While things remain quiet overall with Linux development, I took some time out to interview a good friend of mine who I have known for many years through a popular Australian Bulletin Board. I know him as a passionate user and advocate for free and open-source software.
But as I learned whilst conducting this very interview, there’s always a lot more to a fellow Linux user other than Linux itself. As you’ll read in the interview ahead, Daniel Mons has worked with some interesting companies, which we’ve decided to leave the company names vacant from the interview for obvious reasons. Dan has been involved with Unix, Linux, Windows and Mac OS X based systems for most of his career. So his experience and knowledge is vast. And for a man only just approaching his mid-30’s, he has certainly seen a lot of changes in the Unix and Linux industry. But what really provided the basis for this interview, I knew Dan could offer some interesting insight to the Unix versus Linux in the enterprise sector debate. His experience and positive views make him well-deserving of my respect:
Chris: Introduce yourself to Unixmen and tell us a little bit about yourself?
Dan: My name is Daniel Mons, and I’m a 33 year old father of three living in Brisbane, Australia. Other than open-source, my passions are retro gaming and fitness.
Chris: When did you first get involved with Unix and/or Linux on an enterprise level?
Dan: 2001 was my first full-time paid role using SCO UNIX, which was awful (both the job itself, and having to use SCO after learning Digital UNIX and Linux at University). Later that same year I switched to a much better job and had the opportunity to replace a lot of legacy Microsoft software with Linux pieces. (File servers, proxy servers, mail servers and others). It was great fun and it saved the business a lot of cash.
Chris: Are you currently working in a role with Unix and/or Linux? If so, what Unix and/or Linux systems and distributions are you involved with?
Dan: The last two years have seen a fair bit of change for me. I went from a 4 year role working on RHEL and HP-UX with applications written in COBOL and Java/JBoss for a large financial firm, to a company that builds web-based gaming systems that shuffle a lot of funds around the globe running community/unsupported versions of Linux, JBoss and Tomcat. Recently I switched again to build a Visual Effects Studio inside a larger production company, again with a heavy emphasis on Linux, but also integrating a lot of Mac OS X and Windows based systems as well. The studio is gearing up for a feature Hollywood film, so a combination of automatically deployed workstations, clustered storage and a render farm all need to be built in preparation.
Chris: What do you think of the current state of Unix specific deployment at an enterprise level? And do you think that Unix has a long-term future in the enterprise sector or do you believe the future lies with Linux based server systems?
Dan: UNIX will certainly be around for a while. Particularly in large organizations, change is a very difficult thing to push through. Part of my role back for the first financial organization I mentioned before, was to migrate them off HP-UX and on to RHEL for their largest COBOL application. After 4 years, the team had built plenty of new RHEL infrastructure for new services. But the migration of the legacy applications was totally reversed, and the decision was made to stick with HP-UX. Despite constant probing by the Linux/UNIX team, it seems the decision was made with little regard to technology, and much more to do with keeping a status quo.
However, particularly in a post-GFC world, the cost of UNIX is beginning to add up. While die-hard UNIX fans will claim the cost is worth the extra safety and peace of mind, enterprise Linux fans will understand that commodity x86 hardware offers redundancy outside of a single box. Linux’s lower hardware and software licensing and support costs mean that even if it requires 3 Linux boxes to deliver the same redundancy and safety of a single UNIX box, the cost is still much likely to be lower.
Add to that the ability for Linux to exist far more easily in a mixed environment (sitting side by side with Windows servers on VMWare infrastructure, for example), and the cost to deploy Linux drops once again. Linux’s flexibility in tackling either enormous enterprise workloads or small appliance-style tasks means it’s a flexible choice that covers a lot of ground without the need for yet another specialized knowledge base or department to look after it. And the popularity and growth of cloud technologies mean the mass deployment and configuration tools are making their way back into enterprise setups, making Linux even easier to manage at large scale.
With all of that in mind, I can’t help but think there’s an enormous future for Linux, and that UNIX sales will be under pressure in the coming years. I don’t think UNIX will die or anything dramatic like that. The continuing success of Mainframe has shown us that specific workloads are still better suited to these giant expensive boxes, although ironically the availability of Linux on System Z has probably had a lot to do with that success too.
“Big Data” is a buzzword that continues to get thrown around in the corporate sector too. And one that is more and more being pushed towards tools like Mapreduce and Hadoop. Crunching large volumes of data used to be a task that UNIX was designed and built for, but now we’re seeing a corporate acceptance of Linux based technologies capable of high volume number crunching. Ironically it used to be that UNIX was for the big end of town, and Linux for the medium sized commodity tasks.
Speaking in terms of what my current job in the creative industries is like, Linux definitely rules the day. SGI used to own the creative industry when it came to UNIX, but Linux represented a cheaper alternative with a faster turn around on useful changes. Linux is now the de-facto operating system in large creative visual effects and 3D studios around the globe. Studios like ILM, Weta, Dreamworks, Rising Sun Pictures and Animal Logic are all using it extensively.
Chris: I know you’re quite an advocate for free and open-source software, which Linux distribution(s) do you recommend and what distribution do you personally use at home and why?
Dan: For traditional large enterprise server workloads, RHEL is typically what I prefer to recommend to large corporates. It’s got solid vendor support from a company that has recently turned itself into a billion dollar per year business, with support personnel in offices all around the globe. Additionally when it comes to hiring staff, the various RedHat certifications mean that either judging a potential candidates skill base or training existing staff is a relatively easy thing to do.
Where more flexibility is required, I prefer Ubuntu LTS for both server and workstation workloads. The Debian heritage coupled with Canonical’s push to make Linux easy to use on the Desktop mean it’s well suited for a wide range of uses. Commercial support is available if needed. And the enormous userbase means that there’s great community support, and it’s gaining a lot of focus from commercial software vendors as well. Ubuntu offers substantially more software from its internal repositories without needing to find third party sources. It is also more up to date than RHEL (only just last week I met with frustration over RHEL6.3’s 16TB ext4fs limitation due to very outdated e2fsprogs packages), and the LTS offerings are quite sensible for business IT lifecycles.
At home my whole family keeps up to date with the most recent Ubuntu releases. After too much frustration with Microsoft products and with the increasing maturity of Ubuntu, they’ve all converted in recent years.
Chris: And basically just extend your thoughts on the aforementioned topic of HP moving away from HP-UX and towards x86-64 Linux.
Dan: HP have faced a lot of challenges with HP-UX. As mentioned before, the cost of the platform is quite high, and they’ve faced enormous challenges with Itanium’s turn around on features. Every time HP announce a new feature in Itanium (symmetric multi-threading, for instance), it quickly appears in commodity x86 hardware and HP’s competitive advantage is lost once again. Add to that the low clock speed of Itanium, which I know has bitten me personally when it’s come to very serial enterprise workloads that were CPU bound.
On the software side of things, HP-UX continues to integrate more and more open source software into it’s default release. I was quite surprised when I saw 11.31 bundle RedHat Directory Server as an LDAP implementation.
I think HP are trying to be very careful. Oracle challenged them already on their future support of Itanium and HP-UX, which rattled HP’s customer base who relied on both technologies. While HP needs to continue to support their existing customer base, it’s clear the demand for Linux on x86 is enormous, and HP want to bring some of their decades of HP-UX software smarts across to their new Linux customers. I think there definitely will be an inevitable decline in UNIX (particularly HP-UX) use, and a growth in the Linux market.
I believe there will always be large legacy customers that hold on to HP-UX, but new and growing businesses with no legacy requirements would find the cost justification of a UNIX platform a very difficult one, particularly in today’s market.
I really want to thank Dan for taking the time to chat with me for Unixmen. Dan is a busy man and I conside myself lucky that he shared his thoughts for Unixmen.
