IBM BPM 8.5 first impressions

Enter IBM BPM V8.5

As many people may know IBM has released its next major version of the IBM BPM, version 8.5. So far I’ve managed to spend some time evaluating the new product to identify its strengths and weaknesses and determine whether this new release brings the benefits promised by IBM during the lead up to this major release. My first impression of the product were mixed.   There are some important improvements but also so major decisions have been made that could be taken negatively by the existing customer base.  Lets go through them.

The good

Reporting is back!

Now, for those that will read my blog on gaps in IBM BPM, you will note that virtually the entire reporting framework in BPM was depreciated in version 8.0 (depreciated means “no longer strategically viable and likely to be removed in the next release”).  Those of you that have seen open source BPM systems would understand how major this decision was.   A primary value proposition of IBM’s product was functionality such as reporting which differentiated it from its open source counterparts.    There are plenty of solid BPMN based engines on the market so the point of difference was in the polish and additional capabilities that a business could leverage to pay off a significantly higher investment than open source options. With reporting depreciated this left a massive gap in products future, raising the question, why buy something without a reporting strategy? I’m pleased to see in v8.5 the first signs of a replacement.  I will say “first signs” because what I see is only really the reinstatement of the old framework using a new base technology with some bells and whistles.

Googletrends Dojo vs JQuery over the last few years
Googletrends Dojo vs JQuery over the last few years

IBM has made the tough decision to go with the generally unpopular JavaScript framework known as Dojo to implement its new reporting framework. I’d guess this decision was made due to the adoption of JQuery (a very strong alternative framework) by Microsoft, but only IBM would know for certain the reasoning here.  Anyway due to this decision IBM need to spent time and effort building this capability up as the broader development community has moved away from Dojo over the years. (please see google trends graph).   It is however great to see the reporting framework back as at least there is some capability to work with again.

Cache management

Finally, after a great deal of time IBM has admitted that its security group caching doesn’t work and provided a way to manually reset the cache using a exposed API. Wonderful! That being said there hasn’t been a resolution to the actual problems we’ve seen with caching beyond providing a reset mechanism.  There are also additional problems that have surfaced with management of group and LDAP caching that are now cluster related and these aren’t yet addressed directly in either the base release or current fix pack.   Those using Enterprise directory services such as Active Directory will still have to be very careful here.

Evolutionary improvements to coaches

Although I known this was expected there are minor improvements to the dojo coach framework including some changes to dynamic visibility.  Given the initial bugs and issues with dojo coaches improvements are welcome and expected.  That being said I don’t see enough yet that would bring dojo coaches up to parity with heritage coaches which despite their well known limitations are fast to build and can be developed by users with limited experience rather than requiring fairly complex coding skills.

The bad

Hardware demands up and more bloat

This is the first time I’ve install the product and discovered that I simply could not throw enough memory at it to get it running.  The memory footprint of this new version is increased from version 8.0.1.  I found it needed 4-6GB of ram to run acceptably while virtualised.   I’ve not yet investigated if this is just poor defaults configured on the “typical” environment or another example of the often seen unnecessary bloating of software over time by many vendors. No notable performance improvements in general interaction with the tooling  however I’m yet to perform load testing.

Deployment Architecture

The marketing hype around the new deployment architecture and how it simplifies IBM BPM doesn’t translate to much in the way of visible improvements.  Although installation is simplified this doesn’t make it easier to run unfortunately or install plus I encountered issues during the installation that forced me to re-install it about 5 times before I was comfortable with the installation being stable.  We all know that dot zero releases from IBM are not usually production grade but this was significantly less matured than I had expected.   It appears IBM has even taken out the 4 cluster gold topology in the new platform which I am sure will stir up a number of infrastructure architects who invested time in reviewing and approving those typologies which now are no longer recommended.   From a development viewpoint the loss of the single server option for standard and advanced editions is also a negative, meaning all environments use extra (unnecessary) resources to run and take longer to configure and run.  Depending on how the tool is used this can translate to a higher cost to some projects.

Further loss of browser support

It puzzles me why its not understood that moving browser versions is a major (costly, complex and time consuming) effort for many companies, particularly government departments or those businesses with lots of legacy applications.  In Australia IE 7.0 is still common, there are commercial tools tied to this version and web based tools absolutely must be built to support a broad range of browsers to cater for corporate environments.   Version 8.5 of IBPM reduces support again, now IE 8.0 is known not to work for certain functions at all. Tables in particular are noted as not working in IE 8.0, which basically means IE 8.0 support is gone as tables are often needed in coach design.  As can be seen below selecting a dashboard in IE 8.0 intentionally fails to render.  IBM call this “graceful degradation” I call this no support for IE 8.0.

IE8 Support Removed for Dashboard
IE8 Support Removed for Dashboard

Now, even if support had not been “gracefully degraded”, recent performance work I’ve been involved in has shown IE 8.0 is no longer viable for use with IBPM 8.0.1.x, as any use of the expansion capabilities of the portal reduce its performance to levels that make the browser unusable with portal.  (60 plus seconds to render the inbox are clicking show more about 4-5 times and expanding task details)

Still not supporting Australian locale

No improvement that I can see for support of non-US countries (unless its not clearly documented).  So those of us in Australia still have to deal with confusion over time/date formatting, which can create costly mistakes on a day to day use of the tool.


Non Australian Dates
Portal still not supporting localisation

General Impressions

Beyond these positives and negatives there isn’t much change and I get the strong feeling this is an interim release and not the main game.

A few things stood out worth noting:

  1. Participant Groups are gone, replaced with Teams.  It appears that Teams are just participant groups with all the capability that was previous broken removed. (i.e. dynamic expressions are gone in teams which never worked anyway) The change was documented here.
  2. Team’s can have managers defined in a Process Application.  I’m yet to decide if this is a good or bad change, because real world usage patterns I’ve seen do not tend to fit well with IBM’s concept of hard-coding organisational hierarchy in with the process definition (think dynamic reporting, overflow and re-organisation of work independent to code snapshots).  There is however a change to include the ability to invoke services for teams however whether this has adequate performance or flexibility is the key question.
  3. Process Admin has had some minor improvements, such as more controls over the problematic user sync tool.
  4. Portal layout is changed again.  Good or bad I cannot be certain, but change for the sake of change rather than direct evidence such as UX testing is not something I generally agree with.

Beyond again I really don’t see a huge amount to talk about.  There is changes to document connector capabilities and improvements in web services but I still don’t see this being at what I’d consider Enterprise grade yet. Blueworks Live also still lives in the cloud and therefore not meet the requirements of many companies on data privacy, nor do I note any major changes that would solve the single source of truth issue that is caused by using to isolated tools for management of process data.

Overall, reinstatement of reporting was critical but did not justify a new release.  The changes to deployment architecture are likely why this is a major version and frankly I am not convinced the changes will benefit that many customers, this is yet to be seen. Based on what I’ve seen I believe most of the V8.0 and V8.0.1 customers will probably stick to the applying fix packs for the minute and avoid the new version upgrade for a while longer, particularly given that the V8.0.1.1 fix pack is out and stable.  (V8.0.1.2 is also out but I’m not yet convinced this update is a stable upgrade yet.)

What we’d like to see in the next release

Well given this looks like an interim release, perhaps its worth talking through a few things that are missing.  In the next release I’d like to see the following top issues dealt with:

  1. General validation and robustness built into the environment, avoidance of corruption in the BPMN diagrams, deployments and a way to validate the environment is 100% operational. F.Y.I Simple infinite loops in BPDs can take down servers and this is a basic check that could be included in a validation function.
  2. Bring the new coach framework up to parity with the old heritage framework
  3. A solution to problems that involve flow of work with a twist of case management.  Auto-flow as it stands is not a solution.
  4. No more caching, clustering, or enterprise integration problems.  We are well past the Lombardi / IBM merger now so its time for IBM BPM to become a tier one application. This means open up the options on integration to web services, rules engines, security providers etc so they are easy to work with without having to have special case conditions such as needing server restarts every time new data comes through the security provider for example.
  5. Customisable dashboards, without large investment in development time or training
  6. A clear direction on Enterprise Reporting (how to do it with existing Enterprise solutions)
  7. Replacement of the Event Management framework, the UCA approach is complex, combersome and backed by enough bugs to render it useless for any complex processing.   This needs a replacement.
  8. A more flexible portal with consistent UX (including saved searches)
  9. Provide strong migration options to customers