Amazon SimpleDB, Google Knol and more.
This week we chat about hulu.com and viewing long-form videos using flash, VoIP on the iPod Touch, Web services interfaces into databases, the Amazon SimpleDB announcement, KDE4, USB vs. Firewire, “upgrading” to Windows XP, and Google’s unit of knowledge.
In relation to some of the video problems Ben mentioned, I feel compelled to mention Miro.
Miro's built-in BitTorrent client can obviously reduce the bandwidth demands on the server, and hence lower the barrier to entry for video content distribution. Which is a good thing obviously.
I use Miro for quickly and easily downloading favourite video clips from around the web, including YouTube. As it turns out, YouTube provides an RSS feeds for your favourite videos. Just subscribe to this in Miro, and whenever you "favourite" something you automatically mark it for download by Miro on the next poll. Couldn't be easier.
Of course it would be great if it transcoded and synced videos with iPods and Apple TVs, but it's probably up to Apple to provide the open APIs first...
One more thing, I just wanted to correct Ben's comment that Adobe's Flash was available on Linux, MacOS and Windows. A crucial caveat is that on Linux this is restricted to 32-bit architectures only.
Adobe's persistent refusal to support 64-bit OS is a source of continuing frustration, because you're not likely to encounter the limitation until *after* you've installed the OS! And the workarounds are very ugly (eg these instructions for Ubuntu).
It's not as if they haven't had time to fix it. Here is a blog post from Oct 2006 explaining why it is hard - but seriously, over a year's worth of work?
I understand that, particularly with the inclusion of MainConcept's H.264 decoder and other third-party libraries, open sourcing of the Flash plugin is not necessarily possible, even though that would be my preferred option of course.
Thanks for your comments Alastair.
I'm just downloaded Miro, must admit I'd never heard of it before. Looks great!
Totally agree that peer to peer technology is the future of large file distribution.
On 64bit support in Flash player, isn't Apple's Leopard now 64 bit? I guess it somehow makes it transparent to apps or presumably the web browsers are still 32 bit.
Personally, I'm planning to go back to 8 bit. I've ordered an Arduino Diecimila
Flash support on older or non-mainstream Linux distributions is pretty poor. The latest Flash players will crash Firefox on non-ALSA equipped distributions, removing support for the legacy sound architecture was a rather nasty breaking change IMO. They could have at least handled it gracefully.
Call me paranoid, but I don't think I'd trust Amazon with my data. The idea has some merit, and the prices are pretty reasonable, but one wonders how things will change once you are locked into their solution. At the back of my mind I keep recalling Amazon's history of rather anti-social behavior like one-click patents and rather aggressive legal defense of their imagined monopoly on "recommendation technology".
Amazon gets some kudos for using a functional parallelised language, but the actual DB primitives look pretty limited. Document DBs aren't a new idea. "Schema-less" isn't a feature IMO, at least not for any application that is non-trivial. Schemas are there to offer and enforce structure in Relational data, the document model is more like a toy object DB.
I also suspect there will be a lot of coding effort to work around SimpleDB's arbitrary limits, but I'd imagine Amazon will extend some of them (for a price). The cloud convergence delay may present problems, but for applications that are largely read oriented it will likely be a pretty good service. Anything transactional I have my doubts about though. I'd consider investing in strong cryptography if I was putting anything of value in there too, despite what their FAQ says about Amazon not looking at your data.
Knol does seem pretty evil for a company that claims to be driven by just the opposite. I can't imagine anything else but the huge amount of parting clicks to Wikipedia has driven this. It is something I don't think we truly need. Although the content provider-centric model has some business merit, the content-centric model of Wikipedia sits better with me.
That said, Google has always been good at finding ways to commercialise things without being excessively evil, and commercialism does tend to focus human self-interest, frequently providing higher quality results than gift cultures like Wikipedia.
The one thing that bugs me about USB is the 500 mA limit (after negotiation). I can understand it from a safety point of view, 2.5 Watts can't do too much damage (is unlikely to start a fire in a fault condition) but it does limit what you can hang off one port without powered hubs. IBM (et al) did come up with an extension, PoweredUSB, that allowed significant power delivery at several different voltages (e.g. > 100 watts at 24 volts), but it hasn't really taken off.
Strikes me that we should come up with a more consumer friendly ethernet connector and just run power over ethernet for our peripherals.
(Dang I hate those little RJ45 clips!)
Thanks for you thoughtful comments.
I like the RJ series plugs, but that little locking bit is indeed very fragile. I can't count how many I've destroyed accidentally while untangling some cat5 or snaking it through a rack.
There are some boots available that shroud the end of the locking tab and prevent it getting tangled, but most of them are so thick they make it quite difficult to unlock the plug again once it is in something. Especially in high density arrangements like a switch.
There is a non-standard version that has a folded locking tab that can't get tangled. This is probably the best of both worlds, but it is likely quite expensive to produce compared to the standard one.
One very cool thing about all such modern comms cabling is how easy it is to terminate (with the right equipment). I love punch-down and crimp termination with well engineered connectors.
Post a Comment