The Register dances the Tarantella
Interview with Mike Orr
If there were an architect for the SCO deal with Caldera, it would be Mike Orr, president of Tarantella. The Register has been talking to him.
It was his idea for SCO to visit Linux distributors last October, because "anything that increases the credibility of Unix would be good for SCO". Caldera responded warmly, and the result was first the professional services deal and then the bundling of Tarantella.
The next step saw SCO CEO Doug Michels and Caldera CEO Ransom Love get together to see if something closer was possible. It looked good, so bankers were then involved to structure a relationship, with the final shape of the deal being influenced by their advice on the optimisation of the financial aspects.
As for Caldera, Orr agreed that it still has a brand recognition problem, but pointed out that Caldera does more Linux business than Red Hat because it is diversified into eight sectors. Ironically UnixWare is about to go back to Utah to a company fostered by Ray Noorda, who had bought it in the first place for Novell.
SCO had a plan to launch its own Linux which was only pulled at the last minute. Because of the downturn in the Unix business, SCO was starting to get worried about its cash position, so with Caldera having its IPO money, it was logical to work with Caldera.
The Unix slowdown was at first attributed to a Y2k downturn, but it turns out that there is still a general slowdown in the reseller channel, and it is not confined to Unix - Novell and Citrix had also done less business, Orr said.
Even after SCO did a survey of resellers, it was hard to see a pattern except for a slowdown compared with previous years. There was no evidence from resellers that SCO was losing a lot of business to Linux. Orr thought Windows 2000 had had some effect in small rather than large businesses, which he was expecting because of scepticism about W2k in enterprises, but he was surprised that business had been cut in half.
Sun was doing well, but this did not eat into the traditional SCO business, which was mostly one-off vertical business, and nor was SCO competing with IBM or HP, Orr said. There were some signs of business returning, but it was too soon to be sure, although it was not still declining. Orr said he would have to await the feedback since SCO doesn't book a sale when it ships to distributors.
With $2.5 million of business in the last quarter, a loss of about $8 million (which Orr said wryly was better than he had expected), as well as a share price that could only go up to a guaranteed $5 from its present $3.50 as a result of the Caldera deal, Orr explained how he was going to accelerate out of this hole.
He has taken on 60 new sales people (something that should have been done years ago, of course), so it would take some time before they close any substantial business. He didn't expect profitability before 2002 because of this investment and the fact that a sales campaign could easily take 6-9 months.
Tarantella was more like a start-up, and his focus was first on the top line, with some success because five weeks into the present quarter, Tarantella has already passed the sales from the whole of the previous quarter. This quarter, SCO would be reporting its results as before, but after the October completion, the whole business model would change dramatically.
Originally marketing had been focused on large corporations, but the sales cycles were long and he needed direct sales activity until the resellers got up to speed. The business opportunity was very much with ASPs.
Tarantella was now getting five or six new customers a week, with some testing smaller systems before introducing larger systems. Telecom Italia has about 5000 users.
On the OEM front, Hitachi was creating an appliance, with three other appliance vendors evaluating and testing Tarantella. This was something that Citrix could not do as they would have to install it on the application server itself, whereas Tarantella doesn't have to touch the applications or the operating system
SCO has a three-tier model in which a separate application server sits between the servers that run the applications and the clients that access the apps. The protocol for the server and clients is SCO's Adaptive Internet Protocol (AIP), with the app server using X11 to link to Unix or RDP for NT. Tarantella's equivalent of a directory service is specific and doesn't link to LDAP servers, Novell's NetWare, or Microsoft's Active Directory.
Tarantella is still a distant number two to Citrix, which has in MetaFrame a two-tier model with clients connecting to either Unix or Windows on separate servers, with the Independent Computing Architecture (ICA) protocol. With ICA being clearly competitive to the RDP protocol, there must be a certain amount of frostiness in Citrix's relationship with Microsoft.
Citrix's model has less overhead than Tarantella's, although Tarantella reverse-engineered RDP, so it is in fact more friendly to the Microsoft approach. There have been talks about Tarantella licensing RDP on an official basis, and Microsoft seems willing to do this. Although it is still at the discussion level, Orr said an "amicable agreement" was likely.
Tarantella would of course like to get to the point where it's seen to be just as friendly to Microsoft as is Citrix - if not more so.
Orr sees its protocols as an important advantage over Citrix, together with scalability. The limitations for this are not yet known, although tests with Compaq to find the upper limit had passed the 2500 mark on a single system.
Cheaper than Citrix
The launch of next version of Tarantella will be in September, Orr said, and there would be a presentation about it at Forum 2000 next week.
Although the name was not yet announced, it could be thought of as being "Enterprise 3" with major new features and better performance. (Perhaps the graphics will be less jerky in this new version.) Tarantella is significantly less expensive than MetaFrame, with a lower-priced entry system, and presumably will stay less expensive with the new version.
Citrix is the only competitor in the same space, although its Unix version of MetaFrame is only available for Solaris. Citrix also seems to have slipped behind in the directory service area, and has no unification across Unix and Windows, except for anonymous users if this is enabled and security issues are not critical.
It should never be forgotten that Unix has always been multiuser, something that is anathema to Microsoft, and that only grudgingly did Microsoft license Citrix code for Windows Terminal Server. It is the market that seems to have decided to move back to what is effectively a mainframe model. Whether such a partial move would be the result of the complexity, unreliability, cost, security issues, networking problems with NT and W2K, the attraction of an ASP marketing model - or some combination of these - is hard to say, but the trend seems real enough.
The market seems poised to return to the mainframe days, and both Tarantella and Citrix should be well positioned to profit from the expanding market opportunity. ®
Sponsored: Benefits from the lessons learned in HPC