Andy Knaebelis information technology services director/interim CIO for Riverwood International Corp., Atlanta, Ga

Building a solid IT foundation

Build a house’s foundation with-out the right materials and it will fail, bringing the house down with it. Build your information systems without the right infrastructure and your systems will fail, bringing your business down with them! System failures in the paper industry can be very costly to the business and even potentially dangerous to mill and plant employees.

When Riverwood International implemented the SAP enterprise resource planning system in early 1999, the implementation team took care to address the critical issue of building a solid infrastructure or foundation for SAP and all the other critical systems that would be connected to it. Let’s take a closer look at the elements of a house’s foundation to help draw the parallel between a solid foundation and IT stability.

For the house, two elements pull together and form one system: reinforcing bar (rebar) and concrete. The rebar always goes in first. The concrete follows and is made of a carefully balanced mix of aggregate, water, and cement. Once the concrete has dried, the imbedded rebar helps prevent the system from failing under stress. Building a strong systems infrastructure also requires rebar (quality IT processes run by the exceptional people) and correctly mixed concrete made up of aggregate (hardware), water (software), and cement (networking). Now that we understand the elements, let’s take a closer look at achieving IT stability:

1. LAY THE REBAR. Engineer just the right amount of IT people and processes. With too many people and too many processes, you won’t get anything done. Even worse, your foundation will crack under stress. Stay nimble, focus on a few critical processes, and identify the best people to run them. At Riverwood, critical processes include change management, problem management, automated operations, and IT systems performance management.

2. MIX AND POUR THE CONCRETE. Make concrete using two parts hardware, one part software, and three parts network:

Two parts hardware. Always think in pairs when considering your hardware purchases. One of anything spells OUTAGE, while two spells AVAILABILITY. Don’t kid yourself; the best repair contract you can obtain from the most reliable vendor will never save you. And, as for that box full of spare parts that’s in the back of your mind, drop it. In a crunch, the people who can install them will just happen to be on vacation and your systems will be dead. Get it over with, bite the bullet now, and invest in an identical, fully redundant set of processing hardware. Be sure to include your facilities in this assessment. It does no good to have two processors plugged into the same outlet. Get your team to install extra, redundant electrical feeds off of different uninterrupted power supplies (UPSs) and generators.

One part software. Don’t mix and match. Instead, stick with one vendor and one proven product. Save the cute stuff for the application layer, since playing games with your operating system (OS) is a recipe for certain disaster. Unfortunately, even under the best circumstances, many available products in this “information age” still aren’t ready for prime time, so you must do research and spend time hands-on testing the solution as thoroughly as you can. You have to be absolutely sure that the combination of hardware, OS, and fail-over software will work with and support your application. The goal is to have the system recognize that a potentially catastrophic event has occurred and automatically fail-over to the redundant hardware you invested in. This must happen without human intervention or, even better yet, without anyone even knowing.

Three parts network. Local redundancy, logical redundancy, and circuit redundancy are also crucial. Before Riverwood ships a roll off its mill docks, a transaction must go from our manufacturing execution systems in the mills, up to SAP in Atlanta and then return to the mills to print shipping documentation and other instructions. Network outages simply are not acceptable. If you have similar requirements, consider this. Local redundancy means multiple physical wires (or fibers) from the systems in the mill to the routers to the telephone company. Logical redundancy means alternate logical paths through switches and routers. Circuit redundancy is automatic fail-over to backup integrated services digital network (ISDN) or other circuits if your primary network fails.

Look closely at your network. Take the time to search for single points of failure and eliminate them. Your network is the cement that bonds your systems together. With no cement, there is no foundation.