Late last year Apriso launched a new manufacturing process intelligence product we refer to as MPI (software vendors hate to spell out long words). The launch and subsequent market interest in this product has been interesting. From a personal perspective, being the product manager responsible for managing and driving acceptance of this product has been a great experience, one that I have learned a lot during the journey.
Now we are six months into the launch, I thought it might be helpful to document some of my findings so far. As time goes by, who knows, maybe I might even publish a full length paper providing further details and explanations. For now, here is a highly condensed list of just a few top benefits that seem to be universally agreed upon, which is driving considerable purchase discussion, as a complement to a manufacturing execution system.
The five top benefits of MPI are as follows:
- Out of the box analysis – along the lines that simpler is better, we spent the time to ensure industry standard KPI’s are instantly available, so this helps to justify the time to break even; no costly Business Intelligence (BI) project is required to unlock functionality
- Fast integration – as MPI is built on the same unified data model and process framework of FlexNet, the implementation time is pretty quick; data is fed directly from FlexNet, so time to value is considerably condensed, giving users immediate benefits
- The power of intelligence standardization – the value of knowing you are measuring performance across sites based on the same measurements standards and metrics is very powerful; knowing that these measurements are all based consistently on best practices is even better!
- Essential visibility for new product introductions – everything about launching a new product is difficult, from design to engineering to production; any product that can help to simplify and accelerate this process is seen as a win
- Less reliance on IT – we all know that everyone is very busy today, trying to do more with less; IT departments are no different; often it is tough to get visibility into operations intelligence and reporting; MPI removes this constraint by giving the operators total control on what to measure, when and how
Hopefully the above list helps to clarify why better visibility into your manufacturing process intelligence really is a good thing. In fact, after having seen and learned what I have experienced over the past six month, I have to ask myself “why didn’t we launch this sooner?” 
Do you have a story to tell? It would be great to hear from you!
Loading tweets...
6 comments
Skip to comment form ↓
contract manufacturing
July 2, 2012 at 7:21 am (UTC -7) Link to this comment
The Manufacturing process intelligence plan seems to be a profitable way to get the job done. Being able to follow the MPI can ensure the manufacturing is complete in a timely, cost effective , and quality way.
James Mok
May 4, 2012 at 9:57 pm (UTC -7) Link to this comment
Interesting discussions. Any comments regarding the use of Hadoop on in-memory BI to enable efficient search and pattern recognition on unstructured data? It seems that this technology can solved many of the point s mentioned. SAS for example is going for this.
Delcuvellerie
April 30, 2012 at 12:44 pm (UTC -7) Link to this comment
New players in BI are claiming DWH data modeling is “thing of the past” thanks to in-memory associative and column tabular techniques. To my understanding, MPI still uses traditional cubes with dimensions model. What do you think ?
James Montgomery
April 30, 2012 at 9:11 pm (UTC -7) Link to this comment
Great question. In-memory analysis has many benefits, and we seriously considered this. However, in the end we decided against it for a number of reasons. To start, this is NOT the way to go for off-line analysis. The problem is that these systems are bound by your source system’s data, often leaving source systems bloated with unneeded execution data thereby reducing the speed of your execution systems. Moving the data out into a data warehouse allows production systems to be kept Lean, thus increasing speed where you need it most.
Additionally:
1. In-memory analysis still needs to be structured
2. Most of the time the data is cached anyway to reduce the impact again on the source system(s)
3. In-memory cubes are not that well adopted by third party technologies, unlike the Microsoft stack for example
* Reports vs. cockpits vs. dashboards vs. analytics – no one solution can really do it all (despite what everyone claims)
* We wanted ONE source of the information for all reporting and analysis. If you don’t enforce this concept, you then cause all sorts of issues at customer sites, such as inconsistent KPIs, measures and information, when accessed in different ways, ultimately causing inconsistent results
4. Centralized intelligence (one central server fed from many plants) does not work using in-memory analysis. It is dependent on the source systems being the same structure, a very solid connection from the source systems and very advanced data management
5. Translations and automatic Time Zone conversions are often very tough to do when relying upon in-memory analysis
However, the most important reasoning is around historical integrity and accuracy around slowly changing dimensions of (SCD). This is very hard to do without a data warehouse. A good example is a machine that was at workstation1 for three months is in workstation2 for the last month. To do this, the IT staff updated the master record of the machine to point to workstation2 a month ago. The problem is that all this historical information for workstation1 is now pointing to workstation2 as this is where the master machine record is pointing. What is needed is TWO machine records pointing to the correct historical records.
Conclusion: As before, there are some very good advantages to in-memory, and we definitely considered it. However, the cost to the source systems and the lack of central support changed our mind. Additionally, these in-memory solutions can still take advantage of our solution by referencing the data warehouse and not the execution (or archiving) system. The MPI Data Warehouse is kept very clean and is in a highly optimized reporting structure (called star-schema), so an in-memory system can really take advantage of the heavy lifting that has already done by Apriso’s MPI.
Custom supplement manufacturing
March 25, 2012 at 9:11 pm (UTC -7) Link to this comment
So nice blog site……………
most sellable and effective skin analyzer by doctors
March 22, 2012 at 6:46 pm (UTC -7) Link to this comment
thanks for the information….^_^