Bernard Golden is a renowned expert on open source software and author of the excellent "Open Source in the Enterprise" recently published by O'Reilly. We caught up with him at the HP Finanical Services industry Open Source Advisory Council to ask him how Open Source is changing the way Enterprises use software.
Q1: Can you tell us a little bit about yourself and why you wrote the report "Open Source in the Enterprise"?
I have been consulting in open source and enterprises for about six or seven years. I know folks at O'Reilly, and when they decided they wanted to examine the enterprise open source phenomenon, they asked me if I'd participate. I was delighted, especially because O'Reilly has some very innovative data mining techniques that would allow us to examine real data about enterprise adoption beyond the usual anecdotes.
Q2: Why are enterprises adopting more and more open source software?
In the report, we identified six drivers for enterprise open source adoption:
agility and scale
quality and security
breaking vendor lock-in
cost
sovereignty
innovation
Depending upon the enterprise, one or more of these drivers might motivate their use of open source. Often, the initial interest is from a cost perspective, but as the organizations gets more acquainted with open source, the other drivers come to be seen as important.
Q3: How do you measure the use of open source adoption in enterprises?
Rather than conduct surveys, which are usually suspect in their methodology or sample set, we looked at open source recruitment postings in Fortune 1000 companies, reasoning that for a company to get to the point that they want to hire a skill, it's likely that they are using the product. We found that about .2% of all F1000 job postings were for open source-skilled positions (our definition was that two open source products needed to be called out in the posting for it to meet the threshold of being open source-oriented). Given that total IT employment in F1000 companies is about 2.4%, this means that around 10% of all IT jobs are open source-oriented.
Q4: How important do you think being able to inspect an application's source code really is? Is it a red herring or do you have an example of where this has been critical in an enterprise's use of open source software?
Certainly a number of US Government security agencies have used the transparency of open source code as a reason to use the products; they are able to examine the code and satisfy themselves that it is not compromised with respect to security. Beyond those situations, IT organizations find the ability to study the code relevant when integrating with an open source product, so as to understand its behavior better and thereby create a better end product.
Q5: What value do the communities around open source projects have to enterprises?
Community is absolutely critical; absent community, open source is little better than source code availability through escrow. The Open Source Maturity Model, which I created to enable formal assessment of the maturity of open source products, measures the maturity of a number of product elements like support, documentation, and professional services, all of which increase in maturity as the product community grows and matures.
Q6: Will open source ever shake its 'beard and sandles' reputation? Or do you think its slightly alternative image is a strength rather than a weakness?
Its outer image served to attract adherents early in the open source movement. Most technologies, to achieve mainstream adoption, have to be seen as mainly being used by mainstream organizations. I feel that as more 'typical' IT organizations adopt open source as a key part of their strategy, open source will come to be seen as a mainstream choice. I'm sure, however, that there will always be tie-dyed t-shirts in the back room of the open source ballroom!
Q7: In the report you identify six core drivers for open source adoption by enterprises. Which is the most important and why?
As I mentioned earlier, the importance of the drivers varies according to the goals of an IT organization. Cost is an easy driver to initially identify, so many people consider that the primary motivator for open source use; however, as companies become more familiar with open source and integrate it more deeply into their strategies, they find other open source drivers extremely relevant to their situations.
Q8: Why should ISVs consider using open source software as their development platform?
For ISVs, there are three primary reasons:
Cost – with constant pressure on margins, using open source in place of licensed third party software components or self-developed components can reduce COGS significantly;
Time to market – leveraging existing open source rather than developing all necessary product components internally can shorten the time it takes to deliver a product to customers; and
Competitive secrecy – licensing a third-party component can signal market intentions to another company and thereby offer them an opportunity to respond competitively; using open source that is available anonymously allows a company to develop a product without offering information to other interested parties about what it is doing.
Q9: Why does open source licensing improve code quality?
Because so many more people – with many different skills and perspectives – examine an open source product in comparison to a closed source product, more attention can be paid to any particular piece of code, thereby raising the likelihood that bugs will be fixed. In addition, the availability of source code means that anyone that is concerned with a bug can directly work on it, avoiding the need to wait for a vendor to address it.
Q10: How does open source licensing improve security?
Just as quality benefits from, as the saying goes, many eyes looking at the code, the same is true for security. More people, with different skills and perspectives, can examine the code and address security issues. This is far more direct attention than is typically possible within a closed source software company.
Q11: Why does open source licensing enhance the quality of support?
In proprietary software companies, support is usually seen as a necessary evil, and a cost drain, detracting from company profitability. Consequently, support is usually understaffed and peopled by the lowest-possible skill set. Commercial open source companies derive the majority of their revenues from support, so they focus on delivering high-quality support services. This can be evaluated in light of the fact that Red Hat has consistently been awarded very high marks for support quality.
This does not even address the quality of community support, a very active aspect of open source, where peer members using the product can offer real-world and real-time feedback on problems encountered while using a product.
The combination of community support and high-quality commercial open source support provides very effective support, much better than the proprietary alternative.
Q12: In cost terms, what is the critical advantage of open source software over proprietary software?
The absence of license fees means that open source is usually less expensive, since at most only support (subscription) fees are required. Moreover, open source avoids the oft-encountered “mandatory upgrade fee. Finally, because open source is more of a pay-as-you-go cost model (in contrast to the front-loaded fee structure of proprietary software), open source matches value to cost much better.
Q13: How does open source licensing encourage innovation?
Obviously, the ability to modify a product to better suit the use cases of a company enables innovation for it – it can create a product better tuned to the needs of its customers. Beyond that, though, companies can make open source libraries available to let their customers experiment and innovate, thereby leveraging the better insights and inventiveness to improve the original product.
Q14: How do you suggest enterprises go about adopting open source technologies?
In the report, we identify three sets of recommendations: early adopters, mainstream users, and innovation seekers. As you might expect, it's a developmental approach in that early adopters, once they get more experienced with open source, apply it throughout their infrastructure so that it is a mainstream use pattern. Likewise, once a company that applies open source widely throughout its infrastructure examines what other ways they can apply open source, they begin to evaluate how it can help achieve innovation. There isn't enough space to describe the elements of each of the recommendations, but an extract of the report is available to download. Of course, the complete report goes into significant detail about how to implement each of the sets of recommendations.
Q15: What common mistakes do enterprises make when taking on open source software?
The primary mistake most enterprises make regarding open source is failing to understand how much it is already being used throughout the organization – it's just invisible, since it's not wired into the processes of the IT organization. It's critical to put standardized governance into place to ensure open source use moves to transparent from invisible.
About Bernard Golden
Bernard Golden is a recognised authority on Open Source software. Called a "renowned open source expert" (IT Business Edge) and "an open source guru" (SearchCRM.com), he is regularly in magazines like Computerworld, Information Week, and Inc. His blog "The Open Source" is one of the most popular features of CIO magazine's website. He is a frequent speaker at industry conferences like LinuxWorld, the Open Source Business Conference, and the Red Hat Summit. Bernard Golden is the author of Succeeding with Open Source (Addison-Wesley, 2005), used as a course text in universities throughout the world. He is CEO of Navica. a Silicon Valley-based management consulting firm focused on open source and innovation.
A recent post has introduced me to a term with which I was previously unfamiliar. 'Click-bait' was the epithet used by a US reader to describe the title of my blog and it both intrigued and disturbed me. It means self-evidently that the title is more attractive in a sensationalist sense than the article merits. The post below is hardly sensational, it's all about how Linux can speed up your computer hence the sub-title 'Linux feels the Need for Speed'.
However a brief whizz around the computing licencing at CERN (which is currently making headlines with the switch-on of the Large Hadron Collider) shows that the biggest computing development to emerge from there since the World Wide Web is the EU Grid which is released on an Open Source basis. Hence open source computing is arguably at the heart of the LHC and if so will play its part in the 'End of the World' if the more apocalyptic of our brethern are to be believed. And, since the LHC is all about really great speeds the conflation was too tempting. So apologies for the click-bait. What follows is really more prosaically about managing the end of a relationship.
Leaving XP
It's a simple question but difficult to answer: 'How do you get folk who are perfectly happy with Windows XP to change to something else?'.
In a previous post I suggested that to a whole generation, a Windows XP desktop was for them a finished work, the culmination of a succession of exciting upgrades of hardware and software.
In the Open Source world at one time, fairly recently actually, it was received wisdom that although technical superiority would win out in the server market getting ordinary users to change to unfamiliar desktops was a step too far.
I am happy to say Microsoft has run square into the self same unfamiliarity problem with Vista and Office 2007. They are a bit too different to the 'finished work' without offering any must-have extras. It gets even harder for Microsoft when even official Governmental bodies like BECTA advise the public sector procurers not to change. Ironically this stricture appears not to apply to other Government organisations in education such as the QCA (Qualifications and Curriculum Authority) who have clearly more money than sense.
Microsoft though has enough clout to follow alternative strategies for persuading its customers to change when they show signs of dragging their feet. The most obvious is by not allowing vendors to install XP on new machines and making sure lots of stuff, bit by bit, won't work on the old machines (allegedly). The Open Source world in contrast with its plethora of cool Linux distributions and manifest lack of clout (on the desktop) only has the 'hey that's a cool desktop - I must change' strategy to fall back on and that's a pretty weak opener in the desktop wars.
Why indeed would one now change desktops, why in the past were we so willing, eager even, to do just that and now are so reluctant?
Reasons to Upgrade
Readers of a certain vintage will remember the Intel hardware upgrades from the 286 to 386 to 486 to 586 Pentium processors, running parallel, Microsoft's OS went from Windows 3 to 3.2 to 95 to 98 and 2000 and Word went from Word 2 (very good it was too) through 6 to 97 and 2000. The amount of RAM fitted to a PC went up during this time from 16mb to 1gb. We all handed over our cash as soon as we could to experience the latest thing.
Of course power consumption went through the roof at the same time but we did not care much back then... No, what we did care about was speed. 486 owners were lightning fast when running Word 2 when compared to lowly mortals who only had a 286. I really coveted the next upgrade, I really did.
Unfortunately but seemingly inevitably, what was also happening during this period was the emergence of 'bloat ware'. Software got more features and more code to take advantage of the new hardware power until things became absurdly bloated.
The tale is told that even Microsoft's own engineers struggled mightily to upgrade the highly evolved XP to Vista's extra feature list but were defeated by the mighty code base and effectively started again using the simpler 2003 Server code. If true, this story provides an exemplar in what the Science Philosophers call 'paradigm elaboration'. Ultimately the accretion of ad-hoc modifications causes the edifice to collapse.
To cut a long tale short - as a result of bloated software, most computers are no longer fast, period. They are dog slow. MS Office 2007 is huge, so is Open Office (sorry guys). Running say Office 2007 or Open Office 2.4 on a budget laptop with Vista Home Premium is a dismally slow experience compared to the same machine running XPpro and Word 97/2000. Don't try this at home kids.
All good things come to an end however and this applies to this particular upgrade cycle. The emergence of new technologies may complement the status quo but sometimes they disrupt it. Those described below fall into the later category and have occurred at a time of hiatus in the prevailing paradigm.
Top 3 Disruptive technologies
The new Netbooks (so-called) low power (typically 6 watts) ultra cheap sub notebooks with 7-10" screens running a flavour of Linux (or at a push the undead Windows XP reincarnated by MS just for netbooks I wonder why?)
'Lightweight' operating systems and applications which use far less code and system overhead than their bloated equivalents.
Google's Chrome browser (or is it an OS or merely very cool spyware?).
What is so disruptive? Read on and then add them together and it will become clearer.
Quick Web
Just how disruptive these three will be will depend on a range of factors but I think the biggest driver will be as it was before, speed. Google seem to think so too. The open source browser application Chrome, out in beta for Windows, is a really really fast browser. Yes, it has a range of cool new features that will emerge as powerful incentives to use Chrome and yes, it may be the worlds most effective spyware , but the first encounter the user has with the browser produces that great feeling, raw speed.
Chrome on Windows XP now not only speeds up browsing but its 'Add Application' feature makes using on-line applications (eg Google docs) simpler and much faster. One more reason then not to upgrade your PC but to speed it up with software choice. When its out for Linux frankly I can't see why I would use any other browser given that I have already sold my soul to Googleplex's logging computers.
Quick Boot
The netbooks such as the now well known Linux Xandros Asus EeePC which a year ago pioneered their introduction and proved to the world that there was a huge demand for such devices (thank you Asus) are not in themselves as supplied that quick. They after all have modest hardware specifications (and tiny power consumptions) but thanks to their embedded operating systems, boy to they do boot up quickly. Not quite instantly-on but only a dozen or so seconds.
Several mainboard manufacturers (New ASUS mainboard has 5-second bootup) for conventional desktop and laptop computers have cottoned on to the agonizingly slow boot ups endured by users and are increasingly offering 'fast boot 'options: yes you have guessed how, they use light embedded Linux distributions on board which offer the basic applications of word processing, browsing and so on. After a few weeks of no-wait computing, how many of you are going to opt for the 'agonizingly slow boot option' so that you can use Office 2007 to type that memo? A quick boot though is no good if the subsequent experience is slow.
Fortunately thanks to a few 'mad' developers who in the true Open Source tradition, ploughed their own furrow even when their project was not fashionable, we now have software which is truly non-bloat, let's call it 'quickware'.
Quick Software
In recent years a few developers dedicated themselves to stripping down full Linux distributions to produce lightweight distributions capable of running quickly on older computers or very fast on later models. Notable amongst this dedicated group are Puppy Linux and DSL Linux (Damm Small Linux). The entire distributions including applications were under 100mb and ran as so called live distributions. Essentially this meant booting from CD or USB Memory stick and running in RAM space.
These distributions are, or have been to say the least, for the minority... oddball Linuxistas. I confess to be a Puppy fan and have received my unfair share of derision in the Office for my visionary abilities but trust me one day quickware will rule the world.
How many of us Linux users are more productive with Compiz et al and every package under the sun installed as default? Linux has shown it can match and beat the eye candy of Vista Ultimate and MacOSx. Maybe now the thrust will be to make the desktop as fast as stable and as well engineered as our server products.
The following distributions have weighed in and show a very active rate of development:
wattOS : an ultra light weight, Ubuntu-based Linux distro.
I think the point made by this non exhaustive list is the day of the speedy distro is nearly nigh.
Take any of the above and boot it from USB onto a bare metal computer sporting a pentium class processor and half a gig of ram and it will fly; all we need now are some speedy applications to help it on its way. In the Open Source 'Office' portfolio let me suggest AbiWord as a replacement for Open Office or MS Word. It'll do all you want and will open in a twinkle of an eye; ditto Gnumeric instead of OO Spreadsheet or Excel. Don't believe me? Download them and see for yourself. Then there are Inkscape, Scribus....
It's all out there.
Conclusion
There is now the possibility that sanity may be coming to the PC's desktop. In many ways the XP generation with whom I opened this post have a point. XP circa 2001 does all they need it to do. Any improvement (for them) would be merely to do it all faster.
Performance has always characterised the server market and consequently I assert accounts for the superb growth of Open Source server deployments. The desktop has in contrast, especially recently, been characterised by 'features' to the detriment of performance. Computers have grown ever more powerful to take advantage of ever more irrelevant interfaces, or is that vice versa?
Now thanks maybe to an economic downturn or global warming or the imminent destruction of the universe by minitiure black holes or whatever, many users have had enough.
The solution is to break the upgrade cycle; just use free speedy software and become more productive! Open Source will love you, Google will love you, your boss will love you but I can't guarantee that Intel and Microsoft will feel the same.
For me, my next computer will be a GDium 10" notebook booting LDXE from an 64gb USB stick and with a Chrome browser. I can feel your envy.
My principle interest has always been the use of Open Source thin-client solutions in an educational context. Its worth stating, again, what is really the blindingly obvious - thin-client work stations use one twentieth of the power of a typical PC (10-20 watts versus 200-400 watts), require no maintenance or technical per-machine support and Open Source software is free of licence costs.
Thin-client networks should be 'no-brainers' for schools trying to meet carbon targets, giving value for money and eking out scant human resources.
Needless to say thin-client deployments in schools are as rare as hen's teeth!
But thin-client solutions just will not go away, and for good reason, it's just that it is very hard to dislodge the incumbent fat boy PC and their fatter still vendors.
Cries of the death of the PC are frequent and always come to nothing. The only major vendor who has doggedly beaten the thin-client drum is Sun Microsystems with their SunRay workstations.
Microsoft's RDP thin-client servers are admittedly ubiquitous but invariably they are used as remote (expensively licenced) add-ons to a conventional PC network.
The big software vendors fear loss of revenue from diskless low power work stations. Sun's premium 'blue chip' pricing hardly encourages new customers and MS fear the loss of revenue from their per PC licencing.
In other words the market for thin-clients is repeatedly announced and then killed by the interests of corporate business models.
How thin-clients will change education (really)
You can't keep a good concept down however and there was a inevitability that the Open Source community would be at the cutting edge of innovations in this area.
However let's deal with recent hardware developments first as these impact on the whole scene:
Thin-client Laptops
Even 18 months ago it was a difficult search to find thin-client laptops and notebooks. Today this market is overflowing with offers. Wyse, Lenovo, Comet and even Dell (supplied if not branded) offer disk-less wireless notebooks for businesses and schools. The compelling sales pitch is that these devices contain no persistent data that can be left in the taxi or lost in the T5 baggage handling void.
In other words serial data loss incompetence and the fear of future losses from public services, school databases and others has driven a huge change from the PC Laptop. This alone may ensure the rise of the thin-client solution.
Of course the data and applications for these notebooks has to be stored and supplied by a thin-client server. We will deal with this later.
Cost
It is a while now since highly specified thin-client terminals with decent graphics broke the £99 barrier and became available with power consumptions below 10watts. This trend shows no sign of abating as Intel Atom chip is released and commodity hardware costs fall. £50 and 4 watts should be all an office needs for its everyday work station. As we said above this should be a no-brainer set against a £400 PC with 400 power packs and per seat licences.
Gaming Consoles
Gaming consoles are very much overlooked technologies for those of a certain age. Computer games are played on high spec PC's, MS XBox, Sony's PS3 and Nintendo's Wii. PC's Xboxes and PS3's all use much the same power; 200 watts or so when gaming, 1-2 watts on standby. The tiny Wii however uses 10 watts when gaming and 1-2 watts on standby.
The Wii, which is currently the best selling console, is very much a graphics-competent web-facing thin-client. Opera indeed produced and support a version of its browser just for the Wii. This means that, in effect, regarding Web2 applications the Wii is a modern thin-client for everyone.
Forget 17" monitors and think HD TV instead. From an educationalist's point-of-view there are some very interesting downstream consequences of this development, more of that at the end of this post.
Server Software
Thin client hardware is, of course, nothing without server-side software. In the world of Microsoft we are well used to the ageing RDP server and in the Open Source world we have the Linux Terminal Server Project (LTSP).
Other new server-side solutions are very exciting indeed:
An Open-Source Hybrid Thin-Client Project from SafeDesk Safedesk is a new Open Source project that uses Debian Live Net to create a Terminal Services Environment. It claims full local device support such as USB, and a full virtualisation of operating systems such Win XP.
openThinClient A free Java 6 product provides the server and client software which can be installed on any existing hardware.
Xandros Server 2 The Linux distributor Xandros has collaborated with NX NoMachine to produce a thin-client server with powerful virtualisation features and a variety of boot options.
Nydio and Userful Two separate offerings based on Open Source software which effectively are PC replicators. One PC is used to run 2, 4 or 10 users using separate keyboards and monitors.
The products above are very intriguing in themselves. On one hand they represent a renewal in interest in the terminal server, with the added twist of the energy saving virtualised server suite, but they also show an innovative approaches to sharing out the excessive CPU power of the single PC.
Whatever the approaches it seems economic and environmental imperatives now mean that mindset has changed and the emphasis is on making best use of computing power rather than building gas guzzlers with huge operating system overheads such as demanded by Vista.
Web 2.0 and Thin-Clients
The Net-pc came and went 10 years ago. Web applications and revenue models had not advanced to the point of usefulness but all that has changed emphatically now. Perhaps the most powerful illustration of how things have changed can be illustrated by the following (actual) scenario:
The home-educated student has logged onto the web and is using the Open Source program Second Life. She is using her Wii as a thin-client web terminal and her HD TV to attend 'school' where, in addition to accessing her teachers' avatars, she can access educational content through the Open Source VLE Moodle and Google's Apps.
Maybe the classroom of the future will come to you via be Open Source software and Nintendo's hardware!
Summary
The death of the PC is predicted once again. Of course I will be wrong like all the others before me. Personal Computing is so seductive that it will morph into ultra-cheap low-powered devices that hybridise the web thin-client with the personal device. Even Dell are aiming to release a sub $100 Linux (Ubuntu?) notebook. What I can say, however, is that the day of the big beige/black box is stone dead maybe it will take a major operating system vendor with it.
