Amazon's New Silk Redefines Browser Tech 249
angry tapir writes "While the Kindle Fire tablet consumed much of the focus at Amazon's launch event Wednesday in New York, the company also showed off a bit of potentially radical software technology as well, the new browser for the Fire, called Silk. Silk is different from other browsers because it can be configured to let Amazon's cloud service do much of the work assembling complex Web pages. The result is that users may experience much faster load times for Web pages, compared to other mobile devices, according to the company."
No. (Score:5, Informative)
Opera was doing this YEARS ago. As usual.
Frist?
Re: (Score:3, Informative)
Re: (Score:2)
iPhones did something similar (Score:3)
This has been done many times (Score:3)
Opera Mini isn't the only one.
One of the PalmOS browsers worked this way, doing pre-rendering at the other end to help compensate for a slow connection and a small device. If the ISP's server farm went down, so did your web browser.
I've used a satellite Internet provider that did similar as well, parsing the HTML at the provider ground station so it could fetch all the needed objects and send them in a single stream to the sky. This eliminated a lot of repeated fetch requests from the client over sat, whi
Re: (Score:2)
Potential privacy nightmare (Score:5, Interesting)
Nice performance bump for users, and an incredible data mining opportunity for Amazon - who wins more?
They Both Win (Score:3, Insightful)
Re: (Score:3)
Reading comprehension is a blessed thing - my comment wasn't who loses, but who wins more.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
No it doesn't.
I pay you $5 for a painting. Then I sell it for $10 (to a buyer who wouldn't have bought directly from you).
We've both gained $5, we've both got some costs to take out of that. Neither of us loses.
Re: (Score:2)
It's not always lose-win or lose-lose.
These days, it's pretty much a total loss for privacy.
Re: (Score:2)
From the summary:
Silk is different from other browsers because it can be configured to let Amazon's
It can be configured. Or it can be ignored. As long as it's an Opt-In feature, and not an Opt-out one, the right to privacy is preserved.
Real problem comes when they make these things "Opt Out", specially if the opt-out settings are as out of the way (and often buggy) as Facebook's.
Re: (Score:2)
It's not always lose-win or lose-lose.
Perhaps you missed the last word "more."
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
A certain three-letter government agency? I suspect that several such agencies would like a peek.
Re: (Score:2)
When people fear their government more than the government fears its people, it is a sad day.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
sift through a trillion porn
Maybe they want to use amazon as a human filter? You know, to get to the good stuff faster!
Re: (Score:2)
I'm somewhat confused. Why would they care? You're talking about a small percentage of web browser usage compared with all other browsers and platforms. Not only that but this is just a small percentage of the network traffic. What about instant messages, bittorrent and other formats of communication some of which will be completely bespoke?
No, I call bullshit. Some conspiracy theorists will happily sling around that an agency has their claws on the data but when you realise it is such a small percentage of
Re:Potential privacy nightmare (Score:5, Interesting)
Of course, google probably aggregates information from those ten servers anyway, and Amazon probably sells the information they collect on you anyway, and the government is probably monitoring everybody involved in any case...
Re: (Score:3)
Except they also said they were going to proxy your HTTPS traffic by making a connection from their "cloud" to the destination server for you. In some parts they call that a "man in the middle"...
Re: (Score:2)
Re: (Score:3)
From Amazon's own FAQ on Silk:
What about handling secure (https) connections?
We will establish a secure connection from the cloud to the site owner on your behalf for page requests of sites using SSL (e.g. https://siteaddress.com./ [siteaddress.com.]
Still a bit vague, but not the part about "from the cloud to the site owner on your behalf". But in this case nothing can be assumed - it's their browser, so they can implement the client to cloud connection however they want. Let's just hope they do it securely (even if, unlik
Re: (Score:2)
They are using SPDY [chromium.org] for the client to cloud connection, which is not only very fast but also https by design, you cannot have SPDY over plain http.
Re: (Score:3)
I can confidently say that the majority of people using Google aren't concerned about privacy at all. And if there was a major scandal they would either not hear about it at all, or quickly forget it.
At the end of the day most people either don't understand the issues sufficiently to worry about them, or flat out don't care.
The only way your average web user is going to be up in arms about something like this is if the tabloids picked it up and
Re: (Score:3)
Nice performance bump for users, and an incredible data mining opportunity for Amazon - who wins more?
Is it really a performance bump, though? I mean, when have you ever felt the load time for a page accessed through broadband was too slow?
If the Kindle Fire was running on a 33MHz Dragonball and accessing the net through a 14.4kbps modem, I could understand the need for this. But with a dual-core 1.2GHz processor and high-speed broadband, why do we need this? I'm still slightly confused at Amazon using this as a selling point ... (or perhaps it's a case of needing a selling point other than price, and dr
Re: (Score:3)
If it gives as much of a boost as Opera mini then it will be well worth it. One of the biggest performance increases comes from javascript being run on their servers before they send the page to your device. They give any onLoad events 2 seconds to fire and then cancel them so you don't get pages hung up waiting for flaky javascript that has hung for some reason or another. Any on page javascript is also processed on the server which massively reduces the load on the device itself.
It does lead to slightly w
This is a potential method to defear noscript (Score:3)
Users of noscript have long benefitted from fast loading of web pages as distracting ads pulled from other domains were suppressed.
If entire web pages are "constructed in the cloud" and then presented to users, the additional overhead of ads,
including annoying animation, would once again turn perfectly readable pages into aggravating distractions that
eventually drive readers away. Anyone remember answer.com? AskJeeves? Or cnn.com before noscript?
Bah humbug to this "improvement" in technology.
Re:This is a potential method to defeat noscript (Score:2)
Defeat! Sorry for the typo.
Re: (Score:2)
No I doubt most people do. Google.com loads up nice and quick though.
Even on pages that do pull in a lot of ads using javascript if this works anything like Opera Mini they won't be a problem. Opera Mini gives any javascript 2 seconds and then bins it which means an ad server that is slow to respond won't slow the page load down. All the javascript is run in the cloud and only a flat page is pushed out to the device so there isn't any overhead for javascript.
Re: (Score:2)
I grew up attached to a computer. I have ad blocking in my wetware. I simply don't even notice web advertisements anymore, unless they have sound.
I did recently start using adblock, but that was because I started noticing increased load times and companies tracking me with social media buttons. My visual cortex started filtering out banner ads years ago.
Re: (Score:2)
Can you imagine what happens when at an inconvenient time someone goes to their Amazon home page and it says "because you like www.clownsboinkingnuns.com we recommend this selection of clown masks"
Re: (Score:3)
Is that site slashdotted or something? I get a page not found error.
Re: (Score:2)
People who watch tons of porno. Think of how fast it will load.
I think the concern in that case is more the speed of un"load"ing.
Opera Mini (Score:5, Informative)
When you request a page in Opera Mini, the request is sent to the Opera Mini server that then downloads the page from the Internet. The server then packages your page up in a neat little compressed format (we call it OBML), ready to send back to your phone at the speed of ninjas on jetpacks.
Re:Opera Mini (Score:5, Funny)
This is just Opera Mini/Turbo (Score:5, Informative)
Opera released Opera Turbo [wikipedia.org] back in 2009 which does this same thing. As well, Opera Mini [wikipedia.org], their mobile browser, does this as well.
So this isn't really re-defining the browser, it's just bringing the technology more mainstream.
Re: (Score:2)
And the TMobile Sidekick by Danger (now Microsoft) did it many years before Opera. Nothing to see here. Move along.
Re:This is just Opera Mini/Turbo (Score:5, Informative)
You're absolutely correct that the basic "innovation" here is exactly what Opera Mini (note, not Turbo - specifically Mini) has done for ages. So all talk about "redefining the browser tech" is pure marketspeak, and both the submitter and the editor should be ashamed of spinning it the way Amazon PR wanted them to.
However, there is one crucial difference with Mini here: it also does work as a full-fledged local browser. Mini always does layout and other optimizations "in the cloud", and fetches the result. That's why it's so bad at JS, Flash, HTML5 etc - if it's something that has to run locally, it's not supported. Here, they are transparently offloading work on the server, but when there is something in the page that cannot be handled well that way - or when the server is not available - it gets rendered locally, same as in any other browser. So it's supposed to be completely transparent to the user, unlike Opera.
Of course, we haven't actually seen how well that it all works in practice, and I'll reserve my judgement until then. It'll be interesting to sniff traffic and see how much actually gets preprocessed; right now my suspicion is that on any script-heavy website, it'll mostly just do compression.
Re: (Score:2)
Actually it is not only opera turbo, they were the first, but google also has this kind of offloading facility, early versions of the android browser used it.
Pff, Opera does everything earlier (Score:2)
Lets face it, us few Opera users are used to living a couple of years in the future.
Mind you until 27 Septembet 2012 I sometimes got upset but since the new law enacted two days ago put to death all IE/Chrome and Firefox users (Lynx users already got their punishment through usage) I am a lot more mellow about it.
Re: (Score:2)
Please, explain to me how it differs? I've read over the linked article and I don't see anything amazingly different? It can render locally or on the EC2 cloud (like turbo). I will agree there is the addition of "learning". Is there something else we're missing?
Also, not an opera fanboy here. I tried opera mini when it came out. I'm a FF/Chrome fanboy.
Re: (Score:3)
so - if i have a hammer and i use it to nail a chair leg back on and then turn and use it to nail a table leg back on i've "refined" what?
so Opera's reason for doing this was to conserve bandwidth (image compression was only part of it) the other large but was the overhead of the requests and also the optimization of space in the transmission.
Amazon's is to optimize the data prior to the device.
they both do the same thing with minimally different options - so yes Amazon did what Opera has been doing for yea
Re: (Score:2)
As far as I can tell, there is no bandwidth saved using EC2, only processor time (and, in turn, battery life). The pictures aren't degraded in quality like they were on the Turbo.
Except that, according to TFA:
The service also uses content compression techniques, such as re-encoding video and images before sending them to a device.
Re:This is just Opera Mini/Turbo (Score:5, Funny)
Speaking of optimization, you can save a byte with a small change to your sig.
Your sig assembles to: A1 00 4C CD 21 (5 bytes!) whereas:
mov ah, 4ch
int 21h
assembles to: B4 4C CD 21 (4 bytes)
Interrupt 21h won't care what's in al, so you don't need to clear it.
You kids these days code like everyone has megabytes of RAM just lying around.
Re: (Score:2)
Bah, that should be B8 00 4C CD 21
Re:This is just Opera Mini/Turbo (Score:5, Insightful)
Interrupt 21h won't care what's in al, so you don't need to clear it.
Well, whoever spawned the process in question might care, since AL is the return/error code after termination!
You kids these days code like everyone has megabytes of RAM just lying around.
I would have thought you old timers had learned your lessons about skimping on what you assumed to be unimportant bytes ;)
Re: (Score:2)
Opera Turbo was all about compressing when connections were slow, not saving processor cycles for complex content.
Opera Turbo on notebooks, desktops and high end smart phones is mostly about bandwidth reduction, yes. But before introduction of Opera Turbo Opera was already using this technology for Opera Mini.
And Opera Mini uses proxy server to render pages that the CPU in a feature phone or older smart phone can not do. AND reduce bandwidth.
Re: (Score:2)
Opera fanboys.. lol, who would've thought?
If you didn't know Opera had fanbois, you've never participated in a Slashdot discussion about any browser tech before.
Re: (Score:2)
Those devices which existed before iOS/Android/WebOS/whatever are now sometimes called 'feature phones'.
Re: (Score:2)
Not that I approve of windows mobile, and it was a horrible experience, but -
I had a Windows phone in about 2004. I compiled mame for it, and it could play some games tolerably fast. The UI layout was xml based and could be adapted. It had a camera module and a reasonably high def screen, and could take MMC-RS cards to play video and music. Definitely a smartphone by any description you choose to apply to it, and it was far from the only one.
Re: (Score:2)
I don't know how to classify the old Windows Phone, I just know one thing the browser definitely was not smart.
Mandy Rice-Davies comments ... (Score:2)
And to paraphrase the immortal words of Mandy Rice Davies - they would say that, wouldn't they.
Opera? (Score:2, Redundant)
Re:Opera? (Score:5, Funny)
Re:Opera? (Score:4, Insightful)
Not sure if trolling or just stupid...
He was engaging in an obscure practice known as humor [wikipedia.org].
Re: (Score:2)
Yo dawg... (Score:2)
We heard you like the cloud, so we put the cloud in your cloud so you can swear while you disconnect!
Re: (Score:2)
We heard you like the cloud, so we put the cloud in your cloud so you can swear while you disconnect!
The next version of silk won't need a client at all. The cloud will be able to take the place of the user's device as well. Sure, you might end up spending thousands of dollars on items from Amazon you wouldn't have ordered, but think of the time savings!
How exactly is this redefining? (Score:3, Informative)
This is ridiculously old technology. Just about every other mobile browser does this now other than maybe IE on Windows phones and Safari on IOS. BlackBerry's have been doing this since 2005, as someone else mentioned Opera has had it since 2009, Bolt Browser has this feature as well. So I am to believe that a browser technology that's been around for 6 years is redefining browsers now? Way to grab on to an old feature and herald it as something new and ground breaking.
Re: (Score:2)
This is ridiculously old technology...BlackBerry's have been doing this since 2005...
Uh, wow. Maybe I'll hold off altogether on this "new" tech then. Seems in the battle of browser wars to find the fastest, easiest, and most efficient one out there, I don't ever hear someone exclaim "Oooh, you should go get a Blackberry!"...
Amazon Silk + SSL = MITM? (Score:5, Insightful)
What about handling secure (https) connections?
We will establish a secure connection from the cloud to the site owner on your behalf for page requests of sites using SSL (e.g. https://siteaddress.com/ [siteaddress.com] ).
So essentially, they become the man-in-the-middle so they can better cache your HTTPS content? And their browser is programmed to show this is acceptable/secure... What kind of privacy implications does this introduce? Even if their privacy policy says they won't use the data maliciously, cloud computing isn't a bullet-proof system (i.e., leaks, hacking incidents, etc.). Call me paranoid, but if I read this right, this sounds like a frightening idea.
Re: (Score:2)
Agreed. Proxies (like the Silk browser uses, even if fancy like Opera's OBML) *shouly* only proxy/format plain HTTP data. Any HTTPS connections *must* go direct from device to server: end-to-end.
I hope the browser displays a warning every time if it truly proxies HTTPS content! (And iconally shows the 'broken padlock' or jolly rodger.)
I read it as just another hop... (Score:2)
Re: (Score:3)
Call me paranoid, but if I read this right, this sounds like a frightening idea.
That's an entirely appropriate level of paranoia. What they're describing in their own help is exactly a MITM attack and extremely irresponsible. If the browser portrays that as "secure" it's fraud.
Re: (Score:2)
Cross posting from my old comment [slashdot.org]. As per their help [amazon.com]:
What about handling secure (https) connections? We will establish a secure connection from the cloud to the site owner on your behalf for page requests of sites using SSL (e.g. https://siteaddress.com/ [siteaddress.com] ).
So essentially, they become the man-in-the-middle so they can better cache your HTTPS content? And their browser is programmed to show this is acceptable/secure... What kind of privacy implications does this introduce? Even if their privacy policy says they won't use the data maliciously, cloud computing isn't a bullet-proof system (i.e., leaks, hacking incidents, etc.). Call me paranoid, but if I read this right, this sounds like a frightening idea.
Iif they put themselves as a man in the middle that sees your banking account credentials, credit card numbers, etc, all their servers that are involved in this should be subject to the kind of security standards and regulations that are required of sites that handle credit card numbers...
Re: (Score:2)
Re: (Score:2)
If this is only at the TCP level, essentially forwarding all encrypted traffic unaltered, then there is no issue.
But looking at the content is very serious. If the browser shows that it sends the data encrypted to example.com, but in fact it sends them in cleartext to proxy.amazon.com, it's a ridiculous security hole. I doubt they are doing this.
Re: (Score:2)
Not sure how long this will be useful (if at all). (Score:3, Informative)
Basically, what this service does is make a "google maps" version of the webpage -- cutting pages up into tiles (like the Nintendo NES did) and streaming them over a wireless connection from their reserved-for-holidays EC2 data centers. Some localized bastardization is involved, but the "google maps" img tiling is the basis of it.
A quick wget of the cnn.com front page yields 2.10 MB of data. And yes, it's less to tile it -- a screenshot at 1400x900, for about 40% of the page, converts into a lossless PNG file for about 700K of data. A lossy but usable 90-quality JPEG is around 350K. The processing time and RAM to bit blit that client-side of course will be a lot less than a modern ACID 2/3 browser would require.
But as sites become more dynamic, the response time to constantly stream pixels won't be worth it. And a lot of sites rely on being dynamic -- view the HTML source on Facebook some time, it's almost all JS. Even slashdot (famous for being HTML3 well into the 2000's) now feeds its stories dynamically with javascript and HTML5.
This isn't "redefining browser tech," it's probably a stopgap measure for their current market-undercutting $199 tablet processor. Anything JS/HTML5 runs fine on my dated Athlon X2 laptop on Chromium or Iceweasel, and that kind of speed will easily be in tablets in 1-2 years. Amazon says Fire is "dual core" but it's probably skimpy CPU-wise and/or RAM-wise. Or maybe their attempt to reinvent the wheel by rolling their own browser engine under NIH syndrome instead of using Webkit or Gecko just turned out badly.
Re: (Score:2)
The use of SPDY does seem likely:
"Software Development Engineers - SPDY
SPDY is an open source network transport protocol which we have leveraged in the design of Amazon Silk. In this role, you will have end-to-end ownership of our use of SPDY. You will be expected to have strong familiarity with the protocol and to use that knowledge to come up with innovative ways to improve the customer experience. We're looking for strong team players who thrive in a startup-like environment where flexibility is essentia
Awesome! (Score:2)
Newflash! It's WebKit (Score:2)
Injecting ads (Score:2)
Old Fogey... (Score:2)
I'm probably dating myself to all the kids on Slashdot, but when did web pages get so complex that work now needs to be split?
I still write all my HTML by hand, optimize my images, specify the actual size of the image in the IMG tag, yadda yadda.
I *never* understood the purpose of CSS, except that it mucked things up. I was very happy with black text against grey (woot Mosaic). Heck, my favorite browser was Omniweb for NeXT.
When exactly did things get so bad that web browsers are at their limit and these ta
Re: (Score:2)
Also the WAP protocol used in the early days of phone web browsers.
Re:Prior Art - Opera Turbo (Score:5, Interesting)
Opera turbo uses compression via opera's servers. Amazon's thing uses amazon's servers to render. With opera the point is to get around a slow connection on the consumer's side. Amazon's point is to do the render processing on amazon's side. Let's take an annoyingly busy website, for example: http://home.sina.com/ [sina.com] Now this beast can take a while to download and get ready, especially on a low power handheld thing like a tablet. Amazon's silk method should prep all those parts for the displaying device.
Re: (Score:2)
Opera turbo uses compression via opera's servers. Amazon's thing uses amazon's servers to render. With opera the point is to get around a slow connection on the consumer's side. Amazon's point is to do the render processing on amazon's side. Let's take an annoyingly busy website, for example: http://home.sina.com/ [sina.com] Now this beast can take a while to download and get ready, especially on a low power handheld thing like a tablet. Amazon's silk method should prep all those parts for the displaying device.
Um, yeah. So it does work exactly like Opera Turbo does. Opera turbo also down-sampled images to a lower resolution or lower number of colors which helped cut the download sizes quite a bit.
Re: (Score:3)
You do realize that in order to compress the data, Opera's servers have to render it first? The two technologies are more similar than they are dissimilar. From what I recall, Opera's approach is to pre-render on a proxy server, compress the end result, and send down to the device as a compact binary stream, and Silk appears to be doing pretty much the exact same thing but without any additional compression that I saw mentioned.
In any case, both have to pre-render the page and Opera's approach also removes
Re: (Score:2)
new [noo, nyoo] Show IPA adjective, -er, -est, adverb, noun
adjective
1. of recent origin, production, purchase, etc.; having but lately come or been brought into being: a new book.
2. of a kind now existing or appearing for the first time; novel: a new concept of the universe.
3. having but lately or but now come into knowledge: a new chemical element.
4. unfamiliar or strange (often followed by to ): ideas new to us; to visit new lands.
5. having but lately come to a place, position, status, etc.: a reception
Re: (Score:2)
Indeed, it is. Opera is also available on the iPhone. And the blackberry. And featurephones. And Mac. And PC. And Linux. I'm not sure there's any computing platform that it's not available for...
Re: (Score:2)
Opera Mini runs fine on Arm Linux.
Re: (Score:2)
Re: (Score:2)
Which arm platform?
My N900 runs Opera Mobile 11 just fine. You can pick that build up from http://labs.opera.com/downloads/ [opera.com]
They have just dropped support for Windows Mobile 6.5 though.
Re: (Score:2)
Just grab the generic J2ME version, along with the java runtimes and the J2ME wrappers.
Re: (Score:2)
Re: (Score:3)
What magical browser do you use that allows you to surf the web without any network connection?
Re: (Score:2)
Re: (Score:2)
Such as? Sure you can locally cache files but browsing in such a limited way isn't exactly what comes to mind when I think of "surfing" the web.
Re: (Score:3)
Re: (Score:2)
+1 Funny
Re: (Score:2)
I use Opera Mini on a shitty little LG phone. Works quite well, I can even post to Slashdot from it.
Re: (Score:3)
That whole fucking 2 seconds was killing me
seriously were not running 486's here aside from slashdots javabloat every other site does not suck on modern machines, hell if you can stand not pissing you pants even slashdot only takes about a min on a 300mhz PPC
So long?? It takes less than a minute on a Cyrix P-166 running firefox 1.0 and Win95 (yeah, baby!) (don't laugh -- it's the only machine we've got at work that still has ISA slots, which we need for a bit of equipment ....)
But I agree wholeheartedly with the point. Why would anyone wish to jeopardise their privacy to save a few seconds (max) of waiting for a webpage to load? The fact that Amazon can use this as a selling point is a sad statement on current attitudes to privacy.
Re: (Score:2)
It also uses its algorithms to know which links you'll most likely click on (based on what others have clicked on) and starts pre-fetching that data so if/when you click on the link it'll take even less time to load.
Unlike other pre-fetching technology that had no intelligence built in this sounds very awesome.
So not only does Amazon see all the data I'm loading, but they keep a record of it too??? What could possibly go wrong here?
Re: (Score:2)
You get one SSL connection from your phone to the Amazon server, and they get one from the Amazon server to the destination server with your credentials. So you have to trust Amazon not abusing their position by stealing your private keys. But Amazon pretends in their FAQ that their presence in the chain isn't a risk factor at all.