Open source from the inside out

What is the most effective way to implement open source solutions - from the inside-out or the outside-in?

There are two ways to tackle the migration when adopting any new company-wide technology solution into an organisation.

The first option is to begin with the deeper server-type systems at the heart of the infrastructure, and once this move has been completed effectively, start moving slowly from the inside-out to desktop and other peripheral systems.

Alternatively, start with the outer-most systems and work your way in. It would probably never be a good idea to overhaul the entire environment completely. That would entail dealing with shell-shocked employees and administrators.

But what is the more correct approach to these integrations - especially when moving towards open source and away from proprietary solutions?

Starting with server systems will affect IT management first and deal with users later. Starting with desktop systems will first impact users who will be adjusting to a new environment, before even implementing similar technologies at the back-end, which raises issues in terms of compatibility.

In the enterprise world it is interesting to note that the inside-out approach is widely regarded as the safest option, especially for open source. Part of the reason for this is that Linux has not been regarded (by most) to be desktop-ready until now. Open source and Linux-based servers, on the other hand, have been a dominant force in computing for a while and power most mail servers and name-servers, for example.

But the decision to move from the outside in for deploying open source, deals with more than just the value argument. It seems that phasing users in as the final step in the migration, once the server systems have been successfully migrated, just makes more sense.

Now that there are distributions of Linux available that are highly capable on the desktop, a move to open source on the periphery of the environment from an operating system perspective seems feasible as a final step. IT managers are then free to aid users in the migration once the back-end systems are up and running without needing much attention.

In the enterprise world [...] the inside-out approach is widely regarded as the safest option, especially for open source.

Muggie van Staden, MD, Obsidian Systems.

Operating systems aside, it is possible to roll-out open source solutions to desktop systems at the application level without too much disruption - as has been proven by the CSIR in its move to OpenOffice on thousands of desktop systems.

Financial institutions and other private enterprises, however, generally find it more effective on all levels to use open source solutions in the back-end initially and wait before rolling out to desktop systems.

It is therefore interesting to note that in the public sector, and specifically in the case of government, open source implementation is taking place at the desktop level first and making its way to back-end systems later.

The State IT Agency tender for open source was aimed at a desktop roll-out in government departments, and, as it has been awarded, is moving ahead. But why would government choose a different model for implementation to that of large enterprise - where large amounts of money are at stake? Unfortunately that question must be left hanging.

Of course, open source has much to offer desktop users. Now with the launch of Microsoft Windows Vista, which carries a larger price tag and more substantial upgrade requirements than its predecessor, it is highly likely that we will see more migrations to open source operating systems on the desktop, especially when the opposing factors for Vista adoption are coupled with the fact the Linux is widely regarded to now, finally, be desktop ready.

That said, the inside-out model still makes more sense at the moment. Of course, this raises the question - when, if ever, will the opposing model become feasible? It is my opinion that it will be possible in some cases to work from the outside-in, on an operating system level, in a year or two's time. While Linux may be desktop-ready now, it still has to prove itself to enterprise customers. Once it has enjoyed enough time in the wild, so to speak, and proven itself in terms of security and ease of use, enterprise customers might be familiar - and confident, enough to implement the likes of Linux at the periphery of their environment and work their way in.

Of course, this also means that with robust open source solutions available on both front- and back-end systems, it's not entirely impossible that complete overhauls may become possible - if users can be presented with familiar enough functionality.