Sun's Future and Cloud Computing
Sun's recent layoffs (which are said to be affecting people working on OpenJDK, the JCP, J2SE, and desktop Java), and also Sun's recent acquisition of cloud infrastructure vendor Q-Layer, keeps alive the question of how Sun will redefine its strategic direction and choose which of its many technology possibilities it will focus upon. Sun is inextricably linked to the Java community and its recent financial, downsizing, and stock performance problems have been a cause for concern. Last fall, Tim Bray opened a conversation about the Sun's future direction with his post "What Should Sun Do?" Others, notably Ian Skerrett, Stephen O'Grady of Redmonk, Dave Johnson of BloggingRoller and Terrence Barr joined in.
Everyone agreed that Sun has a large and exciting technology portfolio - too large for it to maintain and exploit given its current situation. Everyone agrees that Sun would benefit by focusing on some subset of that portfolio. But which subset?
Tim would have Sun "adopt a laser focus on building a Sun Web Suite and becomming the the Web application deployment platform of choice." This focus would have Sun retain its hardware, OS, HotSpot JVM, and server side technologies (e.g. GlassFish and MySQL) and shed client side technologies (e.g. JavaFX) where he believes Sun cannot compete, NetBeans, and similar non-Web focused technologies. Tim would also have Sun terminate its role as "Steward of Java" by releasing the JCP in the same way that IBM released Eclipse.
Terrence Barr disagreed with most of Tim's recommendations, especially the abandonment of client side technologies. Terrence believes that the client side is what will keep Sun visible and drive traffic to its other, stronger with greater revenue generation potential, technologies.
Both Ian Skerrett and Stephen O'Grady argue for reform in Sun's marketing and its internal culture - emphasizing the need to do a better job at being a business. Instead of letting innovation, technology, and engineering culture drive the company almost completely; figure out how to do a better job at being a business, ala IBM when it faced similar problems a decade ago.
Tim and Stephen O'Grady raised the issue of Sun's participation in cloud computing, an area that is not a current Sun technological stronghold - although they did establish a cloud computing unit last year. Tim first notes some unknowns about cloud computing
- "Will it operate at the level of virtual hardware, like Amazon's AWS, or at Platform as a Service, like Google App engine (and, de facto, much of the PHP community)?
- Will buyers accept a certain amount of lock-in, or will they insist on zero barriers to exit?
- Will those who deploy enterprise applications be willing to let their data offsite and into the cloud? If so, what kinds of privacy guarantees will they require?
- Will those who deploy enterprise applications want to build internal cloud-flavored infrastructure?"
Despite these unkwnowns, Tim concludes:
"Here's something we do know: The business arguments for cloud computing look overwhelmingly attractive. I'm not convinced that Sun can succeed as a large-scale supplier of cloud services, and I'm not even convinced that we need to. I am convinced that we have to go ahead and build some Cloud infrastructure anyhow and operate it and make it pay for itself, so that when the ecosystem does find its shape, we'll understand it and be positioned to sell the Web Suite into it."
Stephen O'Grady adds to Tim's arguments by suggesting that Sun must be a player in this space. Given the nature of the current cloud vendors (Google, Amazon, Microsoft) who are committed to whitebox, commodity, hardware options, Sun cannot even count on being a supplier of choice in this market. The only way for Sun to play at all is to provide the cloud services on its own platforms.
Sun's acquisition of Q-Layer and its NephOS software gives it some cloud technology (and suggests an answer to Tim's question about whether it will operate as virtual hardware) but the acquisition alone does not yet hint at a cloud strategy or direction. Sun's VP of Marketing for its cloud computing unit, Carlos Soto, noted that "Sun was not in position to announce new cloud computing products on the first day of the acquisition but announcements will follow soon."
Except in the area of Cloud Computing, no one cited technlogies past the current product line of Sun to see what might be in its research and development pipeline. For example, Sun's Lively Kernel might have the potential to redefine Sun's client side technology offerings and allow them to penetrate markets where they currently have no presence.
What would you have Sun do?
WEB would be my choice
* Netbeans - too long to go. Imo eclipse beats it almost enywhere (except Ruby and som nice-to-have features) (imo)
* Web .. Maybe it isn't the best choice.But it seems to be rock-solid one.
I think Sun should go safe way now. Will wait and seen what they actually choose.
no silver bullet, but an opportunity not to miss
That won't of course sustain their whole company, but since you ask, I will have told you about a nice opportunity to be taken IMHO:
hell Sun should continue the effort (I just started for fun) to put OpenERP, a brilliant open source ERP that doesn't suck (yeah you heard that) on Jython (e. g. Java) and wire it up to a Java standard servlet container. Then it should try to get somewhat associated to its rising success in large organizations. With nothing less than SAP and Oracle to defeat, there is some large profit to be made.
For long I compared open source ERP's, I found nothing approaching the OpenERP v5, I mean by a fair margin.
New doc provided here for curious people: doc.openerp.com/ and bright new web site expected to shine in a week.
I won't bother proving it to you here (but I've done a 6 months+ comparative study available here www.smile.fr/publications/livres-blancs; disclaimer it had to be in French), so if Sun and other Java players are interested in an open source ERP in really good shape to take the market by storm, they should get their hand dirty, try it, watch its community and think about it. From what I know it's not on sale, but the big players that will bring OpenERP as a standard war container much like JRuby on Rails applications to large organizations (possibly in a scalable Java cloud if you like) will IMHO see some profit. Actually I learned myself that JRuby on Rails was not only portage to be made (while it's probably the smartest one, nothing too foreign to Tim Bray from what I know ;-).
Oh, my modest contribution to that (I'm serous, the amazing stuff is the ERP itself, not running it on Jython):
PS: ah yes, I would have loved too the Java client to success (be it JavaFX or not actually, could have been in 2003 BTW), but I doubt this can happen now: the benefit over DHTML/Ajax running every day faster is not enough to justify a technology shift and Adobe has to many market shares for complex media/animation (while their VM suck, especially outside of windows). I think only very bad deployment strategy of Adobe runtime on next generation mobile phones (like the iPhone) and very good commitment of Sun (unlike the Google Android Davlick thing) could make it happen.
Sun is still a platform vendor
William El Kaim
For me Sun still position itself as a platform vendor.
Sun builds and sells hardware, of course. He also provides in its catalog: an OS, a JVM, an application server, an integration platform and an IAM platform (and soon a portal platform based on liferay). So all stacks needed to run enterprise applications.
So moving to virtualization, cloud computing and PAAS is natural.
Where Sun is not good at is to segment its customers and to see where to earn money.
Clearly he can play a very important role in selling to datacenters, hosting providers, etc. As far as I know, it still lacks a good ESM suite with ITIL support. Sun should also provides production ready appliances with its own product running inside.
The questions are now:
- is Sun good at selling to development teams. Without launching a discussion concerning Eclipse vs. netbeans, I do think, he should continue providing dev tools. Could be great if eclipse and netbeans could re-use a "compatible" plug-in API. But Sun will not earn money with this.
- is Sun good at controlling and moving forward Java language? Like for UML V2, Sun should concentrate on offering a core lightweight and semantically well defined Java kernel. Then, he should leverage all others framework and API on top of it. He should not try to integrate them as much as possible (avoiding war like JPA vs. Hibernate, etc.). I do not think Sun will earn money with this also. But it is its contribution to the Development community and that makes Sun what it is today.
- Its JBI and OsGi support in the new glassfish will make this platform more and more interesting in the future. Java EE6 with profiles will also help the modularity. Then Jersey and Project fuji should deliver very good result this year.
And finally, Sun should not intertwin its R&D work and community based work with full fledged enterprise product. Both have different lifecycles and fit different needs. OpenSSO is the perfect example of this unclear strategy.