Online Apps, Good or Bad?

Filed under Rants, Software Architecture

John Dvorak wrote a very interesting article for PCMag on 8/14, Google Pulls Plug, Everyone Misses Point.

I’ve been saying web apps were overhyped ever since SalesForce first came out and was getting tons of hype back in the 90’s, and that’s not because I used to work at a competitor (gotta love the wayback machine).

On the surface, it’s a great idea. No installation. No deployment headaches. No massive, multi machine upgrade pains. And with AJAX, and now SilverLight, you can get a user experience approaching a traditional fat client. Plus, with broadband nearing ubiquity, some of the bandwidth issues of the past are no more.

And for many installations, where users can connect to a centralized server within the company’s domain, it often is a very smart architecture.

So what’s the problem?

Dvorak points to the demise of the Google DTO/DTR program as exactly the problem. Imagine if it was YOUR BUSINESS that was running off the DTO program and Google pulled the plug. “Oh, you won’t be able to access your customer records after Aug 8th, but here’s a coupon for 2 bucks at Google Checkout you can use for the next 2 months.”

I doubt SalesForce is going anywhere anytime soon, and they do have a compelling product. I’m not even picking on them, per se. But the whole idea of putting that kind of trust into an app that’s running completely NOT under your control is just a little, eh, bothersome?

In fact, I tend to believe that the absolute BEST thing that could happen for everyone is for a few more biggies, like Google Apps, or SalesForce, or Microsoft MSN Mail, etc, to bite the dust and strand millions of people.

It’d hurt, but people would take a serious look at the implications of relying on ASPs to supply services like that.

There were some big reasons back in the early 80’s that the personal computer and the fat client app, as opposed to the mainframe and the dumb terminal, because so popular. Some of them have been forgotten, and unfortunately, it might take a few tough lessons to remind people.


  1. Ralf says:

    Amen. Web is fine for some applications, but is not the holy hand-grenade of business solutions. One of my competitors markets something that does 25% of what I offer, for 5 times the price. They’re doing quite well though, because (ta da!) it’s web-based. Potential customers get that new-boyfriend look, all dewey-eyed and bubbly when describing its wonderfulness:

    – It runs on the INTERNET!
    – I can work from home! (Silent thought bubble: "In my PAJAMAS!")
    – It’s all managed THERE so I don’t have to install software!
    – No more icky server boxes cluttering up my data center!

    …then they start using it, and realize all those things are TRUE, but then the facts you cite kick in I can only imagine some of the internal conversations. "You mean we no longer OWN our data? It’s in a server 2500 miles away?"

    Not to mention how reliable the internet isn’t sometimes (*chuckle*).

    "Smithers! Where’s my report?"
    "Uh… in a server 2500 miles away, with no way to get it?"

    The sad thing is, even when they realize their mistake, they’re locked in. To get their data back they must pay "conversion fees". Break the contract and the servers get turned off. Bad kitty! Software go bye-bye.

  2. Darin says:

    Personally, I think that a large band of business apps would benefit from becoming "browser based" without necessarily being hosted.

    By that I mean, when you buy the app, you buy the server portion that you host on your own server(s). Then, you decide whether it’s available publicly or only behind your VPN.

    Some outfits go that way, but people really seem to like the "hosted" approach, so, if you build it right, you hit both targets with one bullet.

  3. Ralf says:

    Anything of a collaborative nature with buttloads of data is probably best done as hosted app + thin client. It only makes sense. Everything else? I’m not so sure… there’s a reason Adobe Photoshop has to be installed on a workstation.

    If thin client was THE best way to deploy everything, we’d already be doing it. Larry Ellison’s magic box would be everywhere and we’d be laughing at those stupid fat PC’s we used to use. 128 megabytes of RAM? Who would ever need that much?

    The battle being fought now is primarily one of marketing: Web 2.0 RAWKS! If it runs in a browser it’s automatically better than the busted old crap our grandparents had.

    This meme will run its course, people will shrug, and ultimately we’ll go back to what we have now, sort of. Some stuff installed locally, other stuff not, but to the user it’s a seamless experience.

  4. Darin says:

    Any thoughts on these "click once" deployment projects you can do in .NET?

    Supposedly, you can create an app that effectively runs in a "sandbox" within the browser, but is a full on .NET executable, it just can’t touch anything local, only stuff on the server through an inet connection.

    I’ve wanted to play with that model, but I’m wondering if that’s just this year’s version of the old VB6 "User Document" or "Web Class". Never saw many people using those gizmos…

Post a Comment

Your email is never published nor shared. Required fields are marked *