by Sun and not partners HP and Fujitsu. Even better, he could offer a turnkey system with Oracle software pre-installed and fully integrated.
Oracle made a lot of promises, the most around MySQL. The European Commission held up the merger for months because of interference by the likes of Richard Stallman and Ralph Nader purely over concerns of support for MySQL, which was such a miniscule part of the business. Meanwhile, Sun lost $100 million a month in declining sales while it was held in limbo.
Ellison and Oracle made numerous promises around Sun hardware, Java and MySQL. How have the products fared in that time? Let's take a look.
1) Support for MySQL.
In December of 2009, Oracle made 10 commitments to MySQL, ranging from development promises to licensing promises to openness. I'm not really up on the database world, but one database developer feels Oracle has kept many of its promises and did some good things, like adding headcount to the QA department. In the end, though, its mere ownership of MySQL was enough to kill it because people would avoid it via association.
The one thing the writer didn't get into, and I think is a valid case, is the massive shift to unstructured data and Big Data analytics and the rise of NoSQL. After all, the anti-Oracle community forked MySQL just prior to the acquisition to create MariaDB but I still hear very little about MariaDB these days. It's a NoSQL/Hadoop world now.
2) Focus on high-end hardware
From the beginning, Larry said he had no interest in the commodity server market and would phase out Sun's many product lines in favor of very powerful, high end machines. He has certainly kept that promise. The specs on SuperClusters and what are known as engineered appliances like Exadata and Exalogic are off the charts.
Ellison must have seen the writing on the wall for commodity servers. Server unit shipments from 2010 to 2014 are even, according to Gartner, which means at best, servers are being replaced but no new ones are being deployed. Virtualization means you buy one server in place of 10, and the cloud means you dump your servers entirely like many companies are doing. The only people buying commodity servers these days are cloud providers like Amazon, Microsoft and Google, and they are more likely to build their own than buy.
So while Oracle has hardware unit sales in the hundreds or thousands, it's not losing all that much ground dollar-wise because its systems are so decked out. In Sun's last full fiscal year of 2009, it reported hardware sales and support of $6.7 billion. In Oracle's last fiscal year ended June 2014, hardware sales and support was $5.3 billion. But its unit sales have plunged. Cisco, a company associated with networking, has replaced Oracle in the top five of Gartner's server vendors for unit shipments.
Oracle has focused on using its appliances and engineered systems as turnkey products where the OS, database, middleware, applications, Java and everything else the customer needs or wants is pre-installed and fully integrated. Just plug and go. That smacks of the mainframe days but in some cases that's exactly what some people want.
3) Sparc or x86
With the x86 getting more and more features and functions as those found in mainframes, it has slowly pushed its way up the server stack, from departmental to mission critical jobs. Its gain has been RISC's loss, as Sun's Sparc, HP's PA-RISC, IBM POWER and Intel's own Itanium have lost ground.
HP abandoned PA-RISC for Itanium which is for all intents and purposes, dead. IBM is sticking with POWER after selling its x86 server business to Lenovo. And Sparc? Well, Sun hung in there until the end, but Oracle isn't exactly breaking its back to support or advance it. Most of the engineered systems and appliances are x86.
Sun delivered roadmaps on Sparc development back in 2010 and 2011 but has increasingly fallen behind on delivery. It has delivered on the T-Series and M-Series processors in 2011 and 2012 but since then the roadmap is emptying and the release dates are widening. Part of the problem may be that sales simply don't justify funding further development.
4) Solaris
This is even more depressing. Oracle abandoned OpenSolaris, the experimental open source version of the OS almost immediately, thus ending that experiment. Solaris 11 came out in 2011 and since then there has only been two point releases.
Instead, Oracle has put its emphasis on Unbreakable Linux, its own version of Red Hat Enterprise Linux. All of the engineered systems come with it, and Oracle ported DTrace, one of the more notable features of Solaris, to Unbreakable Linux.
5) Java
It wasn't very long after the merger that James Gosling, the inventor of Java, bailed from Oracle along with a lot of other Sun folks. It was no surprise. The culture clash had to be jolting for people like the laid-back Gosling and others.
It's been rough sailing for Java, and blame cannot be laid at the feet of Oracle. It has spent years cleaning up the mess of Java code and its endless parade of critical bugs that got so bad, a security group at Carnegie Melon and funded by the US Department of Homeland security recommended uninstalling Java unless you absolutely needed it.
This has distracted Larry's troops from delivering on new product. Java SE 7 came out in 2011 and Java SE 8 shipped in 2014, a fairly long gestation. Only one version of Java Enterprise Edition, version 7, has shipped since the merger. Java EE 8 isn't expected to ship until sometime in late 2016.
That said, Oracle has been quite good at listening to other JCP partners, especially IBM. Oracle has two very ambitious projects, Project Jigsaw, designed to make Java scale down to smaller devices, and Project Lambda, which seeks to add multicore support to Java 8.
The perception is that Java is in trouble. In reality, Oracle has put a lot of effort into it, it's a central piece of its Fusion product line and turnkey servers, and the editor of Dr. Dobb's Journal noted a few years back that "If Java Is Dying, It Sure Looks Awfully Healthy."
Conclusion
Sun was a mismanaged company. That's often why companies get bought out in the first place. Jonathan Schwartz was the wrong man for the job and made some very poor decisions, like spending $1 billion to buy MySQL, a $50 million company at the time, and putting all his efforts into open sourcing Sun software while sales were on a steady downward spiral.
Oracle brought some discipline and focus that was badly needed. The Silicon Valley has a rep for casualness in the workplace, but Oracle is more like Intel, Apple and Nvidia. It's a tough place to work and no one just punching a clock there. You better be on your A-game daily because the boss still shows up every day despite being 70 years old and worth billions. He could be cruising the world on that battleship he calls a yacht and enjoying the easy life. Instead he's still in the game. The Sun culture was not like that and I'm not surprised a lot of people left.
Oracle is making the most of an old-time Silicon Valley company whose time had come and gone. Sun sold servers. A lot fewer people buy servers at all these days, or they buy few compared to before. Had Oracle not bought it, Sun could have easily withered and died and hardware vendors like IBM, HP and EMC would have fought over its IP.
Sun lives on in a different form, but whether it is delivering to the bottom like as Larry has promised is still debatable.
This story, "Five years later, how is the Oracle-Sun marriage working out?" was originally published by ITworld.
No comments:
Post a Comment