Oracle’s analyst summit in mid-June provided a good look at their plans for Fusion Middleware 11g and WebCenter (released July 1st for download; see summary of features here). Now that we’re out of non-disclosure mode (and into “please disclose!” mode) I’d like to share my high-level impressions. They covered a ton of stuff, but my view is biased towards my coverage area of portals with connections to search, productivity, and collaboration. Other Burton Group analysts were also in attendance from our Identity and Privacy Strategies team and our Application Platform Strategies team (see Anne Thomas Manes’ thoughts here).
First, although Oracle owns 4 portal products, all the portal-related time was spent on WebCenter. Sure, other portals were mentioned in bullets as examples of how they can plug in (or consume WebCenter’s social software), but it was clear WebCenter is the leading actor here (and supporting actor in the stories of the SOA, identity, and enterprise application teams). This confirms what I (and Oracle) has been saying: that WebCenter is the primary portal and that the other 3 (Oracle Portal, WebLogic Portal, and WebCenter Interaction née Plumtree) will be supported and have their die-hard fans but will not be best for new portal projects.
It was helpful to hear Oracle frame its collaboration/portal/search/productivity/social software ambitions in relation to Microsoft SharePoint. For all its plusses and minuses, SharePoint provides a common point of reference against which to measure. They described how they line up with SharePoint as an alternative, can coexist with it, and where they surpass it. This is what IBM should have done with Quickr+Connections at Lotusphere.
As with SharePoint, WebCenter provides an impressive set of functions in one box. There is often better integration between WebCenter and other Oracle assets (like their applications and development tools) than Microsoft where other groups can sometimes get away with ignoring what the SharePoint and Office group does.
There are numerous SharePoint analogies in WebCenter. From conversations with the execs there I found that some are intentional and in other cases they say SharePoint copied them (well, copied AquaLogic User Interaction)!
- Business Dictionary as a role based catalog of information assets. Seems like SharePoint’s Business Data Catalog. This should be an interesting battle since SharePoint’s BDC is clearly a version 1.0 work-in-progress and Oracle has a lot of expertise to bring here being a database company at heart.
- Federated search. ‘Nuff said.
- Office integration. Clients I speak with expect Microsoft will always have the best Office integration, but there are cases where Microsoft’s internal silos or some good ideas can expose openings. Oracle showed a nice Word sidebar for document management that had people, versions, etc.
- Slide sorter. This was a neat feature that SharePoint offered, but Oracle’s version seems to leapfrog it. They demoed picking all the slides for a sales deck. Oracle calls this a “folio” or compound document. Oracle acquired a neat little company called “Outside In” that has sophisticated filters for productivity files. Blending this into Web Center can provide for some good Office integration.
Oracle did a fine job of acknowledging the need to work with SharePoint and others. But the meat boils down to their WSRP producer running on .NET, selective metadata consumption, and Ensemble (a reverse proxy solution). Hopefully this gets beefed up with more programmatic integration, discovery tools, and guidance so it requires less reliance on WSRP.
Of all the competitors, WebCenter is the newest architecture from the ground up. Being the youngest has its advantages. Since WebCenter is newly architected it feels like it more seamlessly integrates new concepts like tagging, linking, social connections, and REST services than IBM and MSFT where it’s more bolted on. So they’re better at utilizing these features across the suite that Microsoft and a little bit better than IBM.
But will Oracle – the whole company – give WebCenter the resources it needs to win the marketplace(not just the resources required to be a good and useful product)? In the Q&A session, Oracle President Charles Phillips said there are “No plans to have middleware broken out in reporting. We have lots of product lines, we’re getting more with Sun… ” This hits at the perennial knock on Oracle’s efforts around knowledge infrastructure – lack of push and commitment. Oracle did talk about how much revenue Fusion pulled in, the growth rate, penetration, etc. That would indicate the company would have to care. But still, Microsoft manages to report on four breakouts (Client, Server and Tools, Online Services Business, Microsoft Business Division, Entertainment and Devices Division). Oracle sticks to two (Applications, Database and Middleware). Sun will add at least one more (servers and hardware). If Oracle is dedicated to the enormous space between enterprise apps and the database, breaking out middleware from the database would be a great way to track and prove this commitment.
Note: This is a cross-posting from the Collaboration and Content Strategies blog
A belated congratulations to the JSR 286 group, which went into final release in June. You can get the details here. Good job to Stefan Hepper of IBM, the specification lead, who must have had a tough time herding the cats on this one. Right out of the gate it’s good to see JSR 286 getting some attention from the portal vendors.
Of course it’s been in development for quite some time. When JSR 168 was being created a number of enhancements were shifted to JSR 286 to keep from slowing down the original portal spec’s ratification. The most important features of JSR 286 are inter-portlet communication (IPC) and alignment with the ongoing work on Web Services for Remote Portlets 2.0 (WSRP 2.0).
Vendor support has been promised soon as well (or is already here in the case of IBM).
- IBM quickly announced support in for WebSphere Portal 6.1 (as well as for WSRP 2.0). Way to track the standards guys!
- Oracle’s ALUI and Oracle Portal don’t have support yet. I’m told Oracle Portal may get it in 11gR1 release or later
- JBoss (Red Hat ditched its own portal in favor of the JBoss portal after acquiring JBoss) is promising support in Version 2.7, due out in Q3 2008 according to CMS Watch.
- A document from someone at SAP coldly stated that “SAP supports and actively participates in this new standard as EG member. As this specification is in draft version, it is not supported in NW CE.” I checked with my contacts at SAP and was told that JSR 286 is currently in scope for NetWeaver 7.0 Enhancement Pack 2, which is expected around October. I also noticed they did not vote on the standard. I wondered if this was a sign of lack of interest or a passive way to vote “no”, but their analyst relations staff said it was in fact because their committee representative was in the hospital.
My guidance to portal architects though is to consider bypassing JSR 168 or 286 portlets and even WSRP and focus on creating web services for information that they want to expose. Once the data portion is available as a web service, any decent portal product can create a quickie portlet (maybe in JSR 168 or some proprietary format) from the WSDL of the web service so it can be used in a portal. And then you have the option of leveraging that interface through non-portal mechanisms as well.
I should mention JSR 301 here as well. Presentation models have changed since the original JSR 168 specification was created. JavaServer Faces (JSF), the most significant of the presentation models for Java and has spawned many proprietary and open source attempts to transform JSF interfaces into JSR 168 portlets. The JSR 301 specification is only in its early stages, but promises to standardize these bridging mechanisms. No solid word yet on what that’s coming though.
I got a chance to talk to Oracle yesterday about how they plan to integrate the BEA middleware assets they picked up in their acquisition into their own middleware portfolio. Oracle let it be known that – no surprise – Oracle strategy is constant and BEA integrates into it. Fusion Middleware is the brand and it’s all about Fusion in this integration strategy. Dictionary.com defines fusion as “a nuclear reaction in which nuclei combine to form more massive nuclei with the simultaneous release of energy.” That seems to be what is happening at Oracle these days as the nucleus of Oracle and the nucleus of BEA (which had already combined with the nucleus from Plumtree) combine to form something pretty massive. And there is certainly a lot of energy being released, so that definition certainly applies.
Still, Oracle assures BEA customers that all products continue under “existing BEA support lifetimes”, there’s no forced migration, and license costs are grandfathered for existing customers. In fact, some support costs may come down since Oracle policy is to price support as a percentage of net price rather than list, but others could increase or decrease a bit since Tuxedo’s pricing tiers get remapped to CPUs.
I won’t comment much on development since that isn’t my coverage area. I will say that JDeveloper is still the flagship development platform for Oracle. BEA Beehive is just in maintenance and Workshop is a freebie in the Eclipse Pack.
And now, the answer to the portal conundrum I wrote about in my “Four Portals of the Apocalypse” posting when Oracle announced its intention to acquire BEA. For portals, as expected, the winner is WebCenter. It’s not that the others are dead, but WebCenter is the “hot” product they want to talk about first, connect everything to, and anoint with all the cool, buzzword-compliant enhancements. Users of the other portals (Oracle Portal, BEA WebLogic Portal, AquaLogic User Interaction aka Plumtree) need to figure out how soon a rewrite is going to be in their future since those products are in the “continue and converge” category (C&C). A C&C portal will keep going forward for existing customers for quite a while (Thomas Kurian publicly stated the lifecycle would be about 9 years), but will not have new customers steered towards it and will eventually be merged into WebCenter.
So, here’s my recommended strategy based on which Oracle portal product you’re using based on what I know so far. I’ve been told that more detailed migration plans will probably come out in 2-3 months:
- Oracle WebCenter: You lucky dog! You picked the winner. It’s a rosy future for you, full of the best piece parts from other portal products, new Enterprise 2.0 functionality, and you’ll meet lots of new friends at each annual WebCenter user’s group meeting. For a new portal project being planned, WebCenter is the only reasonable choice in the Oracle portfolio unless you’re into planned obsolescence.
- Oracle Portal or BEA WebLogic Portal: You’re probably OK coasting along as is unless you fall into one of a few categories. If you’re really thinking of adding the newest functionality (Enterprise 2.0) and architectural standards (REST, RIA) you’ll want to start thinking about migration soon, although Oracle intends to have some WebCenter services plug into WebLogic Portal, Oracle Portal, and ALUI to be leveragable without migrating. And if you are in the rare situation of having a strategic portal with a long lifespan expected ahead of it (5+ years), you’ll have to make a reminder for yourself in 2010 or so to start thinking about migration.
- AquaLogic User Interaction: The picture for ALUI users is pretty similar to that of Oracle Portal and BEA WebLogic Portal – it’s in the C&C category so expect it to be supported for quite a while. Still, it’s my personal opinion that Oracle will have a harder time with ALUI since it will be chopped up into more pieces and there are lots of legacy installations with deep customization. Also, ALUI has a .NET side to its heritage. In the Plumtree days they had spent quite a bit of effort on building out .NET support. For example, the Enterprise Web Development Kit (EDK) could be installed with either a Java or .NET (C#) development and there was a .NET version of the EDK is available as a dynamic link library (DLL). The .NET portlet creation capabilities are going to be rebranded as the Oracle WebCenter .Net Application Accelerator.
Oracle is rolling out a few new SKUs (packages) that will help portal owners.
- WebCenter Services combines the WebCenter Framework with some Oracle pieces (BPEL Worklist, their portlet bridge and JSR 168 container) and sprinkles some ALUI and WebLogic Portal pieces in. Ensemble is renamed Oracle Ensemble and put into services SKU and Analytics gets Oracle branding.
- WebCenter Suite has everything in WebCenter services as well as restricted licenses for content, Oracle Presence, BPEL Process Manager, and search. ALUI is renamed WebCenter Interaction and goes in the suite for now, although as of 11g there’s nothing in ALUI that they’d recommend for new deployments. AL Collaboration is renamed WebCenter Collaboration and goes in the suite until they can roll out new collaboration in WebCenter 11g.
I was a fan of the BEA Pages and Pathways products, both of which will melt into Fusion Middleware. Pages melts into the WebCenter Framework. It will be part of WebCenter Composer for users to create mashups and its wiki and blog capabilities go into WebCenter Services in the 11g timeframe. Pathways melts into two separate places. Its social search merges into secure enterprise search and the social tagging merges into the 11g foundation.
All in all, this roadmap is pretty complete and as good as owners of Oracle Portal or BEA WebLogic Portal could have hoped for. Owners of complex, customized ALUI portals need to see the writing on the wall and plan to migrate to the new WebCenter model (method TBD by Oracle) or re-architect off the Oracle platform entirely. But one more telling statement about nuclear fusion may shed light on Oracle’s strategy of unifying all its pieces under the Fusion umbrella. Wikipedia notes that “Artificial fusion in human enterprises has also been achieved, although not yet completely controlled.” Ah, yes, how true. We still have to see how well this bit of fusion winds up being under control over the next few years.
Note: This is a cross-posting from the Collaboration and Content Strategies Blog
I have been speaking on web and portal governance for 5 years now, and I find it difficult to provide clients with actual examples. It’s not that they aren’t out there, but that presenting on governance involves airing dirty laundry and most organizations don’t want to do that. I recruited a speaker that gave a case study on their portal governance for a conference a few years ago, and it took a lot of searching to find that example (they were a public utility which helped). That’s what made the portal governance presentation I saw at BEA Participate especially rare.
The speaker was Jackie Jajdzik, team lead at Weyerhaeuser. I would like to be able to claim I helped her with her governance strategy, but we’ve never spoken and yet she’s done pretty much all of what I recommend a governance process consist of (hear my podcast on portal governance here). This includes deploying incrementally, putting it into a concise document, making the governance easy to find on the website, gathering and using metrics, and creating feedback loops.
I’ll boil down the approach here. I apologize for any incorrect paraphrasing – these are my categories not hers. But these categories are what I tend to listen for from a client when discussing governance and Jackie had good answers to all of them – I was quite impressed.
Problem statement: 1600 websites, 500mm web pages accessed by 30,000+ users, no common access point, no governance and search indexed everything (for example, “benefits” returned myriad hits, confidential information would show up like resumes and disciplinary actions). Anyone could customize and there were a few standard portlets and communities. Users had total freedom and no standards to tie them down, and yet were unhappy. Key messages (safety being the most important to them) were lost, liability was an issue since everything was searchable and could show up on a home page, and support costs were growing.
Goals: Reflect unified company, manageable number of sites and communities, decrease management costs, easy to navigate, increased productivity. It was obvious from the presentation that these goals were derived from actual pain points and created with buy-in, not just picked from a list from some paper on how to do governance. She talked about how they received executive buy-in that was useful when they had to clamp down on formerly ungoverned sites. Also, they needed to borrow some staff from the business to trim the sites.
What they did: The homepage was reorganized based on usability studies and analytics. Deleted 400 sites, applied navigation to 120 sites, put limits on search. Their governance includes a concept she called “zoning”. Zoning determines which technology to use from 3 options (community/project, conventional website, SharePoint). For example, if you want it indexed by search or have more than 200mb of storage you don’t use SharePoint.
Organizational structure: Executive sponsor, public affairs / intranet mgr, standards and operating committee with operational support from IT and their library (taxonomy, search engine hinting) + task teams. Governance is looser for sites with smaller scope and tighter for enterprisewide sites.
Metrics and feedback loops: They do regular audits and have a site registry to track sites and classify them. There are quarterly meetings of the standards and operating committee. Satisfaction surveys track what users like and dislike, request features (biggest request is for social computing like blogs, wikis, and RSS), and how often they use it (don’t trust weblogs alone for this information).
Outcome: Resulted in more root site usage, less complaints about finding information, less conflict with the business, information is more current.
Now if I can just find those public examples of written governance documents I get asked for too …
This is the first of several posts I’ll be doing about the BEA Participate conference that happened this week. For my first subject, I’ll focus on the biggest issue for me this week: What light does this conference shed on how BEA and Oracle will mix?
This was a strange time for a BEA conference, coming on the 2 week anniversary of the closing of their acquisition by Oracle. There weren’t a lot of balloons or cake to celebrate the acquisition – it was very quiet (as required by law and quarterly reporting deadlines). For example, Mark Carges (EVP Products and GM, BEA) kicked things off and didn’t quite seem up to referring to BEA and Oracle as “we” yet. There was simply a dry reference at the beginning to “Our new owners, who you will meet later”. After the one quick reference, the rest of the presentation was BEA business-as-usual.
The second presenter was indeed “the new owners” in the form of Hasan Rizvi, VP of Fusion middleware at Oracle. He showed a chart of all their middleware products and saying (I’m paraphrasing here) they are “best of breed, but of course BEA also has best of breed products in all these categories so that’s why this is such a good combination”. He said they will be doing “Welcome BEA Customers” events at 25 cities in the US/Canada and 25 in EMEA.
He introduced the BEA crowd to Oracle Fusion middleware and their tools. The crowd didn’t seem very partisan, was attentive and soaked in the information. From my 7 years (on and off) of going to BEA conferences I can say that, like past BEA conferences, the vibe is one of a mature, techie environment. While JavaOne may have techies lounging on beanbags, playing videogames, and eating kiddie snacks while challenging each other to coding duels, the BEA conference attendees are mature programmers, in the second or third decade of their careers, confident in their abilities, who tend to understand the value of well architected systems. They like BEA although they don’t need expensively produced videos making fun of their competition and they don’t paint their hair in the company colors (both of which I’ve seen at other vendor conferences).
It’s appropriate that the Oracle and BEA colors are within a few Pantone shades of the same red since the company colors don’t represent a religious issue like it would be in some other acquisitions (the blue and purple devotees of Microsoft and Yahoo! wouldn’t have mixed as easily). However, the BEA audience is technically adept and will require practical reasons and detailed roadmaps if they are going to buy into new Oracle+BEA solutions rather than shifting their development platforms, application server, portal, and BPM to IBM or open source. I can’t wait to see these roadmaps, particularly in my area of portals, since there are many overlaps that will require sacrifices to resolve. The sacrifices will take the form of placing some products and customers on the sacrificial altar or placing profits on the alter due to the long-term inefficiencies of maintaining redundant products.
And, just to prove that in the end some people always stay loyal, I counted about 300 people in the packed “What’s new with AquaLogic User Interaction” session. When the announcer asked “how many of you still call ALUI Plumtree?” about 75% of the audience raised their hands.
ZDNet writes today that Oracle is making a bid for BEA for the demonic price of $6.66 billion. Well, the devil is in the details and there are plenty of them to hash out if this deal ever flies. There is a large degree of overlap in the application server, development tool, collaboration, and portal spaces between BEA and Oracle. At least BEA doesn’t have a content management offering too, so Stellent is safe for now. I find it rather ironic that after writing in May about “How Many Portals Should a Vendor Offer?” that both Oracle and BEA have two sets of portal products on offer, now Oracle could have four (4!) portal products in its portfolio. I’d love to be a fly on the wall as they try to rationalize that one!
There have been rumors of Oracle (or others) buying BEA for quite a while, but it hasn’t happened yet. For example, in May of 2004, Java.net reported “Faced with a tough, if not losing, battle for PeopleSoft, Oracle executives said Thursday that they were considering other acquisitions… Henley (CFO, Chairman) said Oracle had been ‘looking at a variety of areas’ but didn’t identify any potential takeover targets. It has expressed interest in BEA as well as reserve over the price.” Of course BEA was trading at about $8.00 a share in mid-2004 versus $18.05 at the moment.
Time will tell if this deal shall come to pass. It is not for me to judge the will of Carl Icahn, who has reportedly been pushing for a sale for quite some time now. This simply shows that Oracle is continuing to seek growth through acquisition and that BEA has been undervalued, which should came as no revelation to anyone.
In my previous posting on How Many Portals Should a Vendor Offer I talked about how both Oracle and BEA have dual portal strategies. Since then I had a conversation with Shane Pearson of BEA on this topic. I still stand by my posting, but Shane did point out that BEA is dedicated to maintaining 2 SKUs. This means they are dedicated to 2 purchasable products to meet different needs. They are moving forward over time to unify the infrastructure underneath them and new add-ons will likely support both.
To clarify my posting, I don’t think BEA has done anything wrong. There’s no magic wand that can be waved when an acquisition occurs to instantly rationalize all the personnel, products, and technologies and integrate everything. My point is twofold:
1. A roadmap should be forthcoming from a vendor within a short time after an acquisition (about 3 to 4 months) to reassure existing and imminent customers that their current path will continue or let them know frankly that a product or technology will no longer be supported so they can plan for migration or another purchase. No one wants to be the last one to buy a product before it is discontinued.
2. It is my opinion that, as a buyer’s advocate, software purchasers should be aware that as much as a vendor tries to make redundant services (caused by dual path strategies) under the covers transpearant, cracks start to form over time. These cracks are caused by fragmentation and take the form of higher costs (more specifically, costs that don’t decrease over time as quickly as market pressures encourage), higher risk that a piece of redundant infrastructure that a user is dependant on will be eliminated, slightly lower support costs (especially as one set of infrastructure becomes more rare in the marketplace), and slightly more difficulty finding consulting and integration services. I don’t know for sure that BEA will encounter this issue, but it is the norm and provides a reason for customers to be pessimistic in the long run.
I ran a breakfast roundtable here at the SharedInsights Portal, Collaboration, and Content conference this morning and found it to be quite enjoyable and enlightening (despite the hotel dragging away the coffee and tea service too quickly…). The table had about a dozen attendees, mostly architects, and was a good cross section of portal implementations. A large food franchise, a few large government agencies, a major retailer, a vendor (not a portal vendor), a real estate firm, and an international utility were represented.
The primary issue they were all faced turned out to be integrating service oriented architecture (SOA) concepts into their portal environments. Point goes to Plumtree a few years ago who dedicated themselves to the “enterprise service bus” concept. Portals were originally created as “Swiss army knives” that could connect and adapt to all sorts of identity management, security, content management, and application products. It seems that need is still prevalent. Unfortunately, the vendors have been slipping into the mode of integrating the portals into their infrastructure stacks and playing favorites by connecting to their own infrastructure first and then allowing a standards based connector to everything else (blaming it on the other vendors if they can’t take advantage of JSR 170, JSR 168, LDAP, etc.).
It was interesting that while the momentum is certainly in favor of Microsoft and IBM right now, none of the people at the table reported using them as their main portal. Instead it was BEA, Vignette, and a few Oracle. One was interested in open source as well. The table felt that the reason is that legacy environments are not going away, particularly for content management and portal. While many vendors can show a nice, unified stack now, that doesn’t help the practical reality that organizations face with the large amount of built up legacy infrastructure. Governance is a key success factor then in getting each part of the organization to agree to corporate standards even if it is a little less useful for them. Optimizing enterprise-wide sometimes means a sub-optimal environment for those that heavily use another app that offers a portal.
Workflow was also important to the attendees. A few had workflow/BPM tools they wanted to hook up to the portal such as TIBCO and Ultimus, while others were interested in the more simple capabilities you get out of a portal itself.
I’ve often been asked how many portals a corporation should have (see my posting on “Portal Factories” for my answer to that one). But today the issue of how many portals a vendor should offer came into clear focus. Oracle and BEA both have two and, at least publicly, still insist that’s an OK answer. I disagree, as a conversation with a client today put into focus for me.
So what justification is given?
Let’s start with Oracle. The official word in the Oracle press release was
The WebCenter Framework will support portal and content integration standards including JSR 168, JSR 170 and WSRP 2.0 and will inter-operate with standards-based portals including Oracle Portal.
More recently, Peter Moskovits at Oracle said in InfoWorld
Peter Moskovits, product manager for WebCenter, replied that Oracle Portal and WebCenter Suite would coexist at least “for the foreseeable future.”
Now on to BEA. They made similar statements when they acquired Plumtree
Because the two portals target fundamentally different aspects of the enterprise, BEA officials said the company plans to keep the Plumtree and BEA portal product lines separate.
WebLogic Portal targets companies using transactional portals in a J2EE application development environment, according to Mark Carges, BEA’s CTO and executive vice president., Plumtree’s portal is designed for business users in a collaborative workgroup setting,
“We are keeping both [BEA and Plumtree products] as two separate portal product lines for as long as we can see,” he said. “Each product is a leader in its own right.”
Last year, BEA CEO Alfred Chuang reiterated their commitment to the parallel path, with an honest nod to the challenges
“We could integrate them but some companies are buying both, and nobody’s bitching. Yes, it would be much cleaner if we only had one product to support but at the moment I can sell some customers both – I’m trying to sell everyone who has one the other, and it’s going well.”
The justification in both cases is that one product is more developer focused and the other more user-focused (collaboration, intranet). There may indeed be different segments, but then the correct engineering response is to create a unified infrastructure platform that serves up JSR 168 portlets, WSRP, has identity management hooks, administration screens, delegated administration, and personalization capabilities and then offer different layers and packages on top of that to meet market needs. I suspect at some point BEA and Oracle will have to do that by establishing the core of one of their products to be the underlying infrastructure and keeping the other product as a different skin or layer that is productized on top of it.
Parallel paths just don’t work. Unlike consumer packaged goods, where Proctor & Gamble can gain marketshare by offering Tide, Gain, Cheer, etc. , buyers of complex technological products like portals expect their vendor to dedicate significant resources to keeping it up to date and supporting it. This is difficult with a parallel path. Granted, neither BEA nor Oracle have shown strain at supporting both yet (although Chaung’s statement above hints at that), but I believe it will start eating away at expenses over time. IBM is just extricating itself from the confusion caused in the Notes/Domino world over its parallel path strategy and has gotten kudos all around for offering a unified vision.
Sure, integration is difficult but it’s worth an effort. Vignette did not try to maintain parallel paths when buying Epicentric back in 2002 and went with Epicentric. All is not going swimmingly for Vignette at the moment, but competition in the market and not selection of a single path is the cause of that.