Sun

Apr 13
2008

Jesse Robbins

Jesse Robbins

You Become what You Disrupt - (part two)

Google's GrandCentral (Radar coverage) was down over the weekend resulting in missed calls and other phone problems for its users.

This is very similar to the the two day Skype outage last year where I said that "You Become what You Disrupt". I've spoken about this issue several times, most recently at the Princeton CITP "Computing in the Cloud" workshop.

The problem is that it's not particularly clear at what point a disruptive innovation becomes a utility. As innovators it's important that we recognize that this point will arrive and prepare for it. I believe that we have a responsibility to be good stewards of the technologies we create, and to take responsibility for protecting people who come to rely on those technologies to live their daily lives. When we fail to do that, we may find ourselves being cast as either fools or villains who must be regulated and controlled.

Ultimately, I think we will evolve a set of safety standards very similar to building codes. For instance, it appears that a multi-datacenter strategy would have prevented the GrandCentral outage. (As I've said many times before: Datacenters are a Single Point of Failure!)

Cofounder Craig Walker writes: "I wanted to write a quick note to all the GC users and apologize for the service interruption this morning. We had a power issue at our current colo facility and it knocked us off line for a few hours. Unfortunately I’ve been up in the mountains with the family this weekend and had no cell/internet coverage so couldn’t respond earlier. I did want to let you know that we were able to restore the service by noon today and are working extremely diligently to make sure this won’t occur in the future. We’ll do a better job keeping you informed in the future, not only about service related issues but also about upcoming features, soliciting your feedback, and generally making sure that you, the GC user, is well informed as to what’s going on with the service."

Will better industry standards, best-practices, and independent certifying authorities emerge for these new utilities without innovation-stifling regulation? I hope so.


 
Previous  |  Next

0 TrackBacks

TrackBack URL for this entry: http://blogs.oreilly.com/cgi-bin/mt/mt-t.cgi/6435

Comments: 10

  Thomas Lord [04.13.08 05:54 PM]

Where are you going to find the experts to develop "building codes"?

Upstream from the horrifying "technologist" and engineering cultures that produced the problems you are describing is the meta-process by which experts come to be recognized or shunned, respected or treated as irrelevant, etc. Who are the "alphas" and why?

Also: how do your ideas here apply to O'Reilly's own "house"?

-t

  David Ulevitch [04.13.08 06:26 PM]

Great post. Sums up some serious issues nicely.

  Thomas Lord [04.13.08 06:33 PM]

David:

We don't train engineers well, these days. And Capital doesn't use them well. And people in general don't understand what engineering is so they form bogus reliances and ask for all the wrong things. And a few people make a heck of a lot of money at the margins of that mess.

Regulations ain't gonna fix it.

-t

  Jesse Robbins [04.13.08 06:45 PM]

Where are you going to find the experts to develop "building codes"? Upstream from the horrifying "technologist" and engineering cultures that produced the problems you are describing is the meta-process by which experts come to be recognized or shunned, respected or treated as irrelevant, etc. Who are the "alphas" and why?

This is the main challenge, and part of why I'm so passionate about projects like the Velocity Conference. The reality is that academia isn't really addressing this problem as far I know (but I'm not an academic, so I don't know very much). Worse, most companies that are good at it regard these "best practices" as a source of competitive advantage.

I'm hoping to make it safe/possible for the existing "alphas" to get some slack from their corporate keepers and start fixing this problem now. It's an uphill battle, but one that is in our collective best interest I think.

Also: how do your ideas here apply to O'Reilly's own "house"?

I don't really know. I'm not an O'Reilly employee and have very limited visibility into the company beyond Radar.

  Thomas Lord [04.13.08 07:12 PM]

Do you not see the conflict that arises immediately when you cite a problem and then use that to advertise a conference that is an example of the dynamic that drove the investment that caused the problem you cited in the first place?

I agree that academia isn't fixing the problem but "more of the same" is surely not the solution.

-t

  USA [04.14.08 12:06 AM]

We are in the dawn of these technologies society has to be understanding if they want to use them in while still in their relative infancy.

Even after many decades and trillions of dollars in infrastructure investments, the traditional utilities have still occasional outages

  Simon Wardley [04.14.08 02:02 AM]

"I think we will evolve a set of safety standards very similar to building codes"

The building industry are well ahead of IT. They've been doing "mashups" of commodity like services such as bricks and water works for centuries, ever since Marcus Vitruvius Pollio (Author of De Architectura) and other early architects pointed out that you didn't need a brand new aquaduct for every villa.

Unfortunately in the world of IT, we're only starting to learn about a service oriented approach to design and discovering that we don't need to build everything for every new project and that we can use pre-existing services and infrastructure.

I completely agree with you, but we've got a lot of catching up to do yet. IT is still the new kid on the block.


"Will better industry standards, best-practices, and independent certifying authorities emerge for these new utilities without innovation-stifling regulation?"

You hit the nail on the head. In my view it's a race between the creation of a competitive utility computing market based upon open source standards (of which GoogleAppEngine's open SDK is one) or alternatively discrete but essential services with no competition that ultimately require government intervention.

  James Urquhart [04.14.08 05:38 AM]

As Simon himself would point out, there is a curve that all technologies go through from innovative to commodity. While we all want to think of compute utilities as commodity driven today, the fact is that we are still in a highly innovative stage.

A stifling regulatory system will fail to gain support both legislatively and practically until the services under review are so well defined that innovation becomes a lower priority than stability. I, for one, believe that this will happen just as it has in building, power, water and other infrastructure: with a great amount of teeth gnashing, but with long term benefit to society.

Its all just one big complex adaptive system, folks, and it will tend towards a complex structure that meets or exceeds the needs of its constituent agents.

  Sean Lally [04.14.08 08:38 AM]

I haven't seen any mention of what anyone pays to have this service in a multi-datacenter HA setup? I thought GC was free? Is it even ad supported?

Sean

Post A Comment:

 (please be patient, comments may take awhile to post)






Type the characters you see in the picture above.

RECOMMENDED FOR YOU

RECENT COMMENTS