Comments for RIA: What is it good for?

RIA: What is it good for?

2007-06-01T06:10:11Z

This year there has been an explosion in Rich Internet Application (RIA) frameworks. The first major player to announce one was Adobe and its Apollo framework, which Richard covered back in March and I wrote about during ETech. Richard also wrote an explanatory post on Apollo. Next was Microsoft's announcement of Silverlight during Mix 2007. Mozilla has plans to implement offline mode in the next version of Firefox and this week Google jumped into the game with their offline framework called Gears.

Why this storm of RIA and offline functionality? As long as broadband, WiFi and mobile are all on the rise, why are we even talking about Desktop applications? Surely there must be good reasons and use cases for these frameworks. In this post we take a closer look at the intended use and marketing behind these RIA frameworks.

]]> What is RIA?

To understand these frameworks and the use cases, we need basic definitions. We start with the definition of RIA from Wikipedia:

"Rich Internet applications (RIA) are Web applications that have the features and functionality of traditional desktop applications. RIAs typically transfer the processing necessary for the user interface to the Web client but keep the bulk of the data (i.e., maintaining the state of the program, the data etc) back on the application server.

RIAs typically:

run in a Web browser, or do not require software installation

run locally in a secure environment called a sandbox

can be "occasionally connected" wandering in and out of hot-spots or from office to office."

The takeaway is that RIA is defined as a mix of three things: desktop-like UI online, offline apps that look like online apps, and online applications that can go offline. Together, these things are a bit confusing. But perhaps if we just said: a browser app with a rich user interface that has offline mode, then things would be simpler. So, to belabor the point, an RIA is a browser app - but it works like a desktop app (e.g. a rich email browser app that works like Outlook).

What is Apollo?

Adobe defines Apollo like this:

Apollo is the code name for a cross-operating system runtime being developed by Adobe that allows developers to leverage their existing web development skills (Flash, Flex, HTML, JavaScript, Ajax) to build and deploy rich Internet applications (RIAs) to the desktop.

Adobe emphasizes that their solution as a runtime. The company is marketing it as an RIA framework because it hits all three aspects of the RIA definition. The runtime takes care of the desktop and offline mode, while existing Flash plugins support the same function in the browser. But the company does emphasize the importance of downloading the runtime, which is separate from the browser.

What is Silverlight?

Microsoft defines Silverlight like this:

Silverlight is a cross-browser, cross-platform plug-in for delivering the next generation of Microsoft .NET-based media experiences and rich interactive applications for the Web.

Unlike Apollo, Silverlight is just a browser add-on and is not a separate runtime environment. Instead, it packages a .NET interpreter right inside the browser, making all .Net languages available to developers. And it achieves the same RIA goals - creating rich experiences in the browser while supporting an offline mode.

What is Google Gears?

Google defines Gears like this:

Google Gears (BETA) is an open source browser extension that enables web applications to provide offline functionality using following JavaScript APIs: Store and serve application resources locally, Store data locally in a fully-searchable relational database, Run asynchronous Javascript to improve application responsiveness.

Google's solution focuses primarily on offline mode. From that point of view, it is not really an RIA framework - instead it's an offline mode extension to the browser. It is not standalone, so like Silverlight it just enhances the browser capabilities.

So what are the use cases?

No doubt that these are all impressive frameworks, but what do we need them for? The companies have not gone to the trouble of doing all this just to satisfy the R in RIA. More likely, the primary target is support of applications in offline mode. A classic example is accessing email on a plane. For example, the fact that GMail is not available offline precludes corporations from dumping Outlook in favor of Gmail. Other examples where online apps could make use of offline functionality include project management applications, like Basecamp, and calendaring.

So being able to access some web apps in offline mode is useful. But is it also useful to build desktop applications that look and act like online ones? The answer to this question is far less clear. Why would we do this? the reason must be to leverage desktop functions such as access to local storage and the network. But this is also possible with Silverlight and Gears, right from the browser. So the use case of a full desktop application that looks like an online one, but does not run inside the browser, does not seem to be compelling.

Are all these RIA frameworks overkill?

So the major use case is offline support. While it is true that we still spend time offline, increasingly many of us are almost always online. With broadband, WiFi and mobile devices experiencing rapid growth, being offline is less of an issue. The major gap happens when you fly, although it appears that the solution to this problem is also on the way (i.e. on-board wireless internet access). When the internet reaches the skies, we will run out of reasons to be offline. So where does this put all of these RIA frameworks?

Conclusion

While there is an explosion of activity amongst major players delivering RIA frameworks, there seems to be just a single major use case - support of the offline mode. From this point of view, Apollo is a miss, but Silverlight and Gears are on the right track. Looking at the development side of this, Microsoft's solution is superior because it allows developers to build .Net applications. This means more robust (compared to JavaScript) programming languages and tools. But beyond this, if we are increasingly more and more online, why do we care about the offline mode?

Perhaps it is optimistic to think that offline mode will disappear that quickly - time will tell. But even today, it seems to me that we have already proven that web applications are better than desktop apps at juggling our information.

What do you think about this? What real RIA use cases do see out there? And what platform makes the most sense to you?

Comment from Alexander Marktl on 2007-06-03

2007-06-03T12:56:48Z

You have totally forgotten about the most important RIA technology that's out there: Flash

With the Flex framework you can build Flash based RIAs to the Browser, the Desktop and in future even to mobile phones.
Instead of betting on a technology that binds you either to the Browser or the Desktop, you just can deliver it all with one technology and let the user decide!

Excuse me, but you have totally missintepreted the whole RIA space.

-
Alex
http://www.apollohunter.com

Comment from 123fun on 2007-06-03

2007-06-03T07:07:08Z

I suppose the best case to be made for the desktop applications in RIA is portability. It's a compelling story that you have only one platform for all your application development needs, and maybe you can develop private desktop versions and shared web versions of the same app, for users with different levels of privacy concern, for example.

All of this RIA stuff seems to be about what Sun originally promised with Java a decade ago (fancy desktop UI in web sites, "write once, run anywhere", "sandbox" for security).

Comment from Tim on 2007-06-01

2007-06-01T18:00:24Z

Hi,

nice summary, but I'm with Todd Cullen here, that RIA should not be understood as an internet application with offline functionality.
I think we need a new definition for these types of software, which run in a browser or in separate frameworks (Apollo) and have offline functionality to store/retrieve data from local drives. Not to mention correlated ablilities like dragndrop, which just isn't possible in browser (yet).
Perhaps someone should make up a new definition for that, like enhanced or offline RIA.

Tim

Comment from MeTheGeek on 2007-06-01

2007-06-01T16:39:50Z

Good post Alex.

Save for the ppt issue (there is no good on-line alternative yet) I fully agree with Martin Edic: the 'desktop link' is a transition.

Computing is moving to "the browser". The industry knew this for a long time. It just took longer than predicted.

Comment from Peter Antypas on 2007-06-01

2007-06-01T15:49:14Z

It's really interesting I think how computing has shifted back and forth from centralized to distributed processing over the years. It almost feels like the industry can't make up its mind. My take is that there is no "right" way to do this: Whether processing should be server or client based depends largely on the application. I believe that RIA has the potential of making it easy for application developers to decide which model works best for their app and also allow them to change their mind in the process. My prediction: Whoever owns what I call the "automatic load distribution platform" will make the desktop OS irrelevant.

There's an added benefit to RIA apps although hard to prove: Computers consume roughly the same amount of power whether they're actually doing anything or not. Think of the collective processing power of all these desktop and laptop P4s (and Core Duos) which for the most part occupy themselves with rendering CSS. Most of these machines actually do go to sleep when they're not in use. Server farms, on the other hand, run 24x7 at full power. By having the ability to easily shift processing to the client where appropriate, we may be able to avoid adding more servers to data centers and thus do more with the same level of carbon emissions that we have now.

Just a thought.

Comment from John Dowdell on 2007-06-01

2007-06-01T15:48:46Z

"We start with the definition of RIA from Wikipedia...."

uh-oh... Wikipedia is a good place for generating leads, but not for determining reality!

The starting point for "What is RIA?" is the document in which the term was introduced, back in March 2002... here's a summary and link to the full PDF:
http://weblogs.macromedia.com/jd/archives/2005/03/ria_definition.cfm

There are enough other tangents in here that I'm not sure of the core, central point... I skipped to the conclusion and found it a bit counter to reality. But I'll take a stab at one of the questions in the conclusion:

"What real RIA use cases do see out there?" Any place where people need to interact with computer networks, and don't wish to be restricted to just mounds of text and a few pictures.

jd/adobe

Comment from Martin Edic on 2007-06-01

2007-06-01T15:00:55Z

I've gotta weigh in for the average user here. The plane situation is becoming the only 'average user' argument for desktop apps these days and as you point out Internet access on planes already exists, limiting this benefit. What no one here has mentioned is that software as a service (browser-based) is PC agnostic- it doesn't care where you are or what you're accessing it with. This means I'm not lugging my MacBook Pro back and forth from the office everyday because I can access my apps from my home PCs. Having a desktop connection is, IMHO, a transitory strategy tied to a belief that desktop apps are somehow 'better'. Sure if you have an app that requires Photoshop-level horsepower then you need a dedicated processor. However the 'average user' uses Word, PPT and Excel and an email account almost exclusively. All are available in varying forms as web apps.

Comment from Todd Cullen on 2007-06-01

2007-06-01T13:44:30Z

Couple quick points:

The RIA movement began with AJAX/Flash and has nothing to do with offline access to information. That is not to say that an RIA can't go offline but I don't believe thats the defining characteristic. Local system access (call it offline access if you want) is just a logical progression.

Silverlight is a competitor against Flash, the browser plugin, not Apollo the OS runtime.

Gears and Apollo already plan to play nice. Check this out: Article

Apollo is a miss? You can save data to disk for offline work.

Comment from herr ernst on 2007-06-01

2007-06-01T12:51:46Z

I just can't help but I don't like applications running in an browser for different reasons.
First of all a browser was made for browsing, not as an runtime for applications - for browsing documents, e.g. websites. I don't want an application to be in a browser. The browser toolbars take space, there is no fullscreen mode, when I quit the browser it's gone, etc. Second, Javascript is a very bad language for making GUIs.

The solution for me would be the following: Applications that run as desktop applications but can be started from inside the browser with just one click. Java has tried it with Web Start, but it's just too slow, complicated and does look ugly. I'm thinking about an cross-plattform user interface description language like (but not similar) to XUL, GladeXML or something like that. Therein, you describe windows, buttons etc. The runtime renders the widgets according to the operating system. Part of the software runs remotely, communicating via RPC or webservices.
Thus, you can combine the responsiveness of an desktop application with online connection and ubiquitous availability for everybody. This way you can also imagine the possibility for commercial providers to lease software.

Silverlight doesn't fulfill these conditions (only runs inside the browser), nor does Apollo (doesn't use native widgets, but some flash stuff, which is not as responsive and doesn't look like an desktop app; also, it uses Flash or HTML/Javascript for GUI, which is an dirty solution). I don't know enough about JavaFX.

BTW: I do not care about offline mode. It would have been necessary seven years ago, when I surfed via modem (but even then, there existed offline modes, e.g. e-mail via POP3). Now I'm always online, with the upcoming iPhone even when I'm on the road.

Comment from JulesLt on 2007-06-01

2007-06-01T12:28:57Z

As per the comments above, I certainly think a lot of the value in these technologies is opening up application development to web developers.

As for the value of something like Apollo - well, at the moment, RIAs are still in the novel phase, but once you've got 5 or 6 you regularly use, I can see a real value in having them appear as individual programs on your taskbar / icons on your switcher / etc, rather than as tabs within the browser, or one browser window per app.

Then there's being able to, say, drag a folder of pictures onto your Apollo app and have them uploaded to Flickr (or a video to YouTube). In fact, I can imagine an app where you dragged files in, and based on their type, put them with your relevant chosen host.

Of course you could already write something like that using Java, .NET, Obj-C, etc . . . but with Apollo web developers don't have to.

Comment from Valentin on 2007-06-01

2007-06-01T12:07:14Z

You forgot to mention JavaFX (it is afterall Java that pioneered (almost) all of this RIA stuff a decade ago)

Comment from Tushar on 2007-06-01

2007-06-01T11:58:41Z

Alex,

RIA is just not offline access. It is about creating a very rich interface for users to get information and avoid the site glitches. Just look at how adidas has created their shopping experience as an example

http://www.adidas.com/campaigns/verticalsrunning/content/2007/in/main.asp?module=GEAR&sub=footwear&opt=male

This is some fundamental shift in the User experience design where you can apply more effective usability, better productivity for the end-users.

Tushar

Comment from John Ballinger on 2007-06-01

2007-06-01T11:48:11Z

Hi Alex, I think the real problem here is the marketing guys and their use of the word "offline mode" its a really bad term used to describe an application state. So let me back up what I am talking about.

1. Offline mode really means persistent data storage on my hard drive. Apollo and Gears both use SQL lite, which means I can have a really powerfully tool to retrieve data that has been downloaded FAST from the local disk. The real value is not when my app is "offline" but augmenting locally stored data with real time updates from the web. I dont think many apps work at all "offline" even email. I dont ever email offline then try and connect to send emails. I am always like, I need the Internet, I need to see new emails. So offline is not really that important (or interesting).

2. Locally stored data (which is what currently offline mode is called) is a great way of being able to reduce server load and helping with the scaling issue, we have super powerful client computers but we still require all data to come from a hugely powerful server that is hard to scale. Why cannot we leverage the power of the client to help with scaling (easier said than done of course).

3. The browser sucks, it fully relies on the server to remember everything other than some cookie data. Frameworks like Gears and Apollo change this model by saying hey we can use this users disc to save some data. So many RIA are turning the currently security model on its head saying you can save info from the web directly to your disc for later use during this session. The real challenge here is security and keeping computers safe because this is the real challenge of RIA's that they are basically removed from the browsers sandbox which we all trust (to some degree anyway).

4. The best example of an app that I can modify to today is a data heavy Vineyard soil testing and geogrphical reporting tool. This relies heavily on getting the same data from the server in a different format for each query and while quick to start lags between each query and will require more hardware to scale. A more interesting issue would be to sycn the current users data to local database and query directly from the disc. WOW that would be cool. My development model of client server would completely change. Now I can actually develop against 1 code base locally. Now I wouldnt have to worry about the complexities of loading in Asynchronously XML data for processing each time I wanted something I would just query it. (Of course I have to build one generic module to sync my DB with the server but that would not be dependent on how my application was written).

So in summing up these new frameworks give more power to the application than we have previously got from the browser sandbox. Allow structured data to be retrieved via powerful SQL commands. Yet use the same technology that powers the web. This is exciting stuff but the real challengers lie ahead in how security is managed and how this doesnt create huge holes in a users OS with this new browser model. The same as the web page runs in a Restricted User Account it may be possible soon we will see new OS user accounts for RIA's apps to help box these things in.

John Ballinger

Comment from Tushar on 2007-06-01

2007-06-01T11:30:16Z

Alex,
RIA is not just offline access. In my view it is also about enabling rich user experiences and glitch free access. Look at how adidashas built its shopping cart (http://www.adidas.com/campaigns/verticalsrunning/content/2007/in/main.asp?module=GEAR&sub=footwear&opt=male).
That gives some idea of the power behind RIA. Such experiences will change the way in which transactions are done, information is accessed.

Tushar

Comment from Tomislav Pokrajcic on 2007-06-01

2007-06-01T10:15:37Z

Hey Alex, nice article, but it would be better if you took Flex instead of Apollo. Silverlight is comparable with Flex (not yet actually, but some day). On the other side Google Gears and Apollo have some goals in common, but I'm closer to compare Apollo with .NET in some aspects - they both need runtime framework on client machines and have pretty strong language behind the scene.
The difference betwen Microsoft and Adobe technologies is that MS starded on desktop and is expanding to the internet, where Adobe (Macromedia) started on the internet and now tries to take a part of desktop cookies too...

Tomislav Pokrajcic

Comment from James Key Lim on 2007-06-01

2007-06-01T08:40:58Z

This is the best summary of RIA I've seen so far Alex. Thank you.

I've seen an Apollo demo used for eBay where a user could create an auction offline and have it upload and pushed live the instant you get back online.

Although Ryan mentions that RIA is more about being able to build real desktop apps using web technologies... I do think that RIA applications will be very important especially with the increase in Software as a Service (Saas) offerings. One of the biggest complaints about certain Web 2.0 applications is that there is no ability to work offline.

Comment from Ryan Stewart on 2007-06-01

2007-06-01T08:08:23Z

Hey Alex, I'm not sure it's accurate to compare these three technologies because they fill very different needs. Google Gears could (in theory) actually work with Silverlight to enable an offline mode (Silverlight doesn't have that now). And Apollo supports occasionally connected applications, but does so entirely separate of the browser (it allows you to build desktop applications). Google Gears is very much a browser technology, so it doesn't really compare to Apollo.

I'd also venture that aside from Google Gears, the other technologies here aren't really about offline access. Silverlight was built to enable great experiences in the browser. It does things like video and animation which allow for some awesome user interfaces. There isn't really an offline focus for Silverlight. Apollo on the other hand is all about letting web developers build desktop applications. Part of that is being able to create "offline" modes because Apollo has access to the file system, but again, it's more about being able to build real desktop applications using web technologies than about taking web applications offline.

I'd love to chat more about it and get your thoughts, so feel free to email me.

=Ryan Stewart
Adobe Evangelist

Comment from Emre Sokullu on 2007-06-01

2007-06-01T07:22:18Z

Great sum-up Alex. I haven't hacked any of them yet, but sounds like Google has a great advantage here because Gears is the only one which offers a totally smooth transition from online to offline space. The programmer doesn't need to create a new application or write it in a different language. You just add a few more lines of Javascript to make your AJAX code runnable offline as well. That's what we need, nothing more.