Friday, June 12, 2009

Is anyone buying this? Really...?

I stumbled across this article that, quite honestly, at first pissed me off. After a little reflection, I can only laugh. It reminds me of someone trying too hard to sell those $150, short gold plated digital Monster audio cables - Oxygen free or whatever. (If the engineer inside you doesn't laugh, then cry for the sake of the suckers, read on).

Maybe I read in to far, but I see snakeskin oil vendors grasping for air! The piece is an obvious response to Steve Hechtman's very different article on the same topic (hosted on Control Engineering) - he should be flattered. You see, the big vendors, GE Faunc in this case, but the exact same applies to Wonderware and Rockwell have been long committed to the concept of Historians, a glorified and expensive datalogger that includes, and is only meant to work with, a custom version of Microsoft SQL Server (of all product choices...). The problem is that now much cheaper products from companies like Inductive Automation and Software Toolbox can do a better job using any RDBMS (database) system. Being vendor-neutral, inexpensive add-on packages also do much better for things like trending, reporting, and data analysis. The biggest mistake of the current generation of Historian is that they tried to implement and include everything themselves - like making a giant Swiss Army knife with a spork, usb memory stick, and a wine glass. Now they're caught with their pants down, desperately scrambling to recovery their enourmous sunk costs (my favorite business term).

  • You're data is special and requires "plantwide historian" treatment...their example query “What was today’s hourly unit production average compared to where it was a year ago or two years ago?”. I won't even comment...
  • Your database needs to speak specialized industrial protocols (OPC) - There's separation of function by design and for a reason. Besides - this doesn't even make sense.
  • Faster speeds and higher data compression - no way! The historian is wasting CPU cycles in both directions, which obfuscates your data (can no longer use external applications), to do something better achieved by a RDBMS system that supports it.
  • Robust redundancy for high availability - is this a joke? Maybe Amazon.com should migrate their server farms over to GE-flavor SQL Server.
  • Enhanced data security - another losing battle for the historian. The white paper mentions SQL injection attacks - all platforms in question can use stored procedures, and are all subject to this sort of attack. When it comes to up to date patching, arguably the most common vulnerability, SCADA vendors have the absolute worst track record! IT keeps their servers patched as a matter of practice - they're typically afraid to touch the SCADA machines. Ultimately, the "do everything" approach provides many attack vectors.
I can't blame them for playing their hand. I just wonder - will anyone read this white paper and take it at face value?