I have found a lot of issues in Fluid Query and BigInsights Applications to move data in the enterprise version. I have several tickets on these with IBM as they need to be addressed to have a solid implementation. Fortunately none of these are stopping the implementation of BigInsights.

Ambari integration with GPFS monitoring is non-existant. We had network issues which caused GPFS to require recovery. We had no notice from Ambari.

MTR for GPFS took several days with only 30% of space in use. In order for this to be more production ready IBM needs to make this better. Way too long.

• The UI was not interactive: Responses used to be very slow and hang up at times.
• The UI was not really helping to track the real-time jobs and its logs.
• You can bring in a better UI for job management and health checks.
• Developer API documentation needs improvement.

At the time when we tried this product (six months ago), there was lack of technical depth within IBM to support with queries. The service was used from the Bluemix platform and that was also undergoing several changes by the day. We had issues of lockout, service had to be re-configured etc which was more a problem of Bluemix platform stabilizing I think.

We were trying to use the IBM Accelerator for Social Data Analytics that came as part of BigInsights. Initially we tried this with a trial version in our local environment but later for customer POC wanted to move to Bluemix platform. Bluemix at that time offered BigInsights as a service. The interface did not have much of documentation to configure and debug. Later we used a combination of JAQL scripts and BigSheets to do some basic analysis. The Bluemix platform was changing with new services getting added every day and the site also changing. We used to get mails asking us to back up the data else we could lose it as services were re-published.

In essence, the approach IBM took to simplify social media analytics through accelerators was brilliant. Unfortunately the stability of the platform was an issue. Haven't worked on Bluemix recently but presume things will be more stable now. I did see lot more documentation when I searched for this accelerator few days ago.

I have discovered most of the features I need in the version 4.0 of this product, but there is always room for improvements, such as getting better performance on Windows based platform and including compatibility with the new frameworks and technologies related to Hadoop ecosystem.

I'd like to see faster execution time, especially for simple queries that don't touch on many rows and don't involve many operations (Joins, Unions, Groupbys).

Installation process should be improved for IBM ValueAdd components, especially scripts for R/BigR installation. They could add Hue and Ranger to the set of services.

I’d like to see Russian language in supported languages for Watson.