Light Peak: black hole or brilliant beacon?
It's a fight folks, between USB 3.0 and Intel's new Light Peak. USB 3.0 is already shipping and Light Peak has yet to be announced, but by late next year they'll be duking it out for the hearts and minds of consumers. And the winner is…
Light Peak Light Peak is Intel's new 10 Gbit per second - scaling to 100 Gb/sec - fiber optic interconnect. Intel is taking direction from Apple and Sony, both strong with creative professionals, and whose bandwidth requirements are “more.”
The buzz is gathering buzz around Light Peak: the Intel demo at IDF; they're working with storage and display vendors; reports that Light Peak is due on the next generation of MacBook Pros this spring; and Steve Job's comment that he doesn't see USB 3.0 taking off.
Taken individually, none of these are conclusive. But taken as a whole they suggest a strategy that is quite a bit more sophisticated than Intel's past I/O marketing failures such as Infiniband.
Light Peak won't be the only 10 Gb per second interconnect shipping: 10 gigE ports will be down to $50 each by this time next year. But USB 3.0 will be the volume leader, thanks to its backwards compatibility with 1 billion+ USB 2.0 ports.
USB 3.0 But if USB 3.0 is already shipping and has a strong installed base of USB 2.0 devices to work with, how can Light Peak possibly succeed? It won't be a slamdunk, but Light Peak has a decent chance.
What are Light Peak's market advantages? In no particular order:
- High-end focus. Apple owns over 70% of the high-end PC market. Their customers can invest in the new technology.
- The everything interconnect. Light Peak implements the lowest two layers of the networking stack. You can run virtually any protocol stack-ethernet, FireWire, Infiniband andUSB-on top of it. Connect displays, storage, networks or virtual I/O systems at twice the speed of USB 3.0.
- Convenience. Optical cables are thin, light and long distance capable. Plastic is cheap for short distances. Glass is highly reliable for long runs. Many USB 2 printers limit cable lengths today, something USB 3.0 won't improve.
- Technical superiority. Not only is it fast, but fiber is immune to electromagnetic interference and crazy-making electrical problems like ground loops.
- A future. USB 3.0's performance is impressive, but it's half of Light Peak's and the end of the line. Light Peak is just beginning its lifecycle with a roadmap to 100 Gb/s.
Creative pros need a smarter I/O strategy today. The welter of interconnects - USB, FireWire, HDMI, digital audio, eSATA and SDI, to name a few - either have technical problems like noise and distance or cost and availability problems.
The Storage Bits take I wasn't optimistic a year ago about Light Peak, but I'm seeing more intelligent marketing than I'd expected. Light Peak won't replace the cheap and ubiquitous USB anytime soon, if ever. But it doesn't need to.
Today's notebooks often sport 3 or 4 USB ports, eSATA, DisplayPort, HDMI, VGA, Ethernet, FireWire and microphone and speaker jacks. It's almost as bad is the late 90s when PCs had PS/2 ports, parallel printer ports, SCSI, modems and VGA ports - and 2 floppies.
Remember SCSI-based scanners? I do, and it's not a happy memory.
Intel, Apple and Sony can easily fumble the Light Peak opportunity: excessive licensing fees (think FireWire); wonky DRM; cost-curve fails; insufficient vendor support.
But it can clearly be a big win for storage users: higher data integrity; greater bandwidth; longer distances; more convenient hardware; and the ability to integrate fast solid state storage wherever it make sense.
Let's hope they get this right.
Comments welcome, of course. Want to see more? Check out Intel's IDF demo. Update: Good discussion between NonZealot and DeusXMachina on how LP might be used and what it is and isn't deep in the comments. Keep clicking "View more comments" until you get to it. End update.