META Trend: Although Windows-based PCs will remain the dominant information access point for corporate and consumer users through 2006, multiple information consumption devices - tuned to (user-selected) specific lifestyle situations, work needs, and personal preferences - will become essential business tools. By 2005, pervasive computing/communications technologies will blur the line between corporate and consumer computing, ushering in an era of lifestyle computing.
Our work with G2000 corporate accounts indicates strong interest in, but strong aversion to, deployment of Linux for support of end-user operating systems (OSs), Web, mail and collaboration, business intelligence, and mobile applications. Although hobbyists and “religious” IT advocates are deploying Linux OSs and applications, most corporations are rejecting Linux as a corporate end-user platform. In practice, a license-free environment is a fallacy (see Figure 1), and giving Microsoft a “bloody nose” is a less compelling motivation than support, user acceptance, maintainability, total cost of ownership (TCO), and application compatibility concerns. For these reasons, we expect mainstream corporate penetration of Linux in end-user environments to grow extremely slowly. We estimate that corporate Linux deployment for general-purpose end-user platforms will grow from less than 0.1% today to 2% at best by YE05. Although we expect selected segments (e.g., education, government) to show higher deployment (growing from 2% currently to 10% by 2005 and 15%-20% by 2007), it is unlikely that deployments in these environments will seed consumption elsewhere due to different asset management behaviors, different political drivers, and a lower propensity to sacrifice familiarity and functionality (e.g., replacing known applications such as Office and Outlook with similar Linux-based analogs; see Figure 2 for a description of key corporate impediments to Linux deployment). Where we see Linux in our client base, it tends to be deployed primarily as a religious principle or as “under the radar” projects rather than being based on a systematic application, architecture, risk, and cost analysis. In assessing a Linux vs. Windows vs. alternatives decision, IT managers need to evaluate the following factors to determine which applications are suitable for deployment under Linux as well as the risk profile for making the decision.
End-User Platform OS Replacement
Users seeking to deploy Linux as an end-user platform OS should assume they will have to refresh the entire application stack. Although vendors like Lindows offer Windows end-user platform analogs, these are designed as lightweight Windows augmentation rather than as pure Windows replacements. Application emulators (via products such as Click-N-Run and CodeWeavers’ CrossOver Office) are another alternative providing basic support for “vanilla” versions of Office, Outlook, and Notes but should not be used for deployment of specialized application variants (e.g., those using extensive macros and scripting). Unless IT managers are prepared to sanction extensive product modification and application/hardware compatibility testing, they should focus on providing a Linux-based end-user platform OS augmented by native Linux office packages with non-standard applications delivered by Web, Windows Terminal Server/Citrix-based clients or carefully tested emulated versions.
Office
Linux-based office productivity tools such as StarOffice, ThinkFree, WordPerfect Office for Linux, and KOffice all include basic word processing, spreadsheet, and a presentation tool. However, no comparable products are universally available for Office extensions such as Outlook, Access, Project, Visio, FrontPage, Publisher, etc. We also expect Microsoft-Office-to-Linux-office productivity conversions to create serious productivity and cost ramifications comparable to those experienced in the infamous Office-95-to-97 migration. Linux office packages are therefore best suited for closed environments not requiring significant interchange of documents with MS Office users. In reality, most users will find Linux to be more expensive and difficult for office applications rather than less expensive (see Figure 3), leading to a decision to stay with the status quo.
Portals
We expect Linux to have minimal immediate corporate impact on portal application infrastructure. Although a few mainstream vendors (e.g., Computer Associates with CleverPath, IBM with WebSphere Application Server) have a Linux version, we are beginning to see these being used for consumer or supplier/partner-facing B2C and B2B applications rather than B2E applications. By 2005, we expect BEA to follow with Linux portal products. Companies looking at deploying Linux for portal applications should identify the key Web services (e.g., assembly, rendition, authentication, security) that are amenable to being offloaded to Linux appliances or defined function servers. Client devices will use a standard OS, a common language runtime, or a Java Virtual Machine (JVM) to deliver the server-based portal service. In practice, we expect most organizations deploying Linux-based devices for portal applications to use these for portal server augmentation rather than as a replacement for Windows or Unix server portal infrastructure.
Mail and Collaboration
IBM’s Domino excepted, no application vendor provides an established Linux-based collaboration suite. Even for Domino, our research indicates minimal adoption of this variant (less than 3%). Although we see a small group of collaboration products emerging (e.g., SuSE Linux e-mail Server, Jabber for instant messaging), we expect these products to have niche use because of existing investments and support/credibility issues (the same issues that beset Netscape’s collaboration strategy in the mid-1990s). Where we do expect to see deployment these installations will be in “religious” anti-Microsoft sites or in users focused on using pure Internet standards such as IMAP/POP3 mail. Rather than sanctioning a wholesale technology refresh (because most mail-system-to-mail-system migrations offer no net ROI in and of themselves), we recommend that most organizations focus on accruing benefits from server consolidation and improved operations. The only mainstream deployment of Linux we expect will be in provision of Message Transfer Agents (MTAs) and other relays (e.g., Qmail) to provide load balancing and other services. Organizations highly dependent on deep collaboration services (e.g., Web conferencing and instant messaging tied to core applications, workflow, etc.) should actively avoid Linux-based solutions because of viability and functionality concerns.
Business Intelligence
By YE04, we expect most business intelligence (BI) vendors to have mature Linux server products to support end-user analytical applications and reporting. Brio has Brio Reports (SQR) on Linux, and MicroStrategy has announced a Linux version. Although these versions will become mainstream by 2005, we expect Linux to become common for BI servers only if relational database management systems (RDBMSs) become common on Linux. This will enable users to build Linux data marts - and when that happens, BI report and online analytical processing (OLAP) servers will follow. However, we estimate that to be less than 10% of the BI server market by 2005. Although we expect IBM and Oracle to push Linux RDBMSs, we expect user adoption to be slow because of conservatism around reporting infrastructure refreshes. Although the immaturity of current offerings forces caution, in particular because of the lack of RDBMSs, we recommend that IT managers progressively specify Linux support as at least a highly desirable feature - especially where corporate BI requirements favor delegated analytical support (via decentralized data marts). We expect Linux client support to be minimal, with browsers, common language runtimes, and JVMs being used to emulate the fat-client experience.
Business Impact: Linux will have minimal impact on mainstream corporate end-user computing. Business managers should clearly understand the cost impact vs. complexity of deploying Linux-based infrastructure and applications before considering a move from incumbent providers.
Bottom Line: IT managers can exploit Linux for a few niche areas in the end-user computing environment (e.g., Web serving, mail relays). Other than these niches and specific verticals, Linux will have minimal corporate application.
META Group originally published this article on 22 April 2003
){kind=link}
){kind=link}
){kind=link}