This post may be read to the strains of the wonderful Ella Fiztgerald singing “There may be trouble ahead”.

Having slogged through the marathon that was the Oracle-Sun announcement day, we’re left with two immediate impressions:
1.They took a heck of a lot of time to say so little
2.Oracle’s back to the future pitch, while nothing like as badly thought out as some people say, just isn’t going to cut it.

If you’re in a hurry, here are the headlines:

• The back to the 60′s mantra is both more and less nuanced than you might think
• But the integrated stack pitch simply isn’t well enough thought out
• There are two killer reasons why the “single stack” pitch fails: Innovation, and Focus
• Ultimately Oracle’s hardware strategy is simply not convincing enough
• Oracle’s software strategy is much more coherent but Sun brings a mixed bag
• The absence of a services story is the elephant in the room

We’ll be thinking, talking and writing in much more detail about this – so if you’re interested in knowing more, drop me a line – gary@bathwick.com

The back to the 60′s mantra is both more and less nuanced than you might think
Perhaps the most headline grabbing component of the strategy is the notion that Oracle is planning to emulate the IBM of the 1960′s by delivering a highly integrated stack from top to bottom.

Superficially this seems like total nonsense and no doubt the die-hard Oracle knockers will make much of this “backward thinking”. But you have to remember that when Charles Phillips presented the idea he added the caveat “…but built on open standards”. Oracle isn’t advocating a return to the tightly closed proprietary systems of the 60′s; the company is focussing on the “up-side” that came with those integrated systems – improved reliability, better integration, and a single source of support.

The problem with this guiding strategy is that it isn’t anything like as revolutionary as the sound bite might appear. Of course clients want things that are more reliable, offer better performance and deliver better economics (can you imagine anyone saying they want less?). This may be why IBM has been offering tightly integrated hardware and software bundles for a long time, even before it became necessary to describe them as “appliances”.

But the integrated stack pitch simply isn’t well enough thought out
The “integrated from top to bottom” story is superficially appealing, but doesn’t stand up to scrutiny particularly well.

Let’s begin with the “complete systems” chart above. Firstly there is a pretty compelling argument (made by IBM) that there is significant benefit from not having an apps business. IBM argues that customers gain by using a platform that supports the widest possible applications ecosystem – Apps vendors and clients then both gain the best infrastructure expertise and the best industry expertise. When it comes to industry expertise IBM can argue that it offers a breadth of industry expertise that spans a wider range of verticals and runs deeper into the infrastructure than Oracle and Sun’s (with the possible exception of telecoms where Sun really does have deep roots). We were also entertained by Oracle’s characterisation of SAP – they will no doubt be a bit surprised in Walldorf to learn that they don’t have any vertically focussed apps.

Oracle’s primary assertion is that it’s easier to deliver an innovative stack if you own all the components – a point made repeatedly throughout the briefing. In his opening pitch Charles Phillips said this:-

“If you have separate companies at each layer, one company building the DB, another building the storage etc it’s very hard to get those engineers to work together”

He then nodded towards the “big” problem when he said this:

“It’s hard to get those engineers to work together even when they’re in the same company – ask IBM, they know.”

We choked on our collective coffee at this; because IBM does actually know really quite well how challenging it is to get engineers from different groups to work together, because IBM has been doing it for half a century.

In short, Oracle claims that everything will be better in Oracle stack-land because all of the release dates can be synchronised, and integration can be “engineered in ahead of time”. Well, we wish them the best of luck with that; but, saying “we know it’s hard” and even “believing” that it’s hard are a long way from experiencing the challenges associated with coordinating so many different technology life-cycles. All joking aside, as Oracle discovers that this kind of integration really is very hard to do, the company might be well advised to go and ask IBM.

The final point on the ‘total stack’ approach is that some parts of it simply aren’t core to ‘engineering in integration’ – like storage. There’s a reason IBM backed away from manufacturing hard disk drives – If IBM felt that retaining its manufacturing capability would have given it a competitive edge it would have done so. Instead the company sold it, and created a very close partnership with the company that bought it (Hitachi).

There are two killer reasons why the “single stack” pitch fails: Innovation, and Focus
The single stack pitch fails for two key reasons – It doesn’t mean that you can innovate more quickly, and a “focus” on a dozen things nets out as no focus at all.

Innovation first. Oracle is promising to invest $4.3 billion on R&D in 2011 to cover the apps business, middleware, servers, processor design, storage, workstations, and network equipment. By comparison, IBM spent $5.8billion on R&D in 2009, on middleware, servers, processor design, and storage. In other words, IBM is spending $1.5bn more, on fewer things. Oracle won’t be able to invest as much in server design, processor design or middleware development as IBM, either from an overall budget, or a return on investment point of view. IBM’s R&D is already bolstered by partnerships with vendors like Hitachi (hard disks and semiconductors) and Sony (processors) which exist because they deliver far more innovation than IBM (or Hitachi or Sony) could deliver alone.

The further point is that coordinating development to ‘engineer integration in’ means that the whole will generally advance at the speed of the slowest, which will by definition slow innovation down.

Second, the question of focus. Oracle’s acquisition strategy prior to the Sun purchase was absolutely focussed on augmenting Oracle’s already successful applications and middleware businesses. The justification and value creation arguments were clear, and more importantly were focussed on something that Oracle was already very good at. The addition of Sun with its mixture of businesses (and particularly hardware) creates too many areas for Oracle’s senior management to think about – which argues against the ability to focus at all.

Ultimately Oracle’s hardware strategy is simply not convincing enough
While we’re prepared to believe that Oracle will do a better job than Sun did of making their businesses more commercially focussed (although that’s not saying much), we think that at best Oracle will buy a temporary reprieve for the hardware business, rather than preventing a decline that we see as inevitable.

There’s a real danger that Oracle will come to discover, between 12 and 18 months from now, precisely why IBM’s strategy was to walk away from certain elements of the hardware stack and, indeed, why IBM chose not to acquire Sun when it had its chance. In the meantime, IBM (and HP for that matter) will continue to provide a warm welcome to those clients who decide that while the Sun platform isn’t yet “burning”, there’s a distinct whiff of smoke in the air, and they’d rather be on a platform with a clearer more secure future.

One particularly interesting point is that Oracle stated that it’s not all that interested in the X86 market, which implies a level of confidence in Sun’s proprietary processors that the market doesn’t share. Sun cannot ship the volume that is necessary to maintain a viable processor family. A focus on “high value, differentiated products” can easily turn into “unprofitable but expensive niche” unless Oracle can find a way to bankroll the ongoing development of the Sun processor technology similar to the one that IBM found in its deals with the games console suppliers. IBM has shipped tens of millions of processors for games consoles, all of which form part of, and help to fund, the Power family of processors.

Oracle’s software strategy is much more coherent but Sun brings a mixed bag
While Java is the big prize, there are some niche technologies (the identity management technology for example) that add value to Oracle’s software portfolio. Much of Sun’s Java middleware is to be positioned as the reference implementation (which much of it originally was).

Given Oracle’s stewardship of InnoDB we think Oracle will most likely provide a good home for MySQL. But we’re not convinced that technologies like NetBeans are going to enjoy much attention as time goes on.

The absence of a services story is the elephant in the room
For a while now we’ve been characterising this decade as the decade when the power shifts from software to services. In simple terms the 1970′s and 80′s were the decades of hardware, the 90′s and the 00′s the decades of software and we’re now into an era where services holds sway.

It’s a simplistic characterisation of an industry that’s too complex to be defined in such a pat way, but it was interesting to hear the word “solutions” used over and over again by Oracle – as so many others have for the past 20 years – just as vendors like IBM are beginning to talk about “business outcomes”.

Yes, Oracle will be able to find customers that want to buy hardware, and even “solutions”, but the real money, particularly in the high end of the market, increasingly lies in the ability to help customers deliver business outcomes.

This is why, even before Oracle started talking about a return to the 60′s, we felt that the acquisition of a hardware company was basically a retrograde step, and that Oracle’s long term success would have been better served if the company had bought a services organisation. Of course, there’s still plenty of time for Oracle to do that. Watch this space.

If you’re pressed for time :

• It’s sad, and all that, but Sun had it coming : Sun’s performance meant that acquisition was inevitable
• There’s a hard road ahead for Sun employees as Oracle moves to meet the commitments it’s made to shareholders : Once the deal goes through, expect 12-18 months of turmoil within Sun
• There are major opportunities for IBM and HP, while Sparc isn’t yet a “burning platform” there’s a definite smell of smoke in the air
• While there’s no doubting Oracle’s success in software, the company is unproven in hardware and Oracle has plenty of  challenges to focus on over the next five years without adding the saving of a hardware business to the list
• Sun brings a mixed bag of software assets, with a danger that they’ll provide more disruption than revenues
• It’s unlikely that Oracle will “mess up” Java
• It’s unlikely that Oracle will spike MySQL and if it tried to MySQL would survive
• The EU will delay, but not prevent the transaction : That is unless Ellison gets bored with waiting

It’s sad, and all that, but Sun had it coming
I’ve been an admirer of Sun technology since the 90′s. So there is a part of me that’s a little blue at the thought of such an industry icon simply being absorbed into the Oracle machine. That said; Sun had it coming – the company is reaping the rewards of appalling strategy and excessive sentimentality.

Sun has suffered from a desire to “focus” on so many things, that the result has been a lack of proper focus on anything. The adage “Sun invented Java, and then managed to make no money at all out of it” has become a cliché.  Yes, there’s a case that Java drove the Sun brand which in turn drove hardware sales – but no matter how hackneyed a cliché is, there’s nearly always a strong underlying truth to it.

Sun wasn’t in imminent danger, the company had plenty of cash, and if the latest restructuring plan had been well-executed, there was a reasonable likelihood that Sun could have found an even, albeit “smaller”, keel. But, Sun has had plenty of restructuring plans (this latest one is referred to as “Restructuring Plan IX” – or “Plan i” in aphabetic terms) – so it has seemed pretty clear to me for some time that, notwithstanding Jonathan Schwartz’s manifest talents, he was either unable to, unwilling to, or prevented from driving the kind of wholesale change that Sun needed in order to thrive in the post dot-com era.

At the strategic level Sun was too slow to respond to changes in the market – As hardware vendors divided into the “commodity players” (like Dell) or the blended hardware + software + services players like IBM (and to some extent HP) Sun managed to fudge its strategy – by partially committing to the commodidity/volume play on one hand, and by making a big play in the software market on the other.

There’s a hard road ahead for Sun employees as Oracle moves to meet the commitments it’s made to shareholders
Larry Ellison has made some powerful promises to shareholders, the most exciting of these being that he expects the Sun acquisition to add $1.5 billion to Oracle’s non-gaap operating profit in year one, and $2.0 billion in year two. Those aren’t just big numbers – they’re massive. If you just take Sun’s R&D and SG&A expenses (excluding the restructuring charges, and “impairment of goodwill and intangible assets” hocus pocus) Sun has a gross operating loss of $378 million for the 2009 financial year. So in order to deliver $1.5, we’re looking for a combination of increased revenues, and reduced costs that drive an improvement in performance close to $2 billion dollars.

You don’t make those savings by cutting the office supplies budget.

Once the deal goes through, expect 12-18 months of turmoil within Sun
Sun reported that it had 34,000 employees in it’s 2008 annual report. There have already been a number of lay-offs since then – but the total is still close to 30,000. Only recently Sun announced another round of lay-offs, which it chose, absurdly, to blame on the delay in completing the acquisition caused by EU’s investigation. Anyone who believes that those lay-offs would not have happened had Oracle been able to complete the transaction is in for a big big surprise when it does complete.

Despite being relatively close to one another geographically, Oracle and Sun have cultures that are as different as night an day. Oracle’s culture is one that is absolutely focussed on commercial success with no room for sentimentality. While Sun may have been a more “fun” place to work – Oracle is a far better run business.

There is going to be blood on the walls at Sun, because that is the only way Oracle can deliver the financial turn-around it has promised.

I expect Sun’s total employee count to be a lot closer to 20,000 than 30,000 by the end of next year.

There are major opportunities for IBM and HP: While Sparc isn’t yet a “burning platform” there’s a definite smell of smoke in the air
There’s no doubt that Sun’s recent performance has created an opportunity for the likes of IBM and HP to offer Sun customers a more secure long-term vision. The Oracle acquisition adds to the questions surrounding the long-term viability of Sun’s hardware business, and Oracle’s currently stated intent to retain the hardware business raises even more concerns.

Both IBM and HP will win business from Sun over the next two years. Even with a tie-up with the likes of Fujitsu, the future of Sparc is extremely doubtful. Sparc is certainly a great technology but it’s simply not viable over the long term. The level of investment required in order to sustain a processor business demands that you ship in volumes that Sparc simply cannot reach. Intel and AMD get those volumes from the PC / Server markets and IBM’s power processor family benefits enormously from its use in all three of the top games-consoles.

Oracle may be able to push Sparc as the “ideal” or “preferred” processor on which to run the database, and I’m convinced that Oracle’s famously “keen” salespeople will drive some volume, but I don’t believe they’ll drive enough to make Sparc sustainable in the long-term.

While there’s no doubting Oracle’s success in software, the company is unproven in hardware
This is an obvious point – but Oracle is absolutely unproven in the domain of hardware – Indeed in the past, Oracle has done a great job of battering hardware suppliers on price in order to retain margin for their software.

It’s also telling that IBM previously walked away from negotiations with Sun. IBM has far more experience as a hardware/software/services company, and part of me thinks that if IBM didn’t think the deal was worthwhile, it may not actually be worthwhile?

Oracle has plenty of  challenges to focus on over the next five years without adding the saving of a hardware business to the list
The rejuvenation of Sun’s hardware business is going to require an immense amount of management time and effort. In the meantime, the software industry is undergoing massive change, the emergence of SaaS, and cloud computing represent serious threats to the established business models upon which most large software companies base their businesses. If I were an Oracle shareholder – I’d want Larry to be focussed on the likely impact of SaaS and cloud, rather than have him figure out how to sell more tape drives.

Sun brings a mixed bag of software assets, with a danger that they’ll provide more disruption than revenues
Looking at Sun’s software portfolio, it seems that there are only a few really “interesting” bits – DTrace (beloved of kernel developers and Sysops everywhere) and Sun’s identity management technology (especially notable on account of it being profitable) to begin with. Then the two “star attractions” – Java, and MySQL.

There are other assets of course – the glassfish app server (which is unlikely to survive as a commercial offering), NetBeans (a great development tool that totally doesn’t deserve the market failure it’s enjoyed to date), then there are the assets that form Sun’s “stack” – the EAI technology (from the $387 million dollar SeeBeyond acquisition) – these will add little to Oracle’s Fusion middleware stack – which is already considerably enriched as a result of the BEA purchase.

It’s unlikely that Oracle will “mess up” Java
Oracle has been a supporter of Java, and has its own Java IDE (successor to the technology licensed from Borland some time ago). Given the JCP, and the fact that Oracle already has a clear stake in the Java middleware market, it’s hard to see how much more Oracle gets by acquiring the core technology itself. Oracle may see opportunties to drive cash out of the Java platform that Sun never did – but I suspect that if Oracle tries to hike the licensing fees for things like mobile Java the licensees may simply say “No thanks”.

It’s unlikely that Oracle will spike MySQL and if it tried to MySQL would survive
I don’t think MySQL represents a major threat to Oracle’s database business today, and it may be that by owning MySQL Oracle hopes to prevent it from becoming so – But in the mean-time, MySQL has found its way into places that the Oracle database has not. Essentially Oracle now has the number 1 market-share slot in both the commercial and open source database sectors.

I’m curious though, about how Oracle’s salesforce will view a “free database”, having been brought up on the raw steak that is big proprietary software, I doubt they’ll have much appetite for the humous that is open-source.

Ultimately, the “get out of jail” card for MySQL is the fact that it is open sourced. If all else fails, we can create a fork, call it “OurSQL” and get on with it.

The EU will delay, but not prevent the transaction

I’m really not a fan of the EU when it comes to competition regulation in the IT sector. They’ve bungled the Microsoft case time after time, and seem to me to be approximately a decade behind the times in terms of their thinking. I think the most likely outcome is simply a delay (and the opportunity for Oracle/Sun to blame more lay-offs on the EU), but there remains a small chance that  Oracle will have to hive of MySQL, and an even smaller chance that Larry will get irritated with the whole process and walk away entirely. When it comes to EU bureaucrats (and indeed Mr Ellison) you never can tell.

[Edited October 28th - to correct a glaring error - Had listed Crossworlds instead of SeeBeyond as the EAI co bought by Sun - Thanks Mr W-D]

Oracle / BEA

James Governor’s post on the subject is worth a read – He’s absolutely right to mention the Telco market (Big OSS is a major consumer of Object middleware), but there are a couple of things that I’d challenge – Firstly while I’d agree that this move will place IBM and Oracle the leadering infrastructure vendors in SOA (although Microsoft should be mentioned) but I wouldn’t put them in positions #1 and 2# when talking about SOA generally – Home grown SOA (via PHP, python, ruby etc etc) is surely responsible for the majority of service-oriented apps that are out there today?

James also mentions a great comment by Erik Bengtson pretty much nails it in his comment on the story at InfoQ:

my take on this acquisition: OC4J, Oracle ESB, Oracle BPM, Oracle web services manager will be replaced by better BEA products, WebLogic, Aqualogic ESB, Aqualogic BPM, WL Integration, Aqualogic Enterprise Security. Oracle BAM will evolve from microsoft environment and will integrate Aqualogic BPM Bam functionality. Oracle and BEA have both licensed HP Systnet as their Service Registry, so this is an easy integration. BEA engineers will focus into Front End / Web 2.0 era..

All of that makes very good sense. But lordy! That’s not a trivial amount of stuff to do.

Sun / MySQL

Judith Hurwitz is a sceptic like many of us, Judith has watched Sun make a mess of software for a lot of years, and I understand her scepticism. I’m a little more confidant, though, that Sun can make the acquisition work and ought to be able to make it work commercially too – Although Sun really does need to work on the latter – and on explaining “how” they’ll make it work commercially.

Oh and if you like to get an “Ann Coulter”-style take on a topic – check out John Dvorak’s punditry – It’s all a conspiracy by Oracle it seems – as is global warming no doubt.