The Oracle Sun webcast is going now, and I thought I’d blog what was interesting (for me, at least) as it goes. I’m not going to post this live, but I’ll have a go at capturing the key ideas as it goes.
First up is Charles Phillips, President of Oracle.
The new “red blocks” picture for the Oracle stack adds the Operating System, the Virtual Machine, the server and the storage, so it’s no longer just Applications, Middleware and Database. They want to be the IBM of the 1960s, but building on open standards.
In his discussion of the extent of R and D investment, Phillips emphasised that they have substantially increased their expenditure in product development over the last three years. They’re forecasting investment of $4.3bn in FY11, up from $1.5bin in FY05. They take products from the companies they buy and invest in them, and that would be no different with Sun. He states: “The number one server under the Oracle database is Sun Solaris/SPARC”.
Interestingly, in line with the comments in my earlier blog, he’s discussing the benefits of an integrated solution using as examples the Exadata V2 (where he says that demand is outstripping supply) and the Sun Storage 7000 with flash storage optimised for Oracle.
He’s now talking about the Netra range of carrier-grade of Telco equipment, and their expectation that this will be a big area of growth. This wasn’t something I’d noticed as an area of potential synergy, but it could be a substantial opportunity.
Of course, after all that discussion he has to emphasise the continued Oracle commitment to open standards!
Where next? They’ll focus on differentiated products, not the Windows/x86 space. High value technology will be the focus. They will simplify Sun’s supply chain, and eliminate a lot of the SKUs and complexity. They can implement a more attractive support plan than is currently possible because of their ownership of the complete stack. Finally, they are going direct to the major customers (I can see this is going to upset some of the bigger Sun partners). The top 1700 named accounts will have direct account management from within Oracle. They’re hiring, and looking for around 2000 new sales people. (Turns out that he and everyone else are wearing “We’re hiring!“ buttons.)
In respect of partners, a short contribution from Judson Althoff, Vice President, Worldwide Alliances and Channels explains that they are going direct to the largest customers, and while they will work with partners they want to see them participate with large clients on the basis of the value they can add. They still see partners with a major role in small and medium enterprises, as they do today, but based on the new Oracle partner specialisations.
JavaOne is going to continue, and it’s going to be collocated with Oracle OpenWorld but it will be a separate conference, and will go global.
Next up is John Fowler, Head of Engineering at Oracle.
Fowler starts talking about integrated systems, but also points out that the individual components will also be enhanced, like Solaris. They’re going to look for application performance leadership, because that’s what matters to the clients. In line with some of the comments in my earlier blog, they’re going to enhance features in Solaris, like fault management and predictive self-healing to extend to the Oracle database and applications. This will give a complete view of the stack. He emphasises that their focus is SPARC, and they will only work in x64 where they can deliver differentiated value, such as enterprise clusters using Linux and Solaris. They will not go after the broad space of commodity x64 hardware running Windows. Mike Splain, Senior Vice President of Microprocessors, introduces the accelerated SPARC development roadmap. For the UltraSPARC range which drives the T-series products, he claims that they have four new processor designs in development right now, following the currently shipping or announced processors. These have different design goals, including the potential to accelerate specific Oracle applications. The SPARC64 series, produced by Fujitsu, will continue, but in an evolutionary way and seems to be less of a focus for Oracle/Sun.
The Sun Storage 7000 ZFS product range will receive increased investment. Further, they’re going to use it as a single foundation for all storage requirements, which suggests an end to the traditional storage products produced by Sun. Flash storage can do 20,000 ops per sec, compared with around 300 ops per sec for traditional storage; this is a transformational technology, and the products in the entire stack have been and will continue be optimised to use this. Cindy Reese, Vice President for Operations discusses how they will build systems in future. Currently, they use build-to-stock, but this is unsatisfactory where demand is unpredictable. After discussions with Oracle they are going for build-to-order, simplifying and centralizing their supply chain, and reducing the number of products. They are going to close their two distribution centres and ship direct to clients from their suppliers.
Next up is Thomas Kurian, Senior Vice President, Oracle Server Technologies Development at Oracle.
He is talking about the software strategy. For Java, they want to extend it to new development and deployment models, and to extend community participation in the JCP. He summarises the different facets of Java, like the language, VM, and APIs, and discusses the different forms for clients, servers, mobile devices and so on. For Java SE, the focus is on Java SE 7, and the goals don’t seem much different from the currently stated Sun goals. Similarly, for Java EE, the goals don’t seem too different from the current goals. Perhaps the effort is on acceleration.
In respect of Glassfish, it will remain the Reference Implementation for Java EE, and will be supplied for departmental deployment. WebLogic is and will be the strategic product for enterprise applications.
For integrated systems, the first example is Exadata. He talks a while about this, but gets interesting again when he talks about integration with the Solaris capabilities around fault management to produce higher availability.
NetBeans investment continues but as the “lightweight” tool for Java development, and with particular emphasis a small number of key target markets like mobile development. JDeveloper is the strategic development tool for Oracle. Hudson for continuous development will be integrated with JDeveloper, and they will continue to invest here. They will also continue to support Eclipse.
Glassfish will continue as the Reference Implementation for Java EE, and has unique features like the non-blocking I/O, Metro web service layer, and a microkernel architecture that make it a very attractive product. However, WebLogic will continue as the strategic platform for enterprise deployments. There will some sharing of technologies.
For SOA, the Oracle suite will be the strategic platform, but they will continue to support the Sun products, including JCAP, for existing customers.
For LDAP, the Sun Directory server, which runs on a hierarchical file system, will see continued development but primarily for niche applications, and the Oracle Internet Directory (built on a relational database) will be recommended for most enterprise customers.
For systems management, they are looking at the management of the entire hardware and software stack, covering deployment, monitoring, tuning and so on. Oracle Enterprise Manager manages from the database up, and the Sun Op Center product manages from the operating system down. They plan to integrate them and then unify them.
Next up is Edward Screven, Chief Corporate Architect at Oracle.
He begins with Linux and the Oracle VM, as these are part of the Open Source BU, and to this will be added the MySQL business. They are committed to further development of MySQL, and to bringing it into the Oracle support network. OpenOffice will be managed as an independent global business unit, and they are retaining the Sun development team, and will be extending the product for the enterprise. They will add, for example, connectivity to the Oracle content management suite.
For Solaris, they will be able to extend the current Linux development model and provide enhanced support for Solaris. They will also be able to extend this to hardware. The overwhelming majority of customers run on either Linux or Solaris, and potentially both. They are going to carry on with this.
For virtualization they can now deliver the most comprehensive solution set from desktop to server and storage, with management for the entire integrated stack. They will deliver Oracle VM for SPARC, using LDOMs, to complement the Oracle VM for x86 (which is based on Xen), and they can be managed together. With Solaris containers and hardware-based dynamic domains there are a lot of options, but the key is the management tool. Oracle Templates (a collection of VMs with software pre-installed and configured) can be injected into a VM pool and it will just run, and with the new developments they can have templates for SPARC and Solaris as well as x86.
The virtualization story on the client starts with making VirtualBox part of the Oracle family, allowing creation of sandbox development environments that can be deployed to production. He also discusses VDI and Sun Secure Global Desktop, but points out that Sun Ray clients and servers are a much better and lower cost solution.
Next up is Larry Ellison!
He grabs a stool and sits down. Reminds me a bit of a lounge singer. He talks about how excited he is. He mentions the Sun Storage 7000, and says they’re calling it the Oracle ZFS storage appliance. He says that some things that were done in software will be done in silicone. Some things done in the database will be done in the operating system. Their vision for 2010 is the same as IBM’s was in 1960. That strategy made IBM the most important company in history, and Oracle likes that outcome. The Exadata V2 is another example of the benefits of integration of hardware and software. Oracle will still sell components on other platforms. Their goal is to build best-of-breed components, but to offer them as a best-of-breed integrated solution as well.
He is attacking IBM, and telling the audience that their benchmark performance on Solaris, SPARC and Oracle blew IBM out of the water.
Finally, he attacks analysts for their suggestions that Oracle will be making deep cuts to the Sun workforce, and suggests instead that they are hiring. He says the Sun business unit will be contributing to profits in the first month they own the company, and profits will grow. All they have to do is take it to market. They expect to take market share in servers, storage and tape libraries. It’s a short talk and he stops for questions. Some statements from the Q and A follow, largely without edit…
- Exadata pipelines are measured in hundreds of millions of dollars.
- Linux and Solaris are different products serving different markets, and the goal is simply to make them better.
- There’s a lot of money to be made in the Java middleware space, but exactly where the revenue comes from is not so important as the overall Java business making money. They don’t need to make money from Java as such.
- In response to a question about competing other companies producing products like the iPad from Apple, the response was that Oracle focus on what they do well. They have gone into new markets, but really they are in the data centre.
- Solaris will go down the route of being the OS for the mainframe-class server like the M9000, but also as a cluster OS for groups of x64 operating systems. The high end will be clusters of x64 machines or SPARC servers. It will be a long time before Linux catches up. If customers want Linux then Oracle has the best Linux in the world.
- The fact that IBM don’t have a database machine reflects limitations in DB2, and the fact that it can’t scale via grid computing. They are a decade behind Oracle, except on mainframes. Also, IBM don’t have Java.
- On Linux he argues they’ll be compatible with Red Hat, but offer a better Red Hat, without explaining how this will happen within an open source environment.
And that was it!