Saturday, April 12, 2008

More on Google App Engine ...

It's not about whether it's ready for business today, or whether it has this language or that language, or even about lock in issues of SaaS, something which I've talked extensively about.

The innovation in GAE is in the provision of an open SDK and how open sourced standards can provide portability. Whilst I understand the concerns being raised by pundits on various forms of PaaS (platform as a service), I think a significant point is being missed.

People seem to be so immersed in the detail of GAE, that they're missing the bigger picture, even though it's jumping up and down shouting "look at me, look at me!" with a sign saying "big picture" wrapped around its neck.

Either that or I've gone mad.

7 comments:

phil jones said...

I'd say you haven't gone mad. These people have missed the wood for the trees.

The granularity has just shrunk again. It's now realistic to think of getting an idea for a small service, coding it up, and deploying it to the public, in about a *day*.

With that low cost of entry, you can afford to take the risk of lock-in and Google failure ...

If your prototype service shows any sign of traction with users, in week two you can start thinking about contingency plans for Google failure, migrating to other servers etc.

swardley said...

Ok, that's +1 for me not being mad.

I agree with your points, I just want to emphasise that the open SDK for GAE provides a starting point for the portability issues. All that is needed are a few providers to create alternative environments which map to the open SDK.

phil jones said...

Yep, realized that after reading your other posts linked from this.

Rather like the way Bebo allegedly cloned the Facebook API, no?

Interesting to speculate who might do it. I'd guess M$ are too optimistic / confident that they could come up with and own their own standard. Amazon have their own commitments.

The question : who's big enough to pull it off and offer a compelling service, but without sufficiently rival intentions to want to own their own standard.

Or do you see a consortium of smaller players? Such as web-hosting companies?

If it was cheap enough to do I could see someone like WebFaction having the smarts and profile to get into this.

swardley said...

The smart thing would be "a consortium of smaller players" who are under threat from all these large scale utility computing clouds.

"Such as web-hosting companies" - exactly, what other strategy do they have?

The only way that such providers can fight against this is to create a marketplace based upon open sourced standards. As a marketplace they can fight; as lone providers, their days are numbered.

The open SDK is an open sourced standard.

It's not to going change overnight, but if they don't move quickly, they are going to find that their positions erode over time.

I did talk to a number of them about 16 months ago about this, but back then they were of the mind that this was "never going to happen".

I'm sort of expecting that some of them are a bit more awake these days. However that does sort of depend upon people seeing "the wood for the trees".

swardley said...

As for M$ - well, I've posted already on my concerns about whether a proprietary computing mesh is possible and what this means.

It's also why I've commented before on the whole OOXML issue and the problems with believing that "open standards" (as in APIs) alone will create portability.

Now the open SDK at least provides an open sourced standard - it is still a long way to go from here to the portability needed for a competitive utility computing market, but it's a step in the right direction.

friarminor said...

True enough, Phil. GAE serves up the whole cloud computing but I don't think it will be the last of the biggies. Expect Sun, Oracle and even MSFT to join the fray.

But it is worth mentioning that there are 'small fries' making a living of this PaaS thing like Morph eXchange for Rails that is completely open source. Either they get blown away or innovate but the riks and challenge continue to grow bigger.

Best.
alain

swardley said...

Hi friarminor,

No it won't be the last of the big plays by certain and some of those plays may well be very dangerous for freedoms on the internet.

As for Morph eXchange, this is no different from Zimki back in 2006, except that Morph is using Ruby as its language (Zimki was JavaScript both client and server side) and Rails as its framework (Zimki had its own).

A key part of the Zimki strategy was to open source the entire stack in order to provide multiple competitors. Zimki was designed with ease of portability from one provider to another.

Morph itself maybe using open source components such as Rail + Ruby etc, but if it wants to innovate then it should open source itself. Otherwise it's just a SaaS vendor for RoR.

If Morph open sourced entirely and others set-up as Morph providers, they would get a small bit of a bigger pie but also huge support, exchange, compliance and other opportunities.

They talk about SaaS ecosystems, well if they want to do it - then open source entirely and encourage others to compete.

I hope they do well, competition is always good.