Portals Are Operating Systems

February 22, 2010 by John Petersen

In discussions about how the web has affected the operating system business, the browser is generally considered to be the heir to the desktop, and Google the primary pretender to Microsoft's throne. While web technology has no doubt undermined Microsoft's ability to dictate what the preeminent application environment of the future will be, most analysis of the ongoing operating system wars limits itself to client-side technologies and the consumer market. Yet a significant front is being fought on the server side and in the enterprise — the under-appreciated battle to determine what will become the dominant enterprise portal platform.

The web OS of the future may not be called a portal. Naming trends go in and out of fashion, and though there are recent indications that portals are back, portals are cool again, and portal software is enjoying a new renaissance, there is also a vendor-driven trend towards rolling portals up into broader multi-category middleware brands like Microsoft SharePoint, Oracle Fusion, and IBM WebSphere. (Aside to Open Text: Why are you retiring the name LiveLink, which had great potential to become one of these broadly applicable brands?) But the vagaries of naming and marketing don't change the fundamental game: to try to own the server-side web application hosting environment that most business software vendors and developers code to, integrate to, and deploy to. Whatever its name, this new server-side operating system will functionally resemble an enterprise portal.

Vignette Portal Tapestry
Source: Cathy Miller

Multiple maturing technologies — Linux, mobile, virtualization, the cloud, and overarching all, the web itself — have combined to commoditize the traditional operating system. On the client, the web OS has always existed within a browser, but on the server it has migrated up the stack from web servers to application servers to portal servers, each layer adding new infrastructure and services previously considered non-core. Like sediment, the surface sands of software gradually sink, stabilize, and calcify into bedrock as new layers press down from above. This upward evolution has affected repositories: the RDBMS has now sunk below the surface of relevance, displaced from above by the CMS. What has emerged as the de facto operating system in many organizations is something that former Gartner analyst David Gootzit called the "portal fabric" — a web integration layer of interconnected portal servers (increasingly a heterogeneous mix from different vendors), stitching together all of the organization's software into a unified, personalized quilt.

The threads of the portal fabric will bind to any imaginable web-enabled software or API. For fast, lightweight integrations there is widget/gadget/mashup support, exposing and embedding UI snippets via DHTML. Deeper integrations open up with portlets and web parts, which can programmatically interact with a vast array of libraries, APIs, and SOA services. In the Java world, the JSR portlet specifications provide a pluggable, standards-based approach to portlet development. The under-appreciated JSR-286, in particular, represents a major step forward for portal technology. Prior to this spec's adoption, it would have been difficult to call portals true operating systems; but with it, portlets developed independently by different developers or vendors can now communicate with one another on the same web page. JSR-286, especially in combination with AJAX, brings portal pages to life in a way that rivals the desktop, both visually and technically. WSRP is another important portal standard. It has not seen the same level of adoption as the JSR portlet standards, but it should get a boost with SharePoint 2010's support for bi-directional WSRP 2.0. WSRP provides a great way to federate different portal products in increasingly common heterogeneous portal server topologies.

Portal Magic Quadrant
Source: Gartner

Microsoft understands that the operating system is moving up the stack in the enterprise and its solution is SharePoint. After many years of playing catch-up, SharePoint is now a category-busting front-runner across the entire Portal, Content, and Collaboration (PCC) middleware landscape. Taking a page from the MS Office playbook, the SharePoint team rolled mulitple products into a single suite; and while no specific component of that suite can be considered best-of-breed, the combined integration is indeed greater than the sum of its parts. SharePoint's portal component still leaves something to be desired relative to its competitors, both functionally and in terms of lock-in. Without a web part standard as open and portable as Java JSR portlets, there is no viable path for migrating SharePoint web applications onto non-Microsoft server platforms. But that is hardly surprising given that Microsoft of all vendors understands best that portals are operating systems. No-one expects to move their Windows applications onto other operating systems either.

Closing in on $500M in revenues with WebSphere Portal, IBM is the single-product portal market leader. WebSphere Portal evolved rapidly through the mid-2000s, but the pace of innovation has stalled recently on its 6.x series (7.0, though delayed, should launch soon). Yet it is still a highly capable offering with strong support for open standards. Where Microsoft built SharePoint to protect and extend its operating system business, IBM built WebSphere Portal to protect and extend its application server business, which was becoming commoditized at the hands of pure-play portal vendors. IBM has essentially vertically combined its application server with its portal server, and now seems poised to horizontally combine its portal and WCM offerings. Given its painful history with MS-DOS, Windows, and OS/2, IBM knows what is at stake here, and is fighting for a second chance at operating system hegemony.

The next strongest competitor in the portal market is probably Oracle, although with five portal products on its hands now — two homegrown, two as a result of the BEA acquisition (BEA's homegrown product, plus ALUI, formerly Plumtree), and at least one as the result of the Sun acquisition — Oracle Fusion is an integrated middleware suite in name only. There is a lot of great technology kicking around in that portfolio, and the combined market share of Oracle's portals rivals that of IBM WebSphere Portal at almost a third of the market each. However, it will take time to digest these disparate products, and in the meantime Oracle's market share will represent the best opportunity for challengers to make inroads.

Liferay is a great open source success story and probably represents the Linux of the emerging enterprise web operating system. Liferay has clawed its way to industry analyst endorsed respectability over the years, and their portal is now a legitimate alternative to the major proprietary offerings. However, they can't rest on their laurels, even in the open source world, with worthy competitors like GateIn, the newly combined JBoss / eXo offering, and Hippo Portal, built atop a battle-tested Jetspeed core. I'll have more to say about Liferay in a future post; suffice it to say that I have great respect for both the business and the product that the Liferay team has built over the years.

The most interesting portal player to me is of course Open Text, inheritor of Vignette Portal, formerly Epicentric Portal — a storied product lineage that goes back more than ten years to the dawn of the enterprise portal. Vignette's recently departed upper management never really understood the value of this asset, and for several years put it on the back burner, depriving it of resources. But a skeleton crew of believers fought against the tide internally to continue to improve the product, and today it remains feature for feature one of the best. I give Open Text credit for seeing more strategic value in the product than Vignette did. They seem to be moving to make the portal the web front-end for a number of other products. But I still believe that Open Text is under-valuing this important asset. They aren't thinking big enough. Granted, if Open Text were to recognize the rare operating system play that this product represents and decided to pursue it, that could undermine their own efforts to remain a neutral Switzerland amid warring giants, and could put them more directly into conflict with longtime partners like Microsoft and SAP. But the brass ring in this case is worth it. Open Text, you once held a critical strategic asset that you failed to fully appreciate. Don't let that happen again.

Vignette Portal Development Team
Epicentric / Vignette Portal Development Team, November 2003

The last time operating systems were in a comparable state of flux was in the late 80s and early 90s during the transition from console-based operating systems to graphical desktop environments. Microsoft emerged victorious then by focusing on enabling developers and on providing the best business applications. When it became clear that far more applications were available on Windows than on Macintosh, including all of the applications one needed to do one's job, the consumer market came around and the Windows dynasty was born. If during the current historic transition from desktop to web-based operating systems, a dominant enterprise portal emerges by similarly winning over developers and hosting the most critical business applications, it could pivot into the consumer space by offering hosted services directly to end users via private clouds. At that point the distinction between consumer portals (My Yahoo, iGoogle, etc) and enterprise portals would dissolve and the stage would be set for a new Windows-like dynasty. Google is betting on that kind of a future with a host of portal-related initiatives — App Engine, Google Sites, iGoogle, Wave, etc. — several of which it is already attempting to cross over into the enterprise. Their offering is still patchwork and thin compared to the established enterprise portal products, but Google obviously cannot be discounted any more than Microsoft could a few years back before SharePoint caught fire.

More than a decade after Plumtree and Epicentric pioneered the enterprise portal market, the game is heating up again. All of the world's largest software companies are competing, along with a handful of upstarts. The stakes couldn't be higher. I'm lucky to have played a small part in the invention of the enterprise portal, and I hope to help build the web operating system of the future. I can't wait to see what happens next.