Silo-Busting vs. Silo-Bridging

Blogger: Lyn Robison

Eliminating silos has proven to be an intractable problem. Application silos exist in every enterprise. Business functions are typically optimized using an IT system of some sort, and each IT system tends to become its own silo. Our two primary approaches to busting silos – SOA, and Enterprise Software Applications –are proving to be ineffective at busting silos.

SOA is an effort to eliminate process silos, but this effort has stalled, largely because of the difficulties inherent in process integration. Process integration requires distributed transactions across service calls, which are difficult to implement. Another problem with SOA is that one of its promised benefits – the ability to dynamically combine business processes – is not really very useful. In actual business practice, how often do enterprises really need to dynamically combine their business processes? Aren’t business processes usually pretty static and well-defined in big businesses today? Business processes which change a lot or which need to be dynamically combined with other processes are the exception. Most business processes need simple data integration more than they need over-engineered, SOA-based, dynamic process integration.

Enterprise-wide software applications are an effort to bust silos by subsuming small silos into big ones, but in actual practice, this approach complicates instead of solves the problem of silos. For example, enterprise-wide CRM systems are an effort to take one important enterprise data type – customers – and put customer data inside of a well-oiled silo. Okay, how about other important enterprise data types, such as vendors? Vendor information tends to be scattered across modern enterprises too. Are we going to build another big silo for vendor data? How about financial data? And how about employees, fixed assets, inventory, products, accounting, sales, billing, logistics, contracts, or supply chain data? Are we going to build big silos for each of these types of data too? Enterprise data types and the processes that work on them are too varied to be crammed into a small number of big silos – they won’t fit into a few big silos because there are too just many data types in modern enterprises.

Okay instead of busting silos, which SOA and Enterprise Software cannot do for us, what if we bridge silos instead? This new approach, which is well-proven, is to reconcile the data between the silos – make data consistent so that data flows smoothly between silos. The beauty of this approach is that it is based on nothing more than competent data management.

Burton Group will soon publish an overview that describes a case study in which two IT developers successfully bridged nine distinct silos within their enterprise. The overview is entitled, "The XQuery Development Stack: a Toolset for Content-Centric Applications", and will be published in early March. In mid-April, I will do a telebriefing on the topic of bridging silos as well.

Comments

[elliptical reviews]

No offense, but i suggest admin adding a google+ button for easy share!