Is Central Practical, or really Distributed Solution Leveraging Micro Data centers, provides a really the Practical Way to go Industry Solutions

In the last couple of weeks I have come across a number of very large opportunities (10 Million I/O) that challenge the traditional thinking of potentially “central” systems.

The key is these are distributed operational systems, in supervisory control, information systems, and Operational control (MES) systems. With distributed vale generating assets (plants) of different sites, but with the many sites orchestrated together into a working align value chain. But today's production requires flexibility and agility requiring “actionable decisions” to be taken at all levels of the Vale chain.

At the “edge” e.g. the field plants decisions must be able to made, and that means the information, and ability act is local, with a timely response. So while Central data centers with master data storage and applications is logical, it is only practical to have distributed systems.

Last week I was speaking at conference and one of my fellow speakers talk about “Micro Data Centers” as the next wave in industrial computing following the banks etc. This appealed to me as faced these large opportunities where they want 99.99% uptime, and responsive systems. While we have developing the software to addressed distributed “peer to Peer” systems, and solutions, the key is the associated hardware. 

So what is a Micro Data Center???A micro-datacenter is a smaller containerized datacenter system designed to solve different problems or to handle different workloads. A micro-datacenter (MDC) is a smaller, containerized (modular) datacenter system that is designed to solve different sets of problems or to take on different types of workload that cannot be handled by traditional facilities or even large modular datacenters.

Whereas an average container-based datacenter hosts dozens of servers and thousands of virtual machines (VMs) within a 40ft shipping container, a micro-datacenter includes fewer than 10 servers and less than 100 VMs in a single 19in box. Just like containerized datacenters, MDCs come with in-built security systems, cooling systems and flood and fire protection.

This blog is worth a read:http://blog.schneider-electric.com/tag/micro-data-centers/

With some interesting comments:“Their size, versatility and plug-and-play features make them ideal for use in remote locations, for temporary deployments or even for use by businesses temporarily in locations that are in high-risk zones for floods or earthquakes. They could even serve as a mini-datacenter for storage and compute capacity on an oil tanker.We have already discussed that Micro data centers are breakthrough solutions, but we are still in the early adoption phase and the market potential is untold. Where are they already in use? How fast will the market ramp up?Actually, you may already be using micro data centers and don’t even know it. The demands of real-time (or near real-time) data processing needs in environments with factory automation (robots), industrial automation (cranes), and bidding or trading stocks and bonds, for example, call for the capabilities micro data centers provide.The sheer amount of data required in such industries like oil and gas drilling and exploration, construction and mining also require processing to be on site and so they do not go through latency increasing hubs.Other sites may not have as much big data, but micro data centers offer advantages through standardization fast deployment, ease of management and troubleshooting and security, not to mention cost effectiveness.But the use case on the horizon with the greatest potential is a massive distributed network of micro data enters to form a content distribution network. This processing on the edge will support the commercial Internet of Things (IoT), including the fast emerging category of wearable devices.   The processing of data could be reduced to mili-seconds here.
Forbes recently reported on the massive size of the IoT — expecting worldwide market solutions to reach a value of $7.1 trillion by 2020 and connected devices to double by then to 40+ billion. While micro data centers might be a niche today, they will become more ubiquitous as they will be needed to facilitate this unprecedented connectivity.”

So when looking at these distributed Industrial Architectures these “small Bunker” micro Centers have the real opportunity to provide a distributed hardware architecture with practical capability to support a distributed Industrial Application of tiered historians and application servers.

Data Structuring/ Normalization and Validation the Hidden Key to Unified Information Solution!!!

For the last 2 weeks I have been on the road, sorry just did not get to updating blog, but a constant discussion topic that keeps coming up. “How to truly build a Unified Information system that allows a view across their industrial/ production assets, and the ability to compare, and understand current state, capacity and performance quickly and easily.”

People have tried the intelligence, information semantic products which layer on top and bring valuable representation of information, but while the semantic information is a critical role, and aligns data silos, it is only as good as the data.

When discussing strategies with companies in Upstream oil and gas, Mining, Metals, Refining, Power, Water Utilities, and a common theme is the need to get data structuring/ normalization and key is validation of data.

One person commented last week “we cannot just take all the data from our existing SCADA systems/ historians and bring it to central historian due to no alignment, lack of consistent context, and data validation!”

The problem is not just about aligning data for access, the goal today is to make decisions not just about one process or plant but timely decisions across multiple sites. To achieve this it is not just about gathering existing data, as one set of engineers put it, we need to normalized data this means units and measures been aligned for one. They are putting unifying storage systems but they spend a lot of time going to the source of actual data, and then structuring and normalizing measures. Today this is being done in the control layer and supervisory layer. It is also key that the existing running systems on a plant are not disrupted, so the risk to changing these running systems in these systems is way too high. 

In one case a customer talked about 16 versions of one measure across 14 plants. This same story comes all the time, so the questions is how to achieve a result that does not threaten the existing sites, especially when the challenge today is not in a plant but across sites. These sites have been put in at different times with different systems, and often different objectives.

The key is to look at a layer that allows:

·         cross site administration/ configuration e.g. a distributed system

·         ability establish standards for asset structures / kpis / measures/ units, with the ability to centrally administer these standards and enforce governance

·         ability data collect from both real time and historical data stores

·         ability to set up in these structures validation rules that will validate the captured data to trusted before storage

·         distributed data capture and ability to store locally and aggregate providing data integrity.

The key is a 3 layer solution:

• A data collection platform that can be phyically be distributed but centrally managed with standards governance, providing data structure, intellignece, and validation as close to the source as possible. So that different process/ plant data both realtime and historiacl can be aligned across this common namespace.
• A storage layer that the platform now propagates this now normalized information into one or multiple distributed storage historians, providing a consistent information storage across these different control/ automation systems.
• A unifying information layer with a semantic capability to allow this unified storage + the operational storages from different systems to be aligned. 

These three layers enable success as a team, not any one of them, if you are looking for unified information system that can be trusted, so decisions in the now can be made.