Article: Key Challenges in Portal Adoption

Discussions

News: Article: Key Challenges in Portal Adoption

  1. Article: Key Challenges in Portal Adoption (8 messages)

    In "Key Challenges in Portal Adoption," Nikhil Sharma offers an overview of aspects to watch in adopting portals in various organizations and enterprises, as they offer challenges in technology, governance, requirement analysis, integration, and user experience.
    In a selected few cases, where the knowledge and awareness of Portal technology is present in the organization, the team is often found overwhelmed with the Portal features and try to achieve everything in one shot. This leads to the big-band phenomenon, which is particularly chaotic and fatal for Portal projects. I have seen ambitious transformation programs with Portals at their core, usually leading no where even after having spent considerable efforts on analysis and conception phases. Eventually such initiatives are parked / shelved only to be forgotten or in select few cases, broken into smaller parts and taken up in subsequent IS-budget cycles. There are numerous facets to any portal program, ranging from integration of a number of existing systems/products, to content aggregation and repurposing, to UI aspects to rationalization and harmonization among existing functional applications. Each of these should be attacked one by one and a phased approach should be adopted to successfully roll-out Portal in an organization.
  2. Interesting review My only overall comment, is that while portals provide a foundation to a component based developement, there is still resistance from vendors on providing portal components (portlets or WSRP) that can be integrated into our infrastructure. In the world of outsourced development or obtaining 3rd party applications, one of the major hurdles we face is the integration into our "consolidated desktop" model that portal development enables. Yep, sure, open standards (JSR168, etc...), facilitate and simpify doing this, the problem is getting vendors to implement to these standards. The reply we get from vendors (who are developing/selling the end to end solution) is, " Oh sure, we'd like to do a JSR168 implementation, but really, its not on our radar, we are concentrating on the standard Web based app first, but hey, if you pay us for the work (extra of course), we can customise" So fundamentally we are left (at this point in time) with 1 - choose from a MUCH smaller product base from vendors that provide a portal based solution (eg IBM portlet catalog). Thereby severely limiting potential functionality opportunities/requirements coverage. 2 - pay for customisation, the problem being that we have now introduced a maintenance issue. Now, because the customisation is not CORE to the vendor's product, financial pressures on the vendor may result in a "re-prioritization" of the portlet version. 3 - Use IFrame or Web Clipping integration, and IMHO, very much "compromises" While Gartner reports are useful, in the area I am working, ie financial services and heavily transactional systems, where content management is a part (not a large part), but time to market is critical, there is a distinct lack of options (other than 100% custom development).
  3. Thanks Michael!
    ...The reply we get from vendors (who are developing/selling the end to end solution) is,
    " Oh sure, we'd like to do a JSR168 implementation, but really, its not on our radar, we are concentrating on the standard Web based app first, but hey, if you pay us for the work (extra of course), we can customise"
    ...
    The points you have mentioned are very valid. In fact I have got a similar response from a product vendor about compliance with JSR 168 :) for their product and we had to adopt a compromised approach ! But I belive, with standardization, the place is opening and there are more solutions available as OOTB components. Yet, the customization effort required stands out to be significant. -Nikhil.
  4. Hi Michael, You really pointed out a BIG challenge in portal adoption from the implementation point of view. It is hard for us to integrate the legacy systems into portal scope smoothly. Custom JSR portlet is perfect but expensive. IFrame/web clipping is a quick way for integration but also somehow ugly especially when SSO is not working sometimes between portal and the legacy systems. I think the main reason for this problem is the software vendors is making money from their products, not from the portlets. And they want to create the "sticky" software. Products is always more expensive than portlets. :) So I agree with Nikhil that we should take portal as a platform and create standard application (JSR168 portlets) on top of it. Things will be changed if more and more enterprises adopt the portal which support the open standards. On the other hand, the portal vendors should also provide some toolings or out-of-box portal accelerators for mitigate the pain of customers and lower the development cost of portal integration. Furtunately, some portal vendors already did something. For example, IBM provides toolings such as Lotus Component Designer, WebSphere Portlet Factory etc. to help customers to do quicker development. Also there are some portal accelerators on the shelf for the customers. Generally, keeping following standards and open our architecture to more application vendors could give us more opportunity/power to get the integration work done!
  5. In the world of outsourced development or obtaining 3rd party applications, one of the major hurdles we face is the integration into our "consolidated desktop" model that portal development enables.
    Yep, sure, open standards (JSR168, etc...), facilitate and simpify doing this, the problem is getting vendors to implement to these standards.
    The reason for this might be that the JSR168 "standard" is simply totally underspecified and does not integrate well with standard application development. There is no inter-portlet communication, the request and response objects do not extend the standard objects Servlet Request/Response objects. Using another standard technology, like say JSF, in portlets is largely dependent on the Portal Vendor. There are also bugs in various portal products that break the portlet lifecycle and so on. And it is not much better for WSRP where, again, a lot of features that should be mandatory are still optional.
  6. The reason for this might be that the JSR168 "standard" is simply totally underspecified and does not integrate well with standard application development. There is no inter-portlet communication, the request and response objects do not extend the standard objects Servlet Request/Response objects. Using another standard technology, like say JSF, in portlets is largely dependent on the Portal Vendor.

    There are also bugs in various portal products that break the portlet lifecycle and so on. And it is not much better for WSRP where, again, a lot of features that should be mandatory are still optional.
    JSR 286 is expected to make things better for us. But ya! in the existing specs, there are many missing blocks. Different vendors have provided different approaches and proprietary APIs to implement the communication and means to access request/response/session objects. But these take toll on portability of the Portal App. Similarly there is another spec - JSR 301 under discussion which is about Portlet bridge for JSF.
  7. I would like to add the following to the 'underspecified','underimplemented' list 1. Bridging JSF and portlets( Hoping for JSR 301 ). We can't use 286( eventing model ) and the JSF eventing model without a proper bridge. 2. WSRP. Which vendor implements version 1.0 fully ? How does a WSRP Portlet request interact with a AJAX call to the JSF portlet ? Thanks, Mohan
  8. These slick salesmen sold us their portal stack and we were gullible enough to buy in to the hype. When the salesmen did their demo, they showed us awesome stuff that portlets could do. Unbeknown to us, they showed us predominantly Content Management stuff that were largely static news data. We tried building portlets showing transactional tabulated data that end-users could slice and dice. We found the idea of porlets to be not very performant for transactional/dynamic data with lots of user interaction. We realized that portlets are only good for Content Management, but not highly interactive web applications where dynamic data are involved (e.g. bank accounts data that could be sliced and diced a hundred different ways).
  9. We realized that portlets are only good for Content Management, but not highly interactive web applications where dynamic data are involved (e.g. bank accounts data that could be sliced and diced a hundred different ways).
    Thats where a formal product evaluation exercise is needed. Evaluation typically consists of comparing the candidate-product against the other industry leaders for a set of parameters related to the domain of the product. Subsequently, the choosen product should be subjected to a Reference Implementation or Protoype, by a team consisting of your resources as well as vendor resources. This is a generic enough strategy, which usually is followed, but still many times I have seen organizations falling for the Sales-pitch / demo and going for a product in haste :)! thanks, Nikhil.