Wednesday, November 23, 2005

It is and it isn't about the desktop

Industry analysts and tech junkies all seem to agree that the problems staving off Linux desktop adoption are focused on the availability of applications coupled with the complexity of implementing and maintaining those apps. The solution, they say, is to make more Linux-native applications available, and to standardize the technology to a point where mainstream users can really use it. In other words, to make it more like Windows in these respects. This is indeed a lofty goal.

In working with a large community of Linux resellers and system integrators, it becomes clear that the lack of certain applications for Linux does indeed keep customers from adopting the desktop platform. However, what also becomes clear is that the applications desired aren't going to be available for Linux. These key applications are Microsoft Office and QuickBooks. By volume, they are the most widely distributed and utilized applications in small business.

There are numerous alternatives to the Office applications, and the offerings in the open source community represent real contenders. But productivity alone isn't enough to get the desktop platform switched. Accounting applications are key when working with businesses. Finding a contender with similar usability and support to QuickBooks won't happen - Intuit has been at it for too long. And QuickBooks isn't going to be native to Linux any time soon. So, what is the answer?

The answer is utility computing. Sever-based computing, application hosting services, software-as-a-service... whatever the name - it's the answer.

Hosted applications can provide the desktop environment, integration, and applications that users want and that businesses require. When the applications and the computing environment are delivered, the local desktop platform becomes irrelevant and technical implementation becomes inconsequential to the end-user. This solves the two key issues identified as barriers to broad desktop Linux adoption in business.

The desktop hosting model, specifically, offers additional benefits to the business in terms of migrating applications from the desktop to server-based models, and then to web-based models. Once the requirement to locally install and manage applications and services is reduced or eliminated, then the overall network administration becomes easier and less costly. Users are able to adopt an "Internet-based" working model without having to adopt Web-based application technologies at the same time.

Once the business has adopted a virtual working model, then the ability to begin layering other applications and services - based on Web technologies - becomes seamless. Integration of Web-based application data with desktop application data is viable when the desktop application and data are also contained within the datacenter. Now, businesses can extend applications and data to all users in the enterprise, providing access to tools specifically oriented towards the individual job functions. Combinations of Web applications and desktop applications can safely and securely integrate, because they are all present in the managed infrastructure.

So - Hosted application delivery models enable the enterprise to use what it needs and wants in terms of applications, but provides the option of choice for the desktop OS platform. Isn't this really the answer - for Linux adoption and a better way of working overall?

5 comments:

Anonymous said...

The answer is utility computing. Sever-based computing, application hosting services, software-as-a-service... whatever the name - it's the answer.


If you're using networks services then why does Linux matter?

This kind argument seems to be throughout the Linux on the desktop adoption arguments with any attitude of "Linux can operate as an enterprise desktop but you'll just need to change all your applications"

I think an impediment to desktop adoption of Linux is the inconsistency of the user interfaces and their behaviours for the various windows managers.

In addition to that if you're creating an enterprise system you want to be able to provide users with consistent default settings.

But try to get that working.

Gnome and KDE change, they change a lot. Behaviours can change within a version. That combined with reliability problems such as gnome mysteriously freezing on some systems during start up and trying to use usb devices can make a Linux desktop demo turn nasty real quickly.

Seriously get the basics working first. Maybe then Linux on the desktop will happen.

Anonymous said...

DISCLAIMERS:
.This post conveys only my personal view, not my employer's.
.Trademarks belong to respective owners.

My reasoning follows these same general lines of yours, with the following observations/queries:

- people object to Openoffice on the grounds that it consumes a lot more of resources (CPU, memory) than Office. This is an even bigger problem in server-based computing, as differences add up, resulting in less users per server. I suspect it need not to be so; though Oo.o certainly is bigger on a PC, I guess things might work out differently in a multiuser environment, possibly because of different suite construction methods (Oo.o apps, IIRC, share more resources than their Office counterparts, even though these are better integrated to Windows). Does anyone know about a study or benchmark regarding this?

- after pondering about "must-have" apps, and the worthy aim of having free alternatives for those who can use it, there are two scenarios to consider:

a) installing free apps on Windows (on the server) -- like Oo.o/Thunderbird for Windows;
b) using Windows and Linux apps side by side on the same desktop.

The latter is difficult to obtain. The only way I enviewed was doing a two-stage approach, e.g., access a Linux server via NX, and from here a Windows server via ICA -- for instance. This is not a good approach, because of latencies it may introduce. But it would make possible to use Linux-only apps (like Evolution) -- data would be stored on file servers to be opened either by Gnumeric or Excel (for instance). Has anyone tried anything like this?

Thanks in advance.

Anonymous said...

Given that everything works as it is supposed to, then logical arguements against this model cannot exist. However, I have personally encountered those in the Linux Camp that insist this is not the solution...that LINUX programs must be ported or developed to meet these needs. It would seem that brand loyalty may impede natural progression for a time, but I foresee a time where the need of an operating system will give way to the abilities of a service provider. Still, there will be those who are platform-loyal and will fight against what you suggest. It will be interesting to see how long it takes for the dust to settle.

helios

Anonymous said...

Gnome and KDE change, they change a lot. Behaviours can change within a version. That combined with reliability problems such as gnome mysteriously freezing on some systems during start up and trying to use usb devices can make a Linux desktop demo turn nasty real quickly.

I would argue that kde is really going on 0.4.0 and gnome is on 0.2.12

Both of the two desktops have a long ways to go until they work like the command line tools.

That said, it looks like they will eventually get a usable desktop on linux. (and it may be a union of the kde, gnome, and gnustep)

Matthew Miller said...

So.... 1997's Corel Office for Java?