Whether by accident or by design, and despite their presence in almost every other segment of the IT industry, the giants of the infrastructure business such as HP and IBM have steered clear of the business applications market, in recent times at least.
That has meant two things. Firstly, infrastructure providers have remained largely ‘application agnostic’, aware that it is businesses’ desire for applications that justifies their expenditure on infrastructure.
And secondly, it means that there is a logical separation in the IT infrastructure of every enterprise organisation between the components that do the business work – enterprise resource planning, financial budgeting, customer relationship management and more – and the other components that support them.
That arrangement could now be under threat, however, as database, middleware and applications giant Oracle prepares to enter the hardware market. If executed properly, Oracle’s ambitions could reshape not only the enterprise IT industry, but also the very nature of enterprise systems.
“There are limits to how far you can go if you just do software,” said Oracle founder and CEO Larry Ellison as he paraded the company’s new Exadata version 2 database appliance at its OpenWorld user conference in San Francisco in October 2009. And he was talking as much about his own company’s future as he was about maximising system performance.
The Exadata version 2 release is significant in that, unlike its predecessor, it is based on hardware from Sun Microsystems (among others). Oracle announced its intention to acquire the systems vendor in April 2009 and, were it not for the competition watchdogs of the European Commission, who are currently withholding regulatory approval (see ‘Setting Sun’ box), the Exadata 2 appliance would have been largely built on Oracle’s own kit.
And database appliances are only the start. “If you think about the Sun technology that we are bringing to the party, it’s really the [components of the] data centre,” said Sun chairman Scott McNealy as he appeared on stage with Ellison at OpenWorld. “And then you bring the database and the applications and the middleware, and you have a very nice data centre. That is what this combination is all about.”
Selling the singularity
Of course, this vision of a ‘nice data centre’, in which the majority of functional components have been sourced from a single vendor, is many IT decision makers’ worst nightmare.
Oracle insists that it is wedded to the principle of open standards – which allow customers to mix and match components from other vendors – and indeed that the success of its middleware business depends upon that principle. Nevertheless, if it is to capitalise on its breadth, Oracle must convince potential customers that the benefits of adopting the Oracle stack outweigh the strategic risk of relying so heavily upon a single supplier.
That is going to require a long sell, but for now Oracle’s argument is largely built around business intelligence and analytics. (It is not unusual in this regard; systems giant IBM currently puts its marketing muscle behind its ‘Smart Analytics’ systems and services, while Oracle’s chief rival in the applications space, SAP, hopes its BI and performance management toolset will tempt customers into spending more despite the downturn.)
On the one hand, the company is pushing the benefits of BI and analytics to business functions, and at OpenWorld many (often hypothetical) scenarios were presented that link intensive data analysis to business success. Its governance, risk and compliance (GRC) module, for example, could be linked to its procurement application to analyse purchasing patterns for evidence of corrupt employees receiving kickbacks, it says, or market trend analysis could be linked to stock control systems to create an almost autonomous fashion warehouse.
Meanwhile, Oracle is also pushing the value of its new hardware systems to support those analytical functions. “Exadata 2 is the fastest computer that has ever been built to run data warehouse applications,” said Ellison at OpenWorld, although some might question that statement (see ‘Hardwar’ box).
That is not to say that the Exadata box will be the platform of choice for standard business intelligence projects any time soon. Oracle has priced the device for the high end, and only when sub-second response times mean substantial business benefit will the performance enhancements justify the expense.
“The technical challenges we face around BI are more about the integration of the different components than the performance of the hardware,” says Charles Morgan, vice president for strategic enablement at Oracle customer American Express.
However, Oracle’s application strategy points to a time when BI is not the siloed function that it is today, and therein may lie the long-term value of Oracle’s hardware play.
In the company’s much-delayed Fusion range of applications, supposedly due for release in 2010, “BI was not an afterthought”, according to Ellison. “All the BI in Fusion is not separable,” he says. “The user interface is BI driven. You can’t use the system without using BI.”
Should the Fusion range be successful, the kind of online analytical processing tasks that the Exadata 2 is supposed to accelerate will arguably become core to the operation of applications and therefore of business processes.
According to R ‘Ray’ Wang, enterprise software analyst for Altimeter Group, this vision is not merely Oracle’s attempt to sell BI infrastructure. “People need more information than ever to make informed decisions,” he says, predicting that “a new class of analytical applications will emerge into existing systems”.
If that is the case, then Oracle’s hardware-to-application strategy puts it in a strong position.
There are, of course, all manner of obstacles to Oracle’s ability to execute on that strategy, among which is the impact it may have on its partner network. Oracle is a company that is unusually dependent on its partners; last year 80% of its transactions were conducted via partners.
That is because it has a limited services operation for a technology company of its size. “Oracle’s goal is to run its business with a large margin,” explains Forrester Research analyst Stefan Ried, “and services has a low margin.”
Elsewhere in the industry, however, it seems that every technology supplier wants to get into the services business, with Dell and Xerox being two recent examples of hardware suppliers that acquired IT services companies. So as Oracle moves further into competition with its hardware suppliers, it runs the risk of alienating an important route to market.
A pronounced case in point is Hewlett-Packard, whose hardware supports 40% of all Oracle database implementations and which, since the acquisition of IT services company EDS in 2008, employs 3,400 people who specialise in deploying Oracle systems.
But already, Oracle’s hardware ambitions are losing HP business. The first version of the Exadata machines was based on HP technology, which has now been replaced by equivalents from Sun Microsystems.
The IT industry is already a complex mesh of uneasy partnerships, and there is certainly no suggestion that companies such as HP and Oracle would stop cooperating in areas where it is still mutually profitable. But Oracle could find itself at loggerheads with companies that have more direct contact with its own customers than it does.
In theory at least, however, the Oracle and Sun / software and systems combination could give rise to new classes of product, so far prevented by the industry’s historical divisions. According to Forrester’s Ried, these could even include ‘ERP appliances’, enterprise applications that can be bought as hardware, or as cloud computing services secured and managed via hardware devices.
If even a fraction of that potential is realised, Oracle’s acquisition of Sun could prove to be one of the IT industry’s most disruptive yet.
Oracle’s proposed $7.4 billion acquisition of Sun Microsystems was announced in April 2009, but more than six months later the deal’s future still looks uncertain. The obstruction is the European Commission, which – at the time of going to press – is still withholding regulatory approval of the deal.
The EC’s objections concern open source database MySQL, the company behind which Sun acquired in 2008 for $1 billion. It worries that Oracle will cease investment in the product, thereby limiting customer choice.
Oracle denies this. “MySQL is a fantastic piece of technology, and we are going to increase our rate of contribution to that product,” said Ellison at OpenWorld. “I have taken the position that MySQL competes in different markets [to Oracle’s existing database products].”
However, the perception in the open source community is that Oracle is the enemy of open source, says Andy Butler a distinguished analyst at Gartner, although this is not entirely justified. “Everyone sees Oracle as this implacable enemy of open source, but it has made such a vast number of acquisitions in the past 15 years that it already owns a lot of open source technology,” he says.
But perceptions matter, and Oracle will have a difficult job exerting its influence over the MySQL product without alienating the community that supports and develops it.
“An open source product is only as good as the community of people around it,” Butler says. “Oracle wants to influence where MySQL is in the competitive landscape to make sure there isn’t competitive friction [with its existing products]. But creating a close clone of MySQL without any obligation to Oracle will be relatively easy, so they cannot risk alienating the community.”
The future of Sun’s vast range of technologies under Oracle’s ownership has been the subject of considerable speculation. Oracle has said that it will continue to invest in all of Sun’s products, often to a greater degree than Sun is today.
“Even in the worst-case scenario, all Sun technology will be supported through its natural lifetime, so there’s no reason for users to panic,” says Butler. “But not every piece of Sun technology will receive a strategic focus.”
He says that Sun users should postpone decisions about the long-term viability of their Sun investments until after the deal has or has not gone ahead. “[Oracle’s] statements at the moment cannot be anything more than a statement of intent,” he says. “Validated road maps will only become available at least three months after the acquisition takes place.”
“This is not the best time to become a Sun product user for the first time ever,” he says. Indeed, the uncertainty over Sun’s future is playing into the hands of its competitors, Butler says. The company itself has said it is losing $100 million a month in lost business as a result of the delay.
For this reason, Gartner has been critical of the EC’s delay of the Sun acquisition. “We feel that the whole Sun ecosystem is in a form of paralysis; nobody knows what to do,” Butler says.
As it enters the hardware industry, Oracle brings with it a certain combative approach to marketing that vendors in the space have largely been spared so far.
And already, the company has drawn battle lines against one former ally, IBM. Of course, as Ellison tells it, Oracle is the put-upon innocent of the piece.
“As soon as we announced the [Sun Microsystems] merger, IBM started going to Sun customers saying, ‘Oracle is going to get out of the hardware business, they are a software business, so buy the latest IBM kit,’” he said at OpenWorld. “We felt we had to respond. If IBM wants to compete, we are happy to compete.”
Oracle’s response was this: “We are going to challenge any enterprise company to take one of its existing database applications [that runs on IBM equipment] and if we cannot run that database application at least twice as fast on Sun gear we’ll give it $10 million,” Ellison said. “Oh, and IBM, you’re welcome to enter.”
Ellison’s trademark approach has inspired otherwise placid hardware vendors to take up arms. His claim that the Exadata 2 database appliance is “the fastest computer that has ever been built to run data warehouse applications” – thanks to its use of Flash storage media, which allows it to store large chunks of a database in the memory cache – provoked the following response from data warehouse applications vendor Teradata: “We’ve been very successful when we have gone head to head with Exadata,” says Martin Wilcox, director of platform and solutions at Teradata EMEA.
“The fundamental difference is that the Oracle database was built for transaction processing, not analytics. Oracle tries to throw a lot of hardware and CPUs at the problem, but it’s still a transactional database so it has input/output problems and resource contention. Adding Flash cache to the storage layer doesn’t address the fundamental issue that even simple queries require additional processing outside the storage layer – and that this additional processing does not benefit from the flash cache.”
Is this a positive development for the hardware industry? Typically, customers are weary of vendor rivalries, preferring suppliers to spend their time and money solving their problems. That said, anything that makes the often-dry world of hardware product development more interesting to engage with may benefit all parties.