Tuesday, March 27, 2007

We need a portal standard, not just a portlet standard

As I wrote to Punit Pandey (who runs this excellent blog on portals), I wonder at the fact that although so much work has been done on defining portlet standards, we do not have anything like a portal standard. A large part of a portal/portlet application has to do with menu design, page layout, portal themes, portlet skins, etc. These are not covered by the portlet spec and are left to individual portal servers to implement in their own (read: proprietary) ways.

Quite obviously, this impacts the portability of applications. I can trivially port a web application from one server product to another by just deploying a .war file, but I cannot do the same for a portal application. I would need to rebuild the portal pages from scratch and can only reuse the portlets in it. I can't see the portal vendors wringing their hands over this lack of portability. That's how they lock in their customers even while claiming to be standards-compliant.

I believe we need a portal standard in addition to portlet standards JSR 168, JSR 286 and WSRP. But we can't depend on vendors to kickstart the process. It's the users who need to set up a JSR to hammer out a portal standard, because we're the ones who lose from the absence of one.

Friday, March 23, 2007

The worst time in history to choose Microsoft

I've just heard of a company (which shall remain nameless) that has decided to move to Microsoft for their SharePoint collaboration stack, and that pretty much implies that they will be using Microsoft technology pervasively throughout their organisation.

Why, I asked my contact.

Because we've debated this for too long and we need to make a decision and move on, he said matter-of-factly.

I would normally accept this pragmatic wisdom, but I could not help raising an important point. Isn't this precisely the wrong time to be moving towards Microsoft, I asked, when the story in the non-Microsoft (read: Open Source) world has never looked better?

To pre-empt the usual "corporate" arguments against Open Source, I pointed out to him that much of the Open Source software seen today is in fact erstwhile commercial software that has been newly released as Open Source. Examples are OpenOffice.org (ex-StarOffice), Firefox (ex-Netscape), Fedora Directory Server (formerly Netscape Directory Server), Ingres, TerraCotta (clustering software), Eclipse, etc. What's more, the stream of goodies doesn't seem to be ending. Java itself will be Open Sourced this year. So will Sun's portal server. Why, so will Microsoft's FoxPro!

Third parties are rushing in to build integrated stacks from these excellent software products, and offering support at a price. Think Unisys and SpikeSource.

When so much proven software is "falling off the back of a truck", is this organisation's timing wrong in choosing this point in history to look elsewhere?

The software industry's very model seems to be evolving towards "free software, paid services". Shouldn't they be targetting this future instead of the old model of software license fees?

I don't know if my arguments will make any difference to this company, but articulating them has strengthened my conviction that this is the worst time in history to be choosing Microsoft, or any proprietary software vendor, for that matter.

It's like formulating business and/or IT strategy in 1994 without taking the Internet into account. Two years later, people would look at you and go, "What were you thinking?"

We can see the trend towards commercially-supported Open Source taking shape before our eyes today. If we lock ourselves into a multi-year licensing arrangement with Microsoft today, will we end up looking foolish tomorrow when the trend becomes the norm?

Monday, March 19, 2007

Sun should make a comeback this year

Jonathan Schwartz's elevation to CEO has been good for Sun so far, and I suspect it will prove to be even better in the days to come. In contrast to his predecessor Scott McNealy's on-again, off-again style (especially in relation to Open Source), Schwartz has brought a consistency and focus to Sun's actions. Finally, true to the words of a 1990s Sun ad, Sun once again has "all the wood behind one arrowhead."

The nature of the software industry is shifting, with Open Source gaining more legitimacy and services becoming the thing that customers pay for. Sun is moving to this model, too. In the process, they're beginning to give their competitors quite a few headaches.

You've read about Sun's Java and OpenSolaris plans. The latest news from Sun concerns their portal server. The portal space has been one where Open Source products have so far lagged behind their commercial rivals, mainly because of ease-of-use issues and imperfect integration with content management systems. It will take Open Source portals like Liferay and JBoss Portal more than a year to become "good enough". Sun has now weighed into this unequal battle, and will open-source their portal server. True, it hasn't happened yet, but the announcement has been made, and when the code is cleaned up, the product will be released. I'm hoping it will be under the GPL3.

Sun has a bit of housekeeping and tidying up to do before it can really boast of a full, consistent stack of Open Source products, but I'm confident it will get there.

On the client side, they really need to give OpenOffice.org a concerted push to get it over the line. OpenOffice.org has been at version 2.1 for ages now with no apparent progress. Its interface stills lags Microsoft Office and its performance and memory efficiency are features only a mother could love. No one but Sun can fix this, and they really need to do this quickly. There are many supporters of OpenOffice.org, but support could quickly evaporate unless the product performs.

On the server side, Sun should concentrate on two things:

1. Deprive their competitors of oxygen by providing full and credible stacks for most enterprise components, under a consistent licensing scheme (I'd recommend GPL3). That means GNU/OpenSolaris, Java, the GlassFish app server, Tango (the Microsoft-interoperable Web Services stack), Portal server and third-party Open Source software like Fedora Directory Server and Ingres. (My pet peeve: Forget the dalliance with PostgreSQL and get serious about a real database - Ingres. That's a GPL-ed product too, by the way.)

2. Build up a much stronger services arm or tie up with a dedicated services company (à la Accenture but with morals) and spend serious marketing dollars to get business.

Sun has a breathtaking opportunity to regain the initiative this year with bold moves that their competition cannot readily respond to. IBM, HP, Microsoft and Dell, all can be forced onto the back foot if Jonathan Schwartz plays his cards right.

Let me quote an old Irish blessing:
"May you always have work for your hands to do
May your pockets hold always a coin or two
May the Sun <grin> shine bright on your windowpane
May the rainbow be certain to follow each rain."
I'm waiting, and watching, and hoping.

Disclosure: I don't own Sun shares. I own SGI shares, and SGI has been delisted from the NYSE. Just my luck.

Wednesday, March 07, 2007

Desktop Predictions for 2007-2008

This has been stewing in my mind for awhile, and now the picture has crystallised. Here's what I predict will happen in the desktop market over the next two years:

1. Apple will license Mac OS X to OEMs this year (2007) while Vista sales are still sluggish.
2. Mac OS X sales will surge wildly and end up accounting for up to 25% of all new PC shipments by the end of 2007.
3. Emboldened by Apple's success, Sun will push GNU/OpenSolaris to corporations in 2008. The system will probably be called "The Liberty Desktop".
4. Linux will be seen as "ready for the desktop" once desktop diversity becomes the norm. Linux will dominate the low-cost desktop segment thereafter.

I don't believe Apple's management is living under a rock. I'm sure Steve Jobs is planning his next big smug announcement already. It'll be the news event of the year, but you read it here first. After years of tight control of its OS, Apple will finally realise the Zen paradox of gaining by letting go.

After all, if pre-installed Linux is top of the wish list for Dell's customers, what do you think the demand for pre-installed Mac OS X will be, once customers realise that's a real possibility? Midnight queues for Mac OS X will beat those seen for Windows 95, mark my words.

This is Apple's big chance. Windows as a desktop platform has plateaued in features, and Vista relies more on momentum and lack of alternatives than any compelling reasons to upgrade. There is a vacuum here, a hunger, a latent demand for something different and good. The OEMs know it too, and have been pushing Apple for this for years. Well, this year, they may actually get it. (If Apple is smart, the OEM price won't exceed $50. You need to seed the market, guys).

Once Mac OS X for PCs hits the shelves, the resulting media noise will drown out all other events for the year, including significant ones like the GPL releases of Java and OpenSolaris, the release of the next generation of Linux distributions (with Ubuntu-Linspire's upgrade advances) and further decreases in the price of hardware. But make no mistake about it, these events will impact the desktop market in 2008.

Sun is not sitting idly by. CEO Jonathan Schwartz has exceeded my personal expectations and is proving to be far more savvy than his predecessor, Scott "dinosaur" McNealy. I'm sure he sees the potential for Sun's resurgence, and Apple will open some doors for Sun.

You see, although Apple will take the consumer desktop market by storm in 2007 with freely licensed Mac OS X, the corporate market is deeply suspicious of Apple. Apple does not have a great reputation for reliability as a corporate supplier, having proved too fickle in the past and left enterprise customers in the lurch. The Apple desktop revolution will stop at the gates of the enterprise.

That's when I predict Sun will make its move. It will also coincide with the technical maturing of GNU/OpenSolaris as a desktop OS. Watch for Sun to exploit its links with Java, thin clients, SmartCard technology and the Liberty identity management system to push a compelling, low unit cost desktop solution to enterprises. GNU/OpenSolaris will ironically be running on virtualised servers, but visible to users through low-cost thin clients. SmartCards will provide roaming capability (among other things) and Liberty will provide a federated identity system. In my professional opinion, such a system would be architecturally elegant and also easy to justify from an infrastructure investment point of view. Compared to the hardware upgrades required to roll out Vista, this would be a far cheaper option.

For Sun, this solution would really be the Java Desktop System 2.0, but something tells me that "The Liberty Desktop" would go down better. I think the whiz kids in Marketing will finally call it that.

What does all this mean for our own dear Linux? (It's our own because we all own it, silly, even if you don't believe it. Think inclusive ownership.) Will the long-awaited Year of the Linux Desktop pass it by for a bunch of Johnny-come-latelies?

I believe Linux as a desktop OS has been asymptotically approaching readiness for years now, and its capability gaps have all but vanished. At this level of granularity, there are already market segments for which desktop Linux is a perfectly serious option. The major factor holding back mainstream Linux adoption is psychological. Nobody uses Linux on their desktop because nobody else does (Geeks don't count).

But when your PC salesman begins to ask you which OS you'd like with your new PC (Windows or Mac OS X), and you see the price of the OS as a separate line item on your bill, and compare that to the cost of the various hardware components, that's when you would start to ask about cheaper alternatives. The friendlier salesmen would point you in Linux's direction, even if their employers don't offer support for it.

And that's where I believe Linux will find its sweet spot -- the price-sensitive low end of the desktop market. That is also the volume market, so after 2008, the desktop market will have four major players:

  1. Windows (still dominant in terms of installed base, but perceptibly losing share)
  2. Mac OS X (the media darling with 25% market share of new shipments and the "coolest" OS to own)
  3. GNU/OpenSolaris (whatever it's actually called, making strong inroads into corporates)
  4. Linux (at the low end, steadily growing, and as always, with adoption impossible to estimate because it's all under the radar)
Diversity, I salute you.