Today’s supply chains are vast and wide-ranging. But that scale also makes them fertile ground for risk: concerns over fraud, contamination, insecure production sites and unknown product sources trouble both customers and clients.
These are all factors that make transparency both more vital and more complex. To see why having a full understanding of the complexity of detailed stock tracking for regulatory and safety purposes in the process manufacturing industry, let’s take a look at one of those most scrutinised industries of all, pharma.
In the intensely regulated life sciences industry, pharmaceutical companies must be able to identify at any given time where any individual medicine item is. It’s deemed that in the event of a safety issue, it is imperative that items can be quickly removed from the market to minimise the risk to consumers and the cost of redress. In the fightback against counterfeit goods, regulators also require that individual medicine products are clearly verifiable as authentic. As of February this year, for example, the EU’s Falsified Medicines Directive specifies that any medicinal products must carry a unique product identifier code and that manufacturers and distributors must demonstrate detailed record-keeping, while all products must be logged in a central database of drugs sold in EU countries.
The technical challenge of meeting these compliance targets can be onerous. With thousands of product lines produced across multiple sites which are sold into hundreds of markets, keeping track of every stock unit exceeds the scope of the standard way businesses have to organise data, specifically relational database systems (think Oracle or Microsoft SQL Server). The numbers of unique serial codes alone can run into billions, and CIOs need a highly scalable way to manage the vast volumes of serial numbers.
Many organisations still store this information in data silos, making it only available in partial views, though. Even if the data is stored in one database, if it is run on SQL-based database technology, a simple and fast navigation through all the data in order to recognise how a production line or particular pallets and their contents are connected will be next to impossible. With increasing connectivity and a move to things like an Internet of Things, this complexity is unlikely to decrease.
The reason for this is that relational databases, which store information (product, pallet, production site, serial number etc.) in rows and columns, are poorly-equipped for identifying relationships within datasets. And these connections are essential for identifying a specific product’s whereabouts or to monitor, analyse and search the supply chain, and to share significant data about production sites and products.
Making traditional databases work in real time is also problematic, with performance suffering as the total dataset size grows. Increasingly, however, a software called graph database technology is a solution, because of its ability to record complex data interdependencies. The idea is that when you track something, you create a hierarchy or ‘tree’ of data: if you scan the code of a particular pallet, it will automatically recall its contents. Graphs offer a tremendous advantage over traditional relational databases, maintaining high performance even with vast volumes of data.
And instead of using relational tables, graph databases use structures better at analysing interconnections in data, and they also adopt a notational formalism closely aligned with the way humans think about information. Once the data model is coded, a graph database is practically impossible to beat at analysing the relationships between a large number of data points.
Such a relationship-centric approach enables the manufacturer to better manage, read and visualise their data, giving them a truly trackable and in-depth picture of all products, suppliers and facilities and the relationships between them. Using a graph database, manufacturers can typically demonstrate 100 times faster query response speeds than that enabled by SQL RDBMS software. That sort of response time is critical when you need to provide second or sub-second responses, when required to identify a specific product’s whereabouts. And critical in order to comply with the latest global regulations of traceability and to manage that time-critical and reputation-critical product recall effectively.
Graph database technology is a great enabler and effective solution for organisations that need to work with complex supply chains and provide the level of highly granular governance and sourcing capability our global economy demands.
By Emil Eifrem, CEO and Co-Founder of Neo4j, the world’s leading graph database company