Edge Analytical Processing

There’s been a lot of work going on in the BI space for the last few years. Some of it interesting, some not so much. But one thing I find very intriguing is what I call Edge Analytics. Think of Edge Computing applied to business and governmental analytical processing.

EA Processing has many facets and I want to cover just a couple of them in this blog post. So exactly what is EA Processing and what advantages does it bring to my organization? EA Processing, in a broad sense allows a company to process most of their data locally. When the demand is high, meaning lots of jobs consuming lots of CPU cycles and I/O bandwidth, the company has the option of firing up a server on the cloud to extend processing capacity. This might only last a few hours or be something that occurs at month end or year-end processing.

Using EA Processing negates having a secondary server (or sets of servers) on premises that doesn’t take up data center space, electricity and maintenance. EA Processing is useful for allowing consultants and contractors to have access to compute power in a highly controlled environment and that can theoretically be more secure than on premise analytic servers.

When viewed broadly, EA Processing can also serve as a center for disaster recovery. With the hurricanes that hit Houston, Texas and ravaged most of the State of Florida, it has become apparent to many that having EA Processing capability is an important feature to consider when building out your BI stack.

Finally, consider this example of EA Processing that is not truly cloud based but does use remote connectivity to do basically the same thing. Say your company has offices in Munich Germany as well as Los Angeles, California. Your organization has data scientist and data analyst at both sites. The time difference between Munich and Los Angeles is eight (8) hours.

The BI staff in Munich can have access to all the computing power on the analytics servers in Los Angeles since most of the staff isn’t even up or at work when the Munich staff are at the office. The Los Angeles BI Staff (for the most part) also have access to the Munich servers since the Munich staff are already out the door heading home while they are work.

Both locations have access to multiple servers to fill in the need when high process demand from one site or another is required. Both locations have control of their local data, work under an identical security model and each have low latency when running local jobs.

With WPS, you can easily configure remote processing with the workbench. You can use WPS Link to run your jobs on any WPS Server. You can also use WPS Communicate to run jobs or parts of jobs on any WPS Server.

When you have multiple WPS Servers, it’s invaluable to keep them busy. Most companies in my experience have servers sitting idle for a good part of the day. Analysts (and I’m speaking from experience here) get really frustrated and testy having to wait for jobs to execute simply because resources are not properly set up to server the users in the most expedient and cost-conscious way.

About the author: Phil Rack is President of MineQuest Business Analytics, LLC located in beautiful Tucson, Arizona. Phil has been a SAS language developer for more than 25 years. MineQuest provides WPS consulting and contract programming services and is an authorized reseller of WPS in North America.