Saturday, June 26, 2010

What are silos good for?

@markgould13 tells us why the silo doesn't work

Of course they do work - after a fashion. As Mark points out, they are resilient structures, from which we can infer that they serve some purpose for someone in the organization, or the organization as a whole. It is common (Mark calls it a cliché) to reject information or work silos in most organizational contexts. But what exactly is the alternative?

Another useful observation about silos is that they generally represent some attempt, whether planned or emergent, to decompose an organization according to some notion of specialization and clustering. When people complain about silos this could be because they reject any kind of decomposition, but what is more likely is that they dislike this particular decomposition pattern. However, anti-silo rhetoric is often pretty vague about the difference if any between silo (=bad decomposition) and autonomous loosely coupled functional cluster (=good decomposition), and architects who automatically dismiss all previous attempts to structure an organization as "silos" create much the same impression as plumbers and electricians who automatically criticize the existing plumbing and wiring.

The main issue with any decomposition (whether or not we choose to label the chunks as "silos") is the coordination between the chunks, and I think this is Mark's main point. He quotes someone calls Ed Smith as saying

"Insight is the gap and overlap between silos."

Thinking about the requisite coordination between the chunks is an excellent route for understanding whether the architects should pay attention to the shape of the chunks or to improving the links and feedback loops connecting them. Or both. An architecture that is exclusively focused on cutting the enterprise into perfect chunks (relative to a fixed and abstract model of "the requirements") is probably not going to be much use in the long run.


Chris Bird expanded his Observations on Silos (see comment below) on his own blog.

Excellent defence of silos by Venkat, The Silo Reconsidered

Interesting discussion on Linked-In - What are the advantages of working in silos?

1 comment:

  1. Silos, hmm. At the EAC2010 in London in June 2010 we decided that silos are now called "Cylinders of Excellence". This came out of a byplay between Alec Sharp and his audience, reinforced at my session. There is actually an important point here. That is that most "siloes" have a powerful person sitting on top of them. And these people don't want to see their power bases erode. So we almost always see change being resisted until "regime change".

    One organizationsl structure may be no more valid than any other. The only constant is that we see changes. So from an EA perspective, we have to figure out what essential connections we have to make regardless of the functional structures. The tool we often use for this kind of more static partitioning is the Domain Model. We know that Domain Models are helpful - especially when the organization cannot see its "Cylinders of Excellence" represented in them.