Conversation with a successful Linux services entrepreneurFriday March 04, 2005 (12:00 PM GMT)
By: Robin 'Roblimo' Miller
SourceCon Zymaris runs Cybersource, an IT service company in Melbourne, Australia. Cybersource started as a one-man Unix shop in 1991 and has gradually evolved into a decent-sized Linux/FOSS-based business that serves a client base Zymaris says is now 20% government, 20% corporate, and 60% small/medium-sized businesses. (The interview was conducted via IRC and has been lightly edited for spelling, grammar, and continuity.)Roblimo: So tell me about Cybersource. More than 10 years in business, eh? Have you been there from the start? Are you the founder?
Con Zymaris: Yep, actually we started in early 1991. I quit my day job and we started the business.
Roblimo: By yourself or with partners?
Con Zymaris: We started with a group of four friends who had gone through University together. I quit my job first, then over the next 6-12 months as we established the business the others quit their jobs too.
Roblimo: So you started out providing Unix services? Windows? Both?
Con Zymaris: We started by concentrating on Unix for the most part.
A tiny bit of history to establish context:
We had _almost_ done this a couple of years earlier, in late 1989. One of the co-founders, David Keegel, and I shared a house at the time. He had been involved with connecting Melbourne University to the Internet and I'd just glimpsed this amazing new world through that connection. We wanted to start a business to bring this new world to the realm of business.
We were relatively unlucky because we didn't have a sizable chunk of capital to see us launched in a big way. However, we were relatively lucky that we curtailed expenses for the first few years. We grew organically, and we were reasonably profitable from the first 12 months. We really had no choice. We had to be. There was no additional capital or sugar daddy to fall back on.
Roblimo: How many people are involved now?
Con Zymaris: We have about 30 full time and part time people. We had ballooned to about 40 during 1999-2000 period, but we found that was unsustainable for the business model we had.
Roblimo: That business model was?
Con Zymaris: Our initial thought, of building a business around the Internet, proved untenable as we didn't have the necessary capital. We couldn't really afford real Unix boxes when we started. Sun kit was just too expensive at the time. So we got by on 386 Unixes and Minix. So we were _really_ keen when Torvalds popped his head up on comp.os.minix and talked about a platform which might become a successor to Minix.
As for business models, we decided to do a pure consulting/skills-for-hire model. We considered ourselves a boutique consultancy. Few other firms did serious Unix work, outside the main vendors themselves. We also wanted to remain fiercely independent, and not become a reseller or front end to Sun, HP, SGI or IBM.
Roblimo: When did you start adding Linux to your mix?
Con Zymaris: We followed its progress through 1992 and probably started then. I forget which distro it was. Possibly the MCC one. It came on (IIRC) 50 5.25" floppies. This was before Yggdrasil came out on CDROM.
By not having the capital to acquire proprietary Unix we were essentially forced to look at low cost and open alternatives, such as Linux and FreeBSD. We had all come from BSD backgrounds, too. Most of us had been using BSD 4.1 then 4.2 from the mid-80s. We were, however, realists and realised that few businesses could afford Unix workstations for their uses. We did a combination of software development (Motif, X11 as well as Windows client-side) and Unix backend. Our specialty was writing custom Unix backend servers to different front-end clients.
A couple of the founders had coding backgrounds (myself and Ramon Legnaghi) and David had a strong sysadmin background. The fourth partner, who later became my wife, was Jane Zymaris, who managed the books and ran the office.
Roblimo: Today, in 2005, what percentage of your business is Linux?
Con Zymaris: We have only a few people who continue to work on proprietary Unix, and 2-3 people who work on Windows integration. Everyone else now works in the Linux/FOSS space. I would guess 80%.
Roblimo: Wow! There's that much Linux-based training and consulting business in Melbourne? Do you do it all or are there others?
Con Zymaris: Over the years we've become known as 'the' place to go for Linux and open source. But there are actually quite a few others, and that's an excellent sign of the growth in the industry.
We consider ourselves to be a bit of an oddity in that we've been around for a while and we're a bit bigger than the usual five-person consulting startup in this space. But having dozens of other firms join in this space very much helps to legitimize the arena. We can point to over 200 such firms around the country now. We want to ensure that the industry as a whole floats upwards, and to that end we helped form both the local state industry cluster (Open Source Victoria), which has received state government funding, as well as the national industry body, Open Source Industry Australia.
We have players such as MySQL on board, but we have not yet made overtures to any of the 'names' from the Linux space (Red Hat, Novell, IBM, Sun etc.), although all have representatives who keep up on the OSIA mailing lists.
Roblimo: Your Linux vs. Windows TCO Comparison has gotten quite a bit of play. Has it helped bring in business?
Con Zymaris: It has helped in two ways. Firstly, it's taken our name out a little beyond the boundary that we used to be known in. It's also proven a useful tool in offering potential 'converts' a means by which they can determine Linux/FOSS value for themselves.
By that, I mean while there are other TCO studies around, none of them really offer a simple, join-the-dots method for interested organisations to insert their own numbers. They can stretch our TCO study to fit their reality, deleting stuff which doesn't make sense and adding stuff which does.
I've read through many other TCOs and many don't even provide a methodology of how they achieved the numbers. I think that ours helps us as a business sell migration to Linux to firms which are interested, but need solid evidence.
The TCO, as in many of these side projects we do, are undertaken because we feel there's a need for them. If another FOSS player had built one, we obviously wouldn't have bothered. Same goes for our GPL vs EULA licence comparison. We're not a legal firm, so not the best placed to do a licence analysis. However, no one else was taking up the baton, and we thought it important to raise the issue. So, we did enough of it to take around to some actual lawyers to get their input and sanity check.
I guess none of these side projects can be evidenced as providing actual business sales, but by growing the Linux/FOSS balloon, and getting our name out there, we believe that it all actually is worthwhile in a business sense.
Roblimo: How does your business break down? I mean, very roughly, what percentage is government, what percentage big business, what percentage smaller businesses?
Con Zymaris: Government 20%, Corporate 20%, SME 60%. Also, in the past few years we realised that in order to scale, we had to move beyond the standard consulting work that we had always done.
Roblimo: So you're seeing strong SME (Small and Medium-sized Enterprise) uptake, then?
Con Zymaris: Yes. Much of our Linux growth has been in the SME space, say sub-250 seats. That's actually one of the reasons we did the original TCO document in the first place; a client had asked us to justify our suggestion for them to migrate to Linux, and they were around that size.
Roblimo: Did that client migrate?
Con Zymaris: They ended up migrating many servers to Linux and many of the desktops to OpenOffice.org, but they haven't moved to Linux workstations.
We're still working on them though ;-)
Roblimo: Thinking of desktops... Getting any Linux desktop play?
Con Zymaris: We have actually migrated some smaller sites to complete Linux; backend, frontend. It took a few months, and a bit of care.
Roblimo: What's the biggest? Can you name a name?
Con Zymaris: I can probably give you one which would be OK with being named, but they're not that large in scale. A medical research group called the Centre for Molecular Biology and Medicine. They've been a long-time customer; we rolled out their first Linux server circa 1997 IIRC. So we wanted to make sure that the Linux desktops did what they wanted. They had all these specialist medical imaging apps, which we managed to find equivalents for. We also configured the desktops to look like the Win2k theme they were used to.
Roblimo: When did Molecular Biology and Medicine start running Linux on their desktops?
Con Zymaris: Over the past 6 months.
Roblimo: Are they happy? Do any of their people miss Windows?
Con Zymaris: I wasn't one of the people involved in the project directly, so my knowledge isn't absolute, but to my understanding there weren't many user hassles.
It does seem possible that by using the right methodology, many users in many organisations can be migrated to Linux desktops.
Roblimo: And that methodology is?
Con Zymaris: Pretty simple.
Analyse the existing terrain on the desktop. What are the dependencies? Roadblocks?
If there are roadblocks (VB apps, other Windows apps which can't be shown to work under WINE) then map out a migration strategy, which could be 3-5 years. You gotta start somewhere. In the meantime, cost the deployment of VMWare, Win4Lin or Citrix.
When you have the 'terrain map' of what the users must have, you can plan out a migration strategy which you can consider to be like stepping stones to get you across the wide river, to the other side, without getting wet.
That strategy can include simple processes like:
1) Switch to Windows versions of OpenOffice icons and apps in lieu of Microsoft Office.
2) Ensure that all corporate Office templates can work in both Office and OO.o
3) For users which must have Office, leave them with Office. But they get OO.o too, which they can use as PDF creator.
4) When the Office migration has settled, switch Windows versions of Firefox in lieu of IE; Leave the icon as the IE icon.
5) When this has settled down, introduce Citrix etc., and for those apps which can't run under Linux, add icons onto the Windows desktop pointing to those now 'remoted' apps.
6) After a few months, switch the underlying OS to Linux. Keep the same icons in the same positions as they were under the previous Windows OS.
Roblimo: "Remoted apps" are "remoted" as in running on a server?
Con Zymaris: Yes, those Windows apps which can't run under Linux, make them now run off Citrix. Get the Windows users used to running them remotely, even under Windows, so that when they run them remotely under Linux, there's little difference.
Do the final OS migration in stages across the organisation. There will be some users who cannot, for whatever reason, move off Windows. Don't sweat that. Try and break the mindset that the organisation can only be 'efficiently' managed from an IT perspective if it is totally homogenous. This fully plays into Microsoft's strategy, which is to lock organisations into its vertically-integrated software stack.
An important note: Make sure that you try and get the look and feel as close as your can to the version of Windows they had.
Roblimo: How many companies -- roughly -- have you moved at least partly to Linux desktops?
Con Zymaris: I'd have to ask the tech guys, but my guess would be no more than half-dozen firms having deployed Linux desktops in production. It's still early days yet, but the interest is there.
Our organisation has run on essentially Linux desktops since about 1997. After all these years, we still have one holdout: Our financial person's PC still runs Windows (the last remaining system which does so) because the accounting software our accountants asked us to use is Windows-only. We've even got a full, ready-to-go Linux PC running VMWare; running Windows; running that accounting software, but we've been too pre-occupied with work to get that intro production.
Roblimo: Your training offerings: Are they popular? Get many takers?
Con Zymaris: In honesty, we started our Linux training (in 1998 IIRC) because the firms we were deploying Linux servers to demanded it. In a sense, it was a means to an end. Many potential Linux migrators would state, "We can't seriously migrate our production systems to Linux unless our people are trained in the technology," which is fair enough.
So, to try and ensure that there should be few reasons for them not to migrate to Linux/FOSS, we started offering Linux training. It's only ever been a sideline to our business, just to keep clients happy. But it has some advantages in a sense. We do non-denominational training and target the Linux Standards Base as a generic platform. Also, people don't really come to us to get qualifications per se; they go to the LPI, Red Hat or Novell training groups for those. So what we get are techs in organisations who now need to work on Linux on a regular basis, needing to update their skills
Roblimo: Are there any questions I'm too stupid to ask, but should?
Con Zymaris: LOL
Roblimo: Serious. :)
Con Zymaris: Now, that's a dangerous question to ask an opinionated loudmouth like myself. ;)
Here's a few observations.
The industry side of Linux/FOSS really needs to come to the fore from here on in. The community side have done an excellent job to establish the technology, ethos and 'brand.' Many of us in the 'industry' side are also community advocates but advocacy's time, I think, has peaked. What we really need to focus on is building solid businesses around Linux/FOSS -- building and marketing successful products, building and marketing successful services.
The big guys can do it, sure. But Linux/FOSS offers many many other players a chance to grow and contribute.
Learn the language of business. Learn what business wants. Build it and sell it to them.
Learn how to compete for government business. If there are structural impediments in government procurement in your region, try and have them removed.
We need to polish up on our 'bridging' skills; those things that can allow us geeks and our wares to become a safe option to business.