Tech & VC 17 Jan 2007 09:12 pm
Approaching the Limit of Zero
Everyday we are seeing the price slashing effect of Moore’s Law on bandwidth and storage costs. What happens to web services when the cost of bandwidth and storage approaches the limit of zero? I have a few thoughts, but I’m sure you all have many more valid opinions, or would love to call me out on mine. Some that I like:
- Cloud storage. Access your files anywhere. Geographic redundant backup is standard because incremental cost of storage is negligable, so reliability against data loss is 5-nines (%99.999) at minimum. Hard drive failure fades into history like the slide rule.
- Revver no longer offers added value by hosting your video for you. They still offer value by matching content owners with advertisers, but suddenly the importance of distribution displaces the importance of reliable and free hosting. So, instead of Revver doing a rev share with just the content owner, now Revver has to cut the syndication channels a piece of the ad revenue in addition to the content owner.
- Defensibility in infrastructure disappears. Google no longer is defensible in its massive server farm implementations. In fact, Google will be at a disadvantage because margins in hosting will be too thin too justify an in-house solution for an easily outsourced problem. Think I’m on crack on this one? Ask an AOL sysadmin about their dialup farms from the early 90s; they are currently, at best, a waste of space and a liability.
- Data ownership becomes an abstract notion of permissions, not physical ownership of material containing data like it is today. When all my data is in the cloud, there’s no point to physical media like CDRs because physical media is slower and less reliable. Transferring data to another person is the act of granting permissions to that person to access a small corner of the cloud.
5 Responses to “Approaching the Limit of Zero”
on 17 Jan 2007 at 9:35 pm 1.Erik Schwartz said …
Storage is getting cheaper WAY faster then bandwidth. Streaming will become irrelevant, you will soon be able to afford to store local copies of most media ever created.
In a steady state world infinite local storage makes perfect sense. But we’re not in a steady state world, thus limited bandwidth still plays a factor. I’m not sure what I think yet, but the equation includes:
Speed of propogation of new media to remote servers.
Shape of the value vs time graph of the media.
Amount of bandwidth required to get the media to remote storage.
For example, no one should ever stream Casablanca or the Godfather, no one should ever stream anything from the Beatles catalog. They both have essentially infinite shelf life and fairly steady value over time.
But if you have a blob of media (say from American Idol), and it takes 3 days to propogate to infinite storage iPods on the edge of the network, and the shelf life of said media is a week until the next show, then streaming makes sense.
The ideal situation would be a hybrid. The network knows what I like and that alters the delivery order to match my usage patterns. So if my niece is an American Idol fan, AI shows up on her iPod right away, her value v. time graph peaks early, very high and trails off quickly. But me, it can show up 3 weeks later and it won’t really matter, my value v time graph is essentially flat (and low).
on 18 Jan 2007 at 7:56 am 2.Andrew Parker said …
I agree that bandwidth is moving slower than storage, but I suspect bandwidth will be less and less of a limitation going forward.
In the early 90s you could get a 1 gig HD for the same price as a 500gig HD today. (rough estimate)
Similarly, the average house in the early 90s was greatful for a 28.8kbs connection for ~$40 on Compuserv or AOL. Now broadband 1.5mb pipes are ubiquitous at around the ~$40 price point. (again, rough estimate)
In a similar time span, we saw a 500x increase in storage/price ratio and ~52x increase in bandwidth/price ratio. You’re absolutely right that cost of storage is approaching zero much faster than the cost of bandwidth, but both are heading to zero very quickly.
on 18 Jan 2007 at 10:47 am 3.Erik Schwartz said …
I remember it was a big deal when the 1 Gig drive dropped below $1000 (I was still at ICTV so it was 93 or 94), that was $1/megabyte. I can now buy a terabyte for $450, that’s .00045/megabyte. That’s a greater than 2000% decrease in storage price.
Bandwidth (via cable modem) is a contention based system. If your entire neighborhood is using bandwidth at the same time you’ll get no where near 1.5 mb/sec. Taking the next steps in raising bandwidth require a rebuilding if the hybrid fiber/coax cable plants so that each node passes fewer houses. The cable node loop has about 30 mb/sec of bandwidth, a fiber node passes anywhere between 200 and 2000 homes. Rebuilding cable plants to have smaller nodes will be very expensive. Getting order of magnitude increases in bandwidth will require fiber to the home.
on 18 Jan 2007 at 12:44 pm 4.Micki Krimmel said …
Micki here from Revver – Just a bit of clarification: We already share revenue with syndicators. This affiliate system has been in place from the very beginning. That’s the idea behind Revver – to encourage and reward people for sharing media across the open web. This syndication network and our relationships with advertisers is what differentiates us from the other video services which only provide free hosting.
I like your notion of abstract data ownership. I imagine it will take quite a leap of faith though for people to move completely away from physical data storage. There’s a lot of comfort in knowing my media is backed up on a drive at my house.
on 18 Jan 2007 at 12:56 pm 5.Andrew Parker said …
Thanks Micki,
I was just daydreaming because I didn’t have much interesting to post about. Hope you didn’t take it personally in anyway. The whole post is half-baked. I was unaware of the syndication rev share, but it makes sense now that I think about it. Clearly I’m not that familiar with Revver, though I am certainly a fan (otherwise I would have chosen a competitor to talk about).
Thanks for stopping by my little corner of the blogosphere.