One of the most challenging activities in operations and maintenance is the accurate forecasting of when to replace critical technology on a project. Effectively predicting what to buy, what to replace, and when to make a purchase can result in massive cost savings and enhanced mission continuity. Doing it poorly can have disastrous consequences in the other direction.

GDIT uses simple but impactful heuristic data analytics to maintain a custodial account of how, for example, end components in a data center are used so that we can proactively replace equipment or parts before they fail and accurately forecast the precise moment at which to do so.

To do it, we look at things like the mean time between failures, or MTBF, of every component. Then considering the on-board sensor data including usage, vibrations and others metrics that are built into today’s systems and sub-systems. We know how the MBTF varies across manufacturers or in certain environments or temperatures, and we know how each piece of equipment is being used – e.g., is it running 24 hours a day? Is it running over-temperature 80% of the time? How do run-times and temperature affect MBTF? We look at metrics like time to live, or TTL, and similarly forecast how long data will live on a network, how much traffic is on a network and how that taxes the network’s infrastructure overall.

With all of this information we can use algorithms against the data to model when whole pieces of equipment or component parts will need to be replaced. We can then overlay that data with supply chain data – e.g., how long will it take to get something, and when can we get to the customer site to replace it – and plan our procurement schedule accordingly.

Just as important, with all of this real-world data and manufacturer data, we can also predict and plan for potential failures and mitigate them before they happen. This adds enhanced mission continuity to a project and also lets us perform resource planning that tells us which team resources will need to be in which customer locations to perform the maintenance. All of this data, because we are proactively tracking and forecasting it, allows us to be a better, more effective and more efficient partner to customers.

Without question, artificial intelligence and machine learning are generating a lot of buzz and attention – and have been for years. But solid analytics, done well and done consistently, can generate significant positive impacts for programs just the same. It enhances up-time, generates certainty of costs and resource planning, saves money and can dramatically impact mission delivery.

GDIT supports contracts that involve much of this at a massive scale and scope, so when considering headcount, cost and physical infrastructure – this type of heuristic analysis is essential. It involves methodical data collection, monitoring, modeling and planning, but it can have tremendous returns.