For anyone who lives or has worked at all in The Valley, there is a local tourist attraction called The Winchester Mystery House. The house is a 7-story Queen Anne Style Victorian mansion that was commissioned by Sarah Winchester in the late 1800s and early 1900s.
The home is a local legend not because it serves as an architectural marvel (it is more of an architectural eyesore), but because of the stories that surround its creation and purpose. Sarah Winchester was the wife of gun magnate William Wirt Winchester. You might recognize the name because of the rifle that William popularized and bears his last name. The Winchester lever action rifle was the dominant firearm in the days of the Wild West, and it became colloquially known as “the gun that won the west.”
While there are several versions of the story, one of them suggests that upon William’s passing, Sarah went to see a medium (think spiritualist) in Boston. The medium told Sarah that she should build a house and that if construction ever ceased on the house, the spirits of those whose lives were lost at the hands of her husband’s invention would kill her. That began a period of 38-straight years with non-stop, round-the-clock construction on what is now known as the Winchester Mystery House.
The home today is more oddity than anything else. It features staircases to nowhere, windows inside the home that look onto other parts of the house, inverted columns, doorways that open into shafts, and any number of random features that were added over the years. Built without any plans or architectural drawings, the house is really just a collection of add-ons designed to keep the workers building.
And in many regards, the home and its creators are the perfect allegory for the networking industry.
For the past three decades, we have been an industry that has incessantly built our own mystery house without really consulting an architect or drafting up plans. We have traded in our saws and hammers for C and Java, and we have built knobs and protocols in place of staircases and walls. But the results aren’t that different.
Essentially, we have become incrementalists. And we have lost sight of the purpose of what we do. Networking gear is not the mere collection of everything we have learned up until this point in time; its purpose is not to present a complete anthology of all protocols and features ever invented. But increasingly, networking devices have become just that.
The path of the incrementalist always begins where it last left off, and the new path is additive. In product terms, we start with the same basic design (same chassis, same software), and we add some things to it. Maybe we design a new ASIC so we can add a little bit more headroom. Perhaps we add a little bit of LISP here or PIM bidir there. Every addition makes full sense in the context of the problem, but when cobbled together, the solution represents more Mystery House than dream home.
And, by the way, it’s not just vendors. Our collective inability to walk away from things (either old features or old applications running on the network) has led us down this path of lowest common denominators where the best we can hope for is interoperability with everything. The challenge is that the more extra baggage we carry forward, the more encumbered we are and the slower we all move.
The signs are everywhere. Look at the gear that vendors provide. Look at your own network. Are they designed with a purpose? Or are we building our own equivalent of the Winchester Mystery House? Is this really what we aspire to?
Unfortunately, we cannot meaningfully improve as an industry until we free ourselves of our incrementalism.