Content may be king but Context will be emperor

” and its truely the case with the changing nature of how we access information. Typically, there are three ways in which a user arraive at a site:

• Through direct URL
• Search engine
• Through reference or link from other Site

And context is different for all the scenario’s, especially with the tabletization of the world. It has made every pixel of screen count to present information which the user is looking for. Lets look at the above three scenarios:

Direct URL - Users expectation is that site needs to present personalised content based on their previous behaviour and/or user clicks to present most relevant information. People are always saying: “Present me what I am looking for else I will go and find it else where.” This has driven lot of companies to adopt through anonymous user journeys and present a level of personalised content.

Search engine - This is becoming critically important, as search engines is the main traffic generator for a normal site. Search engine context is the the most important one but has not been explored to the fully. Assume that someone is searching a phrase ‘Fatwire Oracle’ and the site get listed in top ten search results. Now from the query, its not obvious what the user is looking for and so search engine will present a link to the site. When the user clicks on the sits URL, it goes the the specific page ln the site. While the information might be relevant, it is quite possible that it is not relevant for him. But if the user is presented with all the other relevant links around Fatwire Oracle, it is possible that user gets deeper into the site and becomes as regular visitor.

Through reference or link from other Site - This pretty much falls in the same bucket as search engine and the above information holds good.

To add another dimension to the context, mobilization and tabletisation present more challenges to maintain websites and present context based information. So, if the user comes to the site from search, the information presented on a mobile phone needs to vary with what is presented on the website, though they need to be context aware.

With the internet world moving towards tabletisation and mobilisation, its time to re-think before starting any project. How it is going to fit on all the possible context the site falls.

CQ Tips and Tricks

I hope it will be useful for Adobe CQ world !

CSDT has been designed keeping in my the new development as the priority with not much thought around the projects upgrading to CS 761. But in my view the focus should have been other way around as there will be lot more upgrades than new implementations as the initial use case for CSDT.

If you are planning to use CSDT across developers setup, here are some pointers to be aware:
• CSDT V1 has a limitation that it expects the work space name to be cs_workspace at the location /export/envision. So, one option to make csdt work is to create a project with the same name at expected location.
• CSDT expects the same folder structure within workspace, src/_metadata and src/jsp. So it is important that if you are trying to manage CSDT export within CVS, setup the same project structure
• When you have all the artefacts you want to manage within CVS, make sure everyone sync their local instances with the CVS export so that everyone are at the same code level before starting any development
• Word ‘Exception’ causes CSDT import to fail. Make sure that during development, do not specify name/ description fields with it. Or use Excaption

If you are upgrading to CS761, there are number of issues which you need to be aware of:
• Fw_uids need to be same across all the upgraded environments otherwise csdt deployments will fail
• As part of best practice, if you have extracted element as cselement to make code go through the publishing process, make sure that all such dependencies are changed manually in Meta information. (.main.XML) file else the import will fail

General CSDT issues:
• Workflows can not be exported and imported through CSDT so there is need to either mirror to server the workflows or use catalog mover to move them
• Categories can not be exported using CSDT
• If Treetabs or start menus or any other artefact is enabled for any site, it is not exported out when we export artefacts for specific site, though they are part of the at site
• modified date command line parameter doesn’t work as expected
• Not applicable for no asset tables. A classic example is usage of SystemLocateString.
• Incremental deployments are tough to achieve and its more manual work to arrive at the dependency set. Full deployment is the easiest way to achieve code drop.
• CSDT doesn’t check if the asset has actually modified and updates all the assets which are part of workspace. So, if you carry out full deployment everytime, one drawback is that it will invalidate whole of the cache.
• Etc..etc.

Even though there are number of issues with the current version of CSDT, it is a good to start using it as it will provide a platform which is going to improve over time. And hopefully version 2 of CSDT is not far which will definitely address some of the above issues, it not all.

Happy CSDTing !

Good

• Follow,to the dot, the separation of content assets from presentation templates which has been in question recently with the more newer Cms’s like Adobe Day CQ
• Publishing mechanisms available out of the box are one of the strengths with recent addition of Real Time Publishing
• Enhanced caching and seperating the cache to each individual nodes has been one of the great improvements to its performance questions
• Availability of system tools in a supported version is always a nice thing to have
• Finally that there is some respect towards developers by coming up with initial version of Content Server Developer Tools (CSDT), though there are number of shortcoming with it
• Introduction of Central Authentication Service (CAS) is a welcome change to the stack which provides much more flexibility around development and extending it beyond stereotype implementations
• etc. etc.

Not so good

• Some of the basic changes within the implementation requires quite a lot of work. A good example is vanity URLs. Most of the clients now a days have this basic requirement especially around blobs but there is none out of the box URL assembler available
• Deployment is one of the most painful area. Even with CSDT, which is just the recent addition to product, there are many area’s which it lacks. The basic use cases are not covered by CSDT
• With CS 761, there are headless installs which is welcome change through if you are upgrading to CS 761 from version below CS 75 there is no other way apart from GUI based installers
• Product customisation is one of the pain area’s. There is no concept of extending the existing elements into your project elements and apply changes. Which means each time you run an upgrade, there is an activity to reapply all the changes again which makes upgrades sooo much error prone and time consuming
• CS Engage, for providing personalised content, has never been an easy product to work with and get your head around it
• There are number of modules on top of core content server but its been utilised scarcely across the globe and tough to find a single case study when in need
• Rollbacks are still is old 90′s fashion way of taking file system and database dump and restoring them back. Cold dumps needs to be taken to get an exact snapshot of the system at that point in time
• Lots of documentation is there but they cover basic information and for anything beyond basic changes require folow-up on yahoo groups or through support
• Support centre has not been supportive. It sometimes feel, just to meet their SLA’s, there are responses which doesn’t even relate back to the actual query
• etc. etc.

Most of the not so good issues has already been addressed by the newer CMS products and what is taken for granted with a product,Fatwire has not been able to address them. It will be good to see how Oracle take them on board and how WebCentre integration brings in to overcome some of the obvious issues.

Usually upgrades shouldn’t be too tricky as most of the upgrade does is change some schema definition and update existing product code base. So why it needs to be considered at CMS selection process? Read on !

Most of the big WCM Implementations need customisation of some sort and so upgrade projects become more tricky when product overwrite these customisations. A lot of it depends on the under lying architecture of the product. Lets consider two well known products: Oracle WebCentre (previously Fatwire) and Adobe Day CQ. Adobe Day CQ provides facility to overwrite the default components and configurations by simply extending them within your own project. Customisations are achieved without modifying any of the system files, provided its implemented based on CQ recommendations. So, upgrade path for CQ is pretty straight forward and shouldn’t cause sleepless nights.

On the other hand, consider Oracle webcentre where customisation could be achieved only by modifying the underlying system elements. Even your implementation follows the Fatwire recommended approach for customisations, it will lead to changes to system elements. What that means is that upgrade is not straight forward and upgrade itself becomes a project in its own where time lines vary with the number of such customisations.

In a short period product A might satisfy your requirements more than product B, but its important to look at the longer picture because something which looks cool and easy to implement and run might not get you the expected returns in future. And considering the way technology is changing, no one can move away from not upgrading their products for long.

financial institutes

WCM products fit in most of the institutions but its always tough to use any product with financial institutes. And these are the main reasons I think:

• There is always extra layer of security applied within financial institutes which not always fits well with most of the products security models
• Each financial institute has rigid auditing requirements of capturing each of the activity that happens within WCM. And to my knowledge, I do not think any WCM really supports required granual level for auditing
• As part of audit requirement, there is always need to generate how the site looked on a certain past date and if we add personalisation angle, it will make all the more
  complex.

Based on my experience, deploying any WCM product within a financial institutute leads to lot of customisations within the core product functionality. And that is always scary because it raise lot of questions around the credibility of the product and its selection process. Also, managability and upgrade of such system is a nightmare.

Even though it is time and money to develop an in-house CMS system, but I think in long run it pays off and the system could be easily designed to meet the above requirements.

• Customer perspective on Oracle acquiring FatWire from Apoorv
• Oracle Buys FatWire, Now What? from Pie

Here are my thoughts based on various Fatwire modules, though it comes a little late in the day :

Content Server: Content Server is the heart of Fatwire product suite. Content Server provides content creation, management and delivery platform, providing basic and flex asset models to design content types. Basic, in a simple term is a row in database table while Flex is managed across various tables, thus providing complexity as well as flexibility. I think the core CS will remain though we need to look out which development and improvement model Oracle adopts: basic (rigid) or flexible. One area to watch out for cs is its integration with Oracle’s Enterprise search engine.

Engage: provides personalization and merchandising features, enabling organizations to gather information about site visitors and customers, evaluate it and provide personalized promotions, offerings. Could Fatwire’s Engage will engage Oracle not to use Oracle CRM on demand? My guess is that we will see more of rules and promotional offerings being evaluated through Oracle CRM on demand with Engage becoming more of gathering engine.

Community Server: Community server is the social offering from Fatwire which includes setting up Blogs, commenting and rating system. This is the biggest area of threat as Oracle already have social collaboration platform in the form of WebCentre. And as I understand, Oracle always wanted to push WebCentre and likely scenario could be tight integration with Webcentre which will become the social collaboration platform for Oracle’s WEM.

Gadget Server: Fatwire’s Gadget server helps organizations to expand their capabilities for delivering a personalized, interactive, web 2.0-style online experience to engage customers, partners and prospects. Not sure what’s going to happen with it. But I think Oracle will try to do everything around web 2.0 through WebCentre.

Content Integration Platform: Offers powerful capabilities for accessing content of all types from EMC Documentum, Microsoft SharePoint, and Windows and Unix file systems, for fast and efficient publishing online. EMC bit will be cut off with Oracle plugging in UCM to manage documents and we can see more of integration modules evolving in the long run.

Satellite Server: Remote caching engine which brings site near to users. Do not see much change happening with it as it complements core content server.

Content Optimizer: Is the measuring tool which helps business users and marketers to measure the effectiveness of online content and campaigns. FatWire Content Optimizer works with FatWire Content Server and FatWire Engage to enable real-time tracking and optimization. It will be interesting to see if Oracle’s Content Analysis tool : Oracle Hyperion Web analytics gradually gets pulgged in Fatwire.

Digital Asset Management: Digital Asset Management (DAM) Solution is provided by EMC Documentum to manage DAM assets. With Oracle, we can see DAM solution integration with UCM and Fatwire-EMC solution and will gradually die off.

CSDT enables developers to work in a distributed environment using tools such as the Eclipse Integrated Development Environment (IDE) and version control system (VCS) integration. CSDT does not interfere or integrate with other development models. Using CSDT, a development team can manage Content Server resources and exchange those resources with other members of the team

One of the important factor which comes along with CSDT is file system representtion of CS repository. What CSDT does is convert the native CS asset representation to .mail.xml files and vice-versa.
This enable developers to work either in CS or in eclipse. Once the resources are in filesystem, its much easy from management and deployment point of view.

Positives

• Easy to setup and get started
• Finally a development tool to make developers life easy with more controlled release and code management
• Develop JSP elements with standard Eclipse features such as tag completion, syntax highlighting, debugging, and so on
• Command line utility to export, import ,list both CS and workspace artefacts
• Manage most of the CS work from within eclipse, including log level changes, access to log files etc.

After playing a bit with the development tool, here are my findings around CSDT OOTB usage:

• While exporting/importing data from CS using command line utility, there is no way to specify absolute path for datastore. It is always relative to export/envision within CS directory
• CSDT import requires workspace to be present on the server before running the script within export/envision. There is no way to specify remote workspace/zip file, though you can start the process remotely
• Incremental builds, for example, only changed files from CVS, can’t be achieved using import command line utility as the utility looks for all the dependency files to be present in workspace even though they have not changed
• “modifiedSince” option on command line only works for Assets and not for other Fatwire artefacts
• Logging available is limited
• Listing commands (listds, lisrcs) doesn’t check for dependency
• There is no option to see if the existing artefacts within a workspace meets all the dependecy checks before importing in an environment, though when you run import it log errors of not locating the dependent files
• There is no support to export/import workflow artefacts from CS and vice-versa
• There is no option available to export users association to sites

So, typical steps for moving a new site from one CS instance to other includes:

1. Export/check-out from CVS @SITE, @ROLE, @ASSET_TYPE, @TREETAB, @STARTMENU, @ELEMENTCATALOG, @SITECATALOG, @ALL_ASSETS for a site
2. Import all the exported artefacts in a specific order with @SITE first and maintaining all dependencies
3. Manually setup workflow in the new environment (assuming there is no connectivity between the instances)
4. Manually configure users to site association

And Bingo, new CS instance is up and running with the latest managed code.

Over the period of time, the above approach has lead to many CMS implementation failures or the thought of using CMS at the first place. Configuration and deployment process is the least which is talked about during product demo’s and presentation but always is an area of concern. One, there is no way to manage your configurable items in a Concurrent Version System. Two, deployments are hand-crafted than any automated tools. Three, there are lot of manual processes required to be in place to manage change control and many such reasons. And if you have a big team, anything more than five, it is really a nightmare. Resolving conflicts, managing communication around changes, keeping track of the changes etc. adds just new dimension to complexity.

Many CMS vendors have realized the problem area and the result is inclusion of development tools/ IDE’s with the product. Adobe’s Day CQ comes with CRXDE which is custom-built,pre-packaged, stand-alone Eclipse application specifically for CQ and CRX and thus enables to efficiently develop project. Fatwire’s 7.6 version now comes with CSDT (Content Server Developer Developer Tools) which enable developers to work in a distributed environment using tools such as the Eclipse Integrated Development Environment (IDE) and version control system (VCS) integration. Both of these developer tools expose the repository structure in file system which enables all the doors for configuration management.

Once the files are available on a file system, it is pretty straight forward to integrate with a CVS system for config management and providing a much cleaner development environment for big teams. It also enables for continuous integration plus automated delpoyments. Config management becomes managable and the complexity suddenly dis-appears.

I think going forward, we will see more development tools evolving for products, enabling customers to focus of real CMS issues than management and release processes with these tools.

Strengths

• Easy to setup and get started
• Architecture (Open source + standards)
• Sling framework is pretty cool ! .html to .xml to .json etc
• Jackrabbit, Felix OSGi bundles JCR 283 Standards compliant
• Default cluster of one, so easy to add remove new instances
• Syndication/ content sync is pretty fast
• Packaging architecture is neat
• System backups
• Blogs (reverse publishing)
• etc

And there are lot of sites which talk about its positivity. There are always two sides of the coin, and product does have some not soo good aspects as well. Find below the list which might be useful for users who are trying to evaluate CQ. The below list is based on my experience and I am happy if any of the CQ guru’s want to correct me:

Weakness

• Separation of component and content: The component model doesn’t really highlight the real value of a CMS product. Most of the components which comes out of the box doesn’t really separate content from presentation and I fould it hard until we adoped the approach of data components and presentation components. For more details, please refer to my old post: Adobe’s Day CQ : Separation of content from components
• Multi Vendor and issue with CSS /Divs: CQ requires HTML, CSS to be defined in specific formats, as it adds a number of divs on its own to establish its paragraph system. If you are working in multi-vendor environment, where a creative agency is developing designs and mock-ups without any clue of CQ, it may sometimes be really hard to adopt it in CQ. It is better to look for some guidelines from CQ around the structure of HTML, CSS to be defined.
• Thin Documentation: Even though dev.day.com has basic set of documentation available, it is really hard if you want to customize even a small bit of it. The architecture is pretty open but unless there are pointers to explore its really a big hinderance. The google groups CQ forum has more of queries than solutions, but recently it has been pretty busy.
• Authoring WCM User Interface: UI for authors is Ajax based, and so it is an issue in organisations where JS is disabled. With JS disabled, the interface comes back as blank page.
• CRXDE: Popular tool for development is CRXDE but we had troubles in its integration with SVN. The updates didn’t work as expected and you can’t check-out a single file. There is no proper guidelines to merge and resolve conflicts apart from doing it outside CRXDE
• Reporting: Out of the box reporting available within CQ 5.3 is pretty thin and you need to write custom code to extract most of the required information. This has changed significantly with V 5.4.
• Package Manager: New package manager with CQ 5.3 has number of issues and in my experience it is better to use the old package manager which is more reliable. Also, the UI for pacakge manager doesn’t refesh and hangs for ages.
• Multiple interfaces: CQ comes with number of interfaces including WCM, CRX, Apache Felix, servlet engine administration. Some of the functionalities re overlapping across interfaces. It will be better to consolidate into one.
• CQ Corruption: We have seen a situation where CQ stopped working just because someone switched off the machine while CQ was running. It happened to be that CRX repository was out of sync on local and shared directories which didn’t allow CQ to start again.
• Sidekick is sleak but annoying and does page refresh for each action

If anyone wants to share their not so good experience with CQ please drop a comment or mail me and I will consolidate as part of the above post.