August 12, 2009

Will Commercial Linux Vendors Benefit Monetarily from the Software/Virtual Appliance Business?

Independent software vendors (ISVs) and users stand to gain immeasurably from software appliances/virtual appliances (SWAs/VAs). But what about Linux operating system (OS) vendors, such as Novell and Red Hat, whose embedded variations of their general purpose OSs, SUSE Linux Enterprise Server (SLES) and Red Hat Enterprise Linux (RHEL), respectively, are powering many of the appliances?

Based on forecasts for the revenue generated by SWAs, it appears that Novell and Red Hat, whose embedded versions of their general purpose OSs will be powering many of the SWAs, will benefit only marginally for the next few years. Any revenue that their OSs gains from powering SWAs/VAs will be a small fraction (low single digit percentages) of the revenue they gain from their general-purpose RHEL and SLES subscriptions.

ISVs receive considerable value from SWAs. For the ISV, SWAs are another route to market for their applications. Prior to SWAs, getting a new application (or existing application on a new platform) up and running required cooperation from the OS vendor, the ISV, and the user. This can be a daunting experience, since performance is almost always a part of the consideration.

ISVs generally receive the following value from SWAs:

  • Relief in the form of fewer platforms on which to deliver applications
  • Reduction in inconsistency problems across multiple platforms
  • Less costly testing across multiple platforms
  • Reduction, possibly elimination, of patching and updating across multiple platforms
  • Control of support for customers
  • Simpler application development--developers can focus on adding value to applications versus addressing multiple platform differences for each supported platform
  • Control of performance and reliability issues--ISV configures what works best for each application; fit to requirements of application

If ISVs begin supporting multiple embedded OSs per application, then some of the advantages listed above begin to disappear.

Users/customers are also the beneficiaries of considerable value from SWAs:

  • Support and maintenance comes from one source, the ISV, generally eliminating the need to manage multiple maintenance streams, licenses, and service contracts (for application and operating system)
  • Components are embedded versus exposed, enhancing security
  • Problems are isolated and resolved for an application and underlying systems before the user gets the application
  • Security and defect patches are collected and integrated by the ISV for all included components
  • SWAs may be less expensive
  • Purchasing or booting a full general purpose OS is not required
  • Bare, standardized hardware can be used
  • Multiple virtual infrastructures can run VAs with little or no effort from the user

Linux OS vendors that partake in SWAs/VAs will also benefit, but their benefits may be somewhat limited for the near future, especially revenue benefits:

  • OS vendor is not first in line to provide maintenance and support
  • OS vendor leaves it to ISV to create satisfactory performance for the application
  • OS vendors can sell application delivery tools, such as rPath, to ISV to generate revenue or receive subscription revenue from the ISV for its embedded OS

Linux vendors that have variations of their operating system embedded in SWAs/VAs can potentially increase their OSs’ market share and generate additional revenue if they do not cannibalize their general purpose OS business. Two questions arise:

  • What level of subscription revenue can Linux OS vendors expect from SWAs?
  • How does the embedded OS subscription revenue compare to the subscription revenue that they get from full-blown general purpose OSs?

OS vendors can receive a small percent of each SWA sale, an embedded OS subscription, as compensation from the ISV for use of a variation of their OS, often referred to as a JeOS (Just enough OS). According to Billy Marshall, former CEO of rPath, and a Novell spokesperson, the subscription for an embedded OS would be a single digit percent, such as two to seven percent, of the ISV's selling price of the SWA/VA. We examine how much this embedded OS subscription revenue could be for commercial Linux vendors, Novell and Red Hat.

To compare potential subscription revenue for embedded OSs and general-purpose OSs, that can also run the same applications as those contained in SWAs, we do some simple arithmetic.

IDC has forecast that the SWA/VA market with paid Linux as the embedded operating system will be over $40 million in 2009 with a forecast of almost $300 million for 2012. SWAs powered by non-paid Linux distributions have revenue numbers similar to those of paid Linux. Novell and Red Hat account for over 90 percent of the paid Linux subscription revenue. To simplify the arithmetic below, let's assume that SLES and RHEL own 100% of the paid Linux market and evenly share the paid Linux subscription revenue for embedded versions of SLES and RHEL in SWAs/VAs.

For 2009, the total paid Linux subscription revenue for SLES and RHEL embedded in SWAs/VAs, at the rate of seven percent of the sale of each SWA/VA, would be just over $2.86 million (0.07 times $40+ million). For 2012, it would be just over $20.3 million.

Novell’s Linux-based subscription revenue for FY2008 (ended October 31, 2008) was $120 million. Since Red Hat acquired JBoss in April 2006, its subscription revenue numbers include both RHEL-based subscription revenue and JBoss subscription revenue. We went back to FY2006 (ended February 28) before Red Hat acquired JBoss to get a RHEL-based subscription revenue number, $230 million. We assumed that RHEL-based subscription revenue increased only 20 percent per year through FY2009 (this is a conservative estimate because Red Hat had subscription revenue increases following FY2006 of 48 percent, 32 percent, and 20 percent, respectively).

We calculate that in FY2009, Red Hat had $390.2 million of RHEL-based revenue from a total subscription revenue (including JBoss subscription revenue) of $541.2 million. We know that the $390.2 million RHEL-based revenue for FY2009 is low, but since Red Hat does not share RHEL-based subscription revenue, we will use the computed number.

We forecast the general-purpose OS revenue, assuming no cannibalization as explained earlier, for each company in 2012 by assuming a rate of growth of 20 percent and then 10 percent. Given the current economic conditions, the 20 percent figure may prove to be too high, but Novell's FY2008 year over year growth rate for SLES was 38 percent. The projected Novell Linux subscription revenue for 2009 and 2012 is $144 million and $248.9 million, respectively. For Red Hat, it is $390.2 million for 2009 and $809.2 million for the end of 2012.

Based on the above forecast revenue numbers for Novell, the subscription revenue for embedded SLES (in SWAs/VAs) is only one percent of the $144 million total general purpose SLES-based subscription revenue for 2009, and only 4.1 percent for 2012. The corresponding percentages for RHEL are 0.37 percent in 2009 and 1.25 percent in 2012.

When we drop the rate of growth of general purpose RHEL-based and SLES-based subscription revenue from 20 percent per year to 10 percent, the subscription revenue for embedded SLES is 5.8 percent of the general purpose SLES-based subscription revenue for 2012 and 1.8 percent for RHEL for 2012.

This forecasted revenue by IDC for SWAs/VAs could be low, but even if it is higher, the embedded OS revenue (for SWA/VA) is still meager at best for both Novell and Red Hat over the next three to four years. This is one reason that Red Hat has shown little interest in its Red Hat Appliance Operating System (RHAOS) since its announcement over a year ago. And it is the reason that Microsoft so far has shown little interest in SWAs/VAs.

We believe that Microsoft and Red Hat’s interests in SWAs will increase when the SWA/VA market takes off and when customers begin asking for Windows and RHEL as the embedded OSs. The Linux OS vendor that dominates the embedded OS market for SWAs/VAs will be able to generate significant revenue over time.

Will it be Novell or Red Hat?