Migrating Plumtree 5.0x to MOSS2007

(Work in progress… Tell me what else you think I need to put in)
Background

Recently, the organisation I work for decided to move away from a long standing relationship with Plumtree Portal and embrace MOSS 2007.
The initial process we undertook was to do a gap analysis of the 2 systems.
We identified the main areas of concern and held workshops for Available Web Parts, Branding & Navigation, Knowledge Directory, Analytics, and Collaboration Server.

The next stages were 2 fold:

1. The business were sent off on a task to archive, delete and generally reduce the total size of the migration.  Initially we were looking at something like 500 communities and 2000 projects.
2. Technical were to produce metrics for the most commonly used Portlets, migrating site structure and security, and Collaboration Server projects, documents and version history which in turn could be extrapolated.

POC and Initial Estimate

The initial process we undertook was to do some ballpark worst case(everything done manually) estimates.  To achieve this we looked at:

• Project Structure
• Project Security
• Average Number of Documents per Project
• Average Number of Versions per Document
• Community Structure
• Community Security
• Community Supporting Article Pages
• Most highly used Portlets

We then ran a small pilot to see how much work it would be to migrate each section.  Once you have the basic you can then extrapolate this based on the expected number of instances per community and the total number of communities.

The numbers we used were:

Item	Time Taken per Item(hrs)
Community/Page creation(includes adding web parts)	1
Community Security(includes resolving nested Plumtree groups)	1
Project Security(includes resolving nested Plumtree groups)	1
Project Documents	0.5/version(Includes download and Re-upload)
Portlets	1 (Depends on what is used)

In our case we create a Team Site for each project which takes about 5mins so wasn’t considered a significant part of the estimation.

After we extrapolated these numbers using an 8 hour day and a 250 working day year we got to a number…… a big number, I believe the projected was somewhere near 22 man years.

If your still reading this you must really want to do this…. or someone isn’t listening to you telling them to start from new.

Worth knowing before you get started

Before moving forward, it’s probably worth mentioning a few restriction Sharepoint has:

1. You can only update the Last Modified/Created date and person by using the API and submitting the document that way.
2. When you crawl documents into Sharepoint you replicate the file systems ACL, you do not get the option to override this at crawler level as you can in Plumtree(this was a huge issue for us as the ACL in many cases was junk)
3. No Sharepoint URL can be greater than 260 characters
4. Sharepoint has no concept of Global groups except for AD
5. You cannot nest Sharepoint groups like you can in Plumtree
6. All Sub Site groups are held within the Site Collection
7. Out of the box Sharepoint blocks certain file types, a full list is available here and is configurable via Administration Site-> Operations -> Blocked File types.  The ones which caused us issues to date are MSI, EXE and CHM.

Knowing some of those points would have saved a lot of issues when designing the system.

Getting Started

When you first get started, you will need to plan to have several tracks running simultaneously within the project.  Within our project we had 3 track leads for Content, Implementation and Architecture.

• Content, covered Branding, which Portlets were to be migrated, what the mapping between Web Parts and Portlets was, and the applicable estimates for any custom work
• Implementation, covered the roles and responsibilities, what the hierarchy of the new system would look like, security management etc.
• Architecture, covered what hardware we needed to provide a robust solution to serve out 10,000 user base

Content

Portlet to Web Part Mapping

The mapping exercise is probably the most critical section from an end user point of view(only because I ran this piece).

We set out with a set of Portlets and asked the business to describe what they used each Portlet for what functionality was absolutely needed and what was optional, basically a business requirement document.  We then passed the work off to an outsource team who set about documenting the functionality as they saw it, eg. what buttons are there and what do they do.  The offshore team, all of whom had Sharepoint knowledge then recommended a Sharepoint solution based on the findings with a %’age match, functionality which was missing, and any additional functionality.  The business was then walked through the solutions and asked which if any recommendation they would take.  If none of the out of the box solutions were suitable the solution was passed to Technical for estimation.  This area is where many battle will be fought.  We started with 34 Portlets and needed to custom deliver 18 Web Parts.  Following on from estimation and various follow-up meetings trying to understand the business requirements we managed to work our way down to 7 custom, 4 were needed immediately and 3 nice to have.

This is where you must do the work.  Getting the business to accept the out of the box solutions will save you many man hours better used writing migration tools or branding.

Time wise we planned for 1/2 a day of time to generate business requirements, 2 days to document actual functionality and proposed solution, and then 1/2 – 1 day per document to present the solution and work through the proposals.

Branding

When it comes to branding this can either be very simple or not.  Recently, my company spent a small fortune getting Plumtree consultants to implement a Design Agencies nightmare and it fell to your truly to not only migrate the design, but also make it faster.

Sharepoint Branding is split into several sections, some will allow you to control colours and widths, others allow you to change the HTML document structure.  The areas are Site Definition, Master Page,Page Layout, Core.css, ~custom~.css, Themes:

• Site Definitions allow you to specify different Master Pages, Page Layouts, Lists, Navigation Items, Default Web Parts and Custom CSS files and Features etc. all of which are applied when the Site/Site Collection is created
• Master pages are selectable at a site level and let you completely change the layout of your page.  Master pages are only applied to Site areas and not to areas such as lists screens.  Do not delete placeholder tags from the master page, use the visible=false attribute and place them all in an asp:panel with the visible=false attribute set, this will ensure they do not render.  Sharepoint expect these placeholders to be there.
• Page Layouts are more important when using Publishing but basically allow you to control the HTML that appears in various sections declared in the master page
• Core.css Changes to this file will impact all Sharepoint sites on the system.  I never change any Sharepoint files as they may get overwritten when the server is patched
• ~custom~.css is a safe way to apply your own CSS file.  The file can be called whatever you choose and you can generate different ones for different scenarios.  When writing custom css remember to only override what you need.  Most people take a copy of core.css, rename it, change the bits they need and delete the rest, otherwise you end up downloading the same file twice
• Themes are a collection of images and css which are applied by the user selecting the appropriate option from a list

Time wise this activity is solely dependent on the requirements.  I spent 7 days modifying the Team Site master page,  creating custom site definitions, modifying the CSS and writing a custom feature which would enable various parts of the site.

Implementation

Structure

When building the Sharepoint hierarchy, keep in mind that Site Collections are a good way of grouping similar sites together and apply the same look and feels quite straightforwardly using the inherit parent master page and inherit parent style sheet options.  The fact Sharepoint stores all sub site management groups in the parent Site Collection will also make things easier if you have lots of separate Site Collections as you will a more appropriate list.  The disadvantages of separate site collections are:

• There will be breadcrumb issues unless you use the Portal Connection, as this component always starts at the root Site Collection
• You cannot share Sharepoint groups between site collections, in this case you will need AD or maintain the user base in 2 locations
• Most web parts don’t allow retrieving content between site collections, in particular the content by query one
• I’m sure there’s more

 

Plumtree Components

Analytics

Sharepoint has usage tracking, but this is much more like an external sites requirement(Referrers, Top Sites etc).  We found that there is a huge gap between what analytics can do and what Sharepoint can.  Sharepoint has by day or by month tracking.  You will not get more than 31 days as the data is deleted from the database by 1 of the Jobs

Collaboration

Due to the size and nature of Collaboration server items, we decided to generate a separate Projects area and setup each project as a standalone Team Site Collection.  This allowed the users to store large quantities of data, provide granular access to security and generate sub sites for any sub projects/work streams.  To accelerate the migration we developed a custom tool which read the Plumtree database, created the Site Collection, mapped security  flattening existing Plumtree groups, exported each file version from the database and in turn uploaded it to the Sharepoint Shared Documents List resetting the Last Modified/Created by and date fields.  Click here for Creating Sites or here for Uploading Files and altering meta information related article

Knowledge Directory

The Knowledge Directory was removed and any documents which were required were removed from the shared drives and uploaded to Sharepoint

Search

Sharepoint out of the box doesn’t support wild card searching.  A free alternative is available from Ontolica which overrides the default search control and enable wildcard searching from the search box.  It does not however allow wildcard searching from the people picker.
After enabling the Ontolica search we are no longer tracking search usage report data.
I have spent several days looking into this and found the cause not to be with Ontolica but with the way Sharepoint builds it’s usage data. 
Basically Sharepoint Usage is based on the Search Text Box.  If you don’t have the Sharepoint  search box web part on the results page then the usage data isn’t tracked. 
I’ve posted this all over the place and I can’t get anyone to tell me I’m wrong… or that I’m right.  I’d love to be told I’m wrong and it’s config, but the usage data is only tracked on our system when the Out of the Box Search Box   is on the results page.

Structure and Content

In the end all the content and Community structure was migrated by hand.  This was a much more accurate/efficient model and in reality we are pushing though several communities a day with 4 resources working on it

User Profile

User profiles within Sharepoint are totally inconsistent.  If you click a user link from the search results page, you’ll probably end up on their my profile page, if you select a user from a user list, contact web part or any of the Last Modified, Create By links then you will be taken to the Sites User Info screen.  Clicking edit on this screen for the actual user does nothing in most scenarios and in most instances is very confusing.  You can work around this by setting the Portal Connection to the My Sites root URL.  This however restricts you from ever having any connected sites and you will always have the Portal Connection text in the breadcrumb trail.
We have worked around this by overriding the Delegate Control on the UserDisp.aspx page.  For more information I have posted an article on this.  But essentially it redirects anyone who hits the UserDisp.aspx page from a specified host to the My Profile screen.  Click here for more information

Conclusion

There are a few companies available, who describe experience in this area.  In most instances we found that they were cost prohibitive and brought little more to the party than we had ourselves.  If you check my posts I have given several examples of how we found ways around common problems which you can re-use and am happy to answer any questions on.

 

Any questions? Anything not clear? Drop me a comment and let me know