When i hear from teams mentioning that they have single PI server with 2 TB ram or 1 TB ram or their organization is having a road map of consolidating all PI servers to single PI server with Large RAM or processing power, i always gets worried about future state of Infrastructure and development woes their IT organizations and Vendor organization have to with. Its good to have things consolidated but the question is what extent of consolidation is good for organization.
Certain questions to ask if consolidation to a single entity is good idea or not in Process Information systems are and their importance.
- How many subset of organizations parent organization is divided into and operating in how many geographical areas.
- What is a real goal of centralization?
- What issues are foreseen before centralization and after centralization?
- Is cost a major factor forcing for centralization?
- Are the IT Teams forcing for centralization?
Often organization move forward for centralization in-order to reduce cost without even looking forward for the overhead costs of data and bandwidth when they centralize everything. Organizations have to look at what is important to be centralized and what not.
Ex: Consider a large mining company operating across multiple geography’s looking forward for centralization. Each geography data is only required by that organization and not all is important at central location, central location really looks forward for having data that is important to them. In this case, they don’t require all process data stored in Terabytes of data but they just need final production numbers of the day which will be in Kilo Bytes and this data size which needs to be exchanged every day will save significant amount of cost to organization.
If data discrepancy are creeped into local data it would be better to isolate them into local than making them global and then kill valuable time of all locations sorting out the issues. Isolation is very much important when it comes to handling process data and centralization of all data thus eliminates this isolation and leads to frequent data outages which brings the overall efficiency down.
Consider an organization business applications accessing data from a central remote location then one at local for critical actions, it not only gives them ability to act faster but also reliable due to reduced latency. This data exchange will become overhead for the organization as not all other sub organizations will be interested in looking at other organizations process data.
In my view centralization have to be limited to organizations keeping only aggregated data centralized and not all the process data to empower local organizations and to reduce bandwidth as well a latency loses.