Win Sun, lose Sun: How Larry's bet on old-world systems hurt Oracle
AWS and Salesforce match up to Ellison's hardware
Oracle-Sun anniversary Feeling calm and relaxed? Join the Reg on a journey to an alternate reality where Oracle’s cloud business is the envy of all of its competitors. It holds two trillion objects and is growing faster now than at any time in its history.
What began as something for developers is becoming an enterprise IT staple – a platform for running critical apps and services for oil companies, retailers and media firms.
Microsoft and IBM, of course, got caught napping and are now hurling the kitchen sink into catching up: Microsoft has quickly built Azure and IBM has spent $1bn building additional cloud data centres to compete with Larry's cloud.
But Oracle is still setting the pace: cutting prices more than 40 times and effortlessly rolling out hundreds of new features.
Wake up, Larry... this is a all a dream – none of this is real. The truth is that we're describing the cloud juggernaut that is Amazon’s AWS. But... all this might have happened.
Let's go back half a decade...
Five years ago, Oracle purchased Sun Microsystems at what we can now see was an inflection point.
Sun-Oracle was old-world M&A – one big-tech firm swallowing another big-tech firm: the world’s biggest database maker gobbling a giant famed for servers and storage, chips and Unix, which, in its turn, had bought tape storage giant StorageTek.
In the old IT world of Compaq buys Tandem, Hewlett-Packard buys Compaq, HP buys Deloitte, Oracle buys PeopleSoft and BEA Systems, Oracle-Sun made perfect sense.
For Oracle’s chief executive, Larry Ellison, it was a steal, an alignment of perfect opportunity. He took Scott McNealy's loss-maker for just under $8bn and in return acquired Sun's considerable customer base.
He also scooped up the assets needed to turn Oracle into the Apple of the enterprise, from processors and servers through to middleware and apps.
Sun gave Ellison the engineering brains in chips through to systems and manufacturing plus the facilities that resulted in Exadata and Exalogic. In the past, Ellison had had to rely on partners such as HP and Fujitsu.
2010 was the realisation of a long-cherished dream of Ellison’s. Since 1998 he’d talked of marrying his software with hardware – a database in a box.
But more judicious thinking might have recognised that in 2010, a page was being turned in the IT playbook – with the cloud coming in and ownership of a systems business going out.
Have you been server-ed?
In the immediate aftermath, three years later, Oracle claimed Sun had paid for itself.
Oracle president Mark Hurd reckoned in 2013 that cash flow derived from Sun had exceeded the $7.4bn purchase price.
Two years later, the state of Oracle’s business says otherwise.
There’s been little to no pay-off from Sun’s mass-market served business.
It was engineered systems that were one of the prime motives behind Larry Ellison's decision to buy Sun: mass-market servers, running Unix on Sparc or Intel, were never a factor.
That business was already faling under Sun and its decline has done nothing but continue under Oracle.
In Sun's last good year before being bought by Oracle, 2007, sales for servers totalled $5.9bn. Of these, $5.2bn's worth were Sparc and it got $700m for x86 systems. In 2013, the last full year for Oracle, total systems sales were $2.3bn – $1.2bn from Sparc and $1.1bn from x86.
Oracle was right to sidestep the server mass market: today’s market would be unrecognisable to the server-pushers of 2010.
Cheaper vendors are invading from Asia, while spending in the enterprise has slowed.
Since 2010 we’ve seen the growth of hyper-scale computing users such as Amazon, Google, Twitter and Facebook. It’s this group that is the locus of server spending.
But this sect of the select simply aren’t buying off-the-shelf servers – rather than accept what they were given, these firms turned the supplier dynamic of old on its head, and are working with third parties who specialised in design and contract manufacturing to build the servers they want.
The problem for the server status quo of Sun-Oracle in 2010 is it is these hyper-scale web firms that are the growth opportunity – they are the ones buying the most servers.
The enterprise server market, by contrast, is down thanks to virtualisation and server and data centre consolidation, with CIOs spending on storage and networking instead.
Scott McNealy's loss-making Sun was a quick opportunity
The customer infection has spread: the Open Data Center Alliance, Open Compute Project and Project Scorpio all sprung up between 2010 and 2014, formalising users into groups working on common standards and designs in servers and data centers to help web-scale compute.
They have yet to make an impact on mass-market server designs, but they are seen as the new reality of IT and server and component makers are members and sponsors.
“The awareness of this other style of web-scale IT in the enterprise is gaining in popularity,” Gartner research director of data centre dynamics Errol Rasit told The Reg.
“Dev ops – thanks to Rackspace and AWS – is the first step in that direction.”
Oracle has now effectively withdrawn from the mass market and put its faith in low-selling, high-price and high-margin integrated systems.
Sun’s engineering brains and design and production brains was a win-win that has allowed Oracle to build the Exadata Machine and Exalogic Engineered Systems – the first Exadata was on HP.
Owning Sun meant Oracle could declare UDI on Exalogic and dump reliance on HP or anybody else.
Exadata has since become the best selling of nine Oracle engineered systems, with ExaLogic second, according to Gartner.
When it comes to selling systems that integrate server, storage and networking, Oracle is beating the competition, the analyst says.
Oracle in 2013 made $855m from what Gartner calls “engineered infrastructure hardware” – excluding the price of software – with IBM in second place on $212m and Hewlett-Packard third on $88m. From a relative perspective, Oracle has therefore gained from owning the Sun hardware business.
Oracle has hundreds of thousands of customers but has sold just a few thousands Exadatas; the growth opportunity is therefore on Oracle’s side. “There is a marketplace that sees the advantage of the performance of Exadata, and Oracle provides a single source of support and is very, very stable in terms of release cycles,” Gartner’s Rasit said.
But measuring up to IBM and HP is the wrong yardstick for Oracle in today’s climate.
It's hard out there for hardware
Oracle’s hardware business barely grew in its most recent quarter – up one per cent. Revenue made from hardware has been up and down since 2010 but the underlying trajectory is down. As a percentage of Oracle’s business, hardware revenue has fallen, too – at 14 per cent of $1.3bn.
Today, Ellison and his new CEOs – Mark Hurd and Safra Catz – make various claims about being the largest cloud SaaS company as well as predictions about how Oracle will become number one in cloud.
Total cloud revenue for the second quarter was up 47 per cent from a year ago in the most recent quarter.
The rhetoric is a mix of Oracle top leadership's bullishness mixed with the fact Oracle’s applications, platform and infrastructure as a service had actually seen more growth than rivals.
Shaking that AaaS
Most application-as-a-service providers like Salesforce and NetSuite are growing at around 30 per cent a quarter. SAP’s apps-as-a-service business is growing at a similar percentage.
But when it comes to raw dollars, the bullishness doesn’t translate. For once, the giant that places a famous premium on profit and measures things in how much it makes is weak.
Oracle’s cloud business raked in $519m in Q2, but there are two problems with that figure. The first is that hundreds of millions of dollars are peanuts for this giant – a mere one to two per cent of total revenue.
Its legacy, on prem-software licences and maintenance for things such as Oracle ERP or database are still the things that count - making up more than 70 per cent of business.
New software licences earned $2bn and $4.7bn respectively in that quarter. Where does Oracle’s cloud business sit vis-à-vis other app-as-a-service rivals? Oracle grew and earned more than NetSuite, but was behind Salesforce and SAP, which are also adapting to the cloud era.
The other problem is more one of embarrassment for Oracle, because it shows the wrong strategic bet was made in 2010 in buying Sun – because, after all, cloud is now growing faster than hardware.
Or, put more bluntly, cloud is growing. The engineered systems sector is not.
It's amazing, that Amazon
First, let’s take AWS, the yard-stick on infrastructure and platform-as-a service.
Not only is Oracle’s cloud revenue peanuts next to that of AWS, but Amazon's web services business is actually making about the same as Larry’s engineered systems business – and it’s growing faster.
That’s an organic business, rather than one which Amazon would have had to spend money on – like, say, the Sun acquisition.
Amazon doesn’t break out the AWS revenue as a separate line in its spreadsheet, reporting it simply as “other”.
That “other” business was growing nearly as fast as Oracle, 37 per cent, in Amazon’s Q3 but making about the same as Oracle’s hardware systems products – $1.34bn.
AWS has now been valued as a possible $38bn business – making it so lucrative that it would be the subject of a bidding war should Amazon spin out the business, according to one analyst.
Amazon’s size is a product of its momentum, which was provided from having started first with AWS – in 2006.
How about software-as-a-service? Salesforce grew 29 per cent in its most recent quarter, again making around the same amount as Larry’s hardware business: $1.38bn.
Again, this has been organic growth for Salesforce in its core – not the product of customer acquisition through M&A.
Two years after AWS and nine years after Salesforce, Ellison was laughing off cloud as a passing fad. Two years after that, Ellison spent nearly $8bn swallowing Sun. His initial response was to sell more Oracle – flog Oracle systems and databases to those building clouds. Now, he’s come full circle, becoming a cloud service provider.
Every cloud service Oracle is now offering, all of which it claims are industry-leading, were launched in the years after following the Sun deal and following Mark Hurd’s 2013 claim of payback.
Oracle’s database-as-a-cloud service, Java cloud, storage cloud, big-data cloud, integration cloud and the rest came between 2012 and September last year.
It was in 2014, too, that Oracle finally rolled out a version of its flagship database capable of multi-tenancy, a cornerstone of Salesforce’s cloud service architecture.
Bezos: the one Ellison should have been watching
Oracle 12c is now the foundation of Oracle’s elastic clouds and available to customers who want to build their own private clouds running on Oracle middleware.
The opportunity is there for Oracle hardware that may yet justify buying the Sun systems business.
Exadata has only been sold to a few thousand customers and the Oracle customer base is 300,000. That’s “opportunity”.
But sales are slow: Oracle is working hard to woo those who are happy buying everything from Oracle for the integrated technology stack or simplified purchasing and support.
It’s not like nobody saw something like AWS coming. Or even Salesforce – Larry was an early, founding investor in the latter. AWS launched in 2006, Salesforce in 1999. Cloud computing was already a “thing” by the mid-2000s.
Ellison isn’t the only one in enterprise IT to have overlooked the rise and importance of cloud computing, or to think it could either be contained or parleyed away.
Like SAP, Hewlett-Packard, IBM and Microsoft, Oracle is scrambling to make money as a provider of hosted cloud services by hosting other people’s data. Surprise – its fastest growth is in selling subscriptions to online versions of its software.
Unlike these others, however, Oracle’s CEO didn’t just avoid a burgeoning market – he closed his eyes more tightly.
Oracle cloud business could still haul past AWS, but it has handicapped and set itself back in getting there. ®