Linux: Ready for Its First Service?

Linux is on a five-year maturity model (currently at mid-Year 3). This compares with Unix, which took more than 10 years to mature. A key driver that helps shorten the Linux maturity cycle is leveraging Unix services and support capabilities. Linux support services are becoming an area of vendor differentiation in an increasingly commoditized Intel world.

META Trend: With distributed n-tier (DBMS, application, Web) server architectures standardizing on Intel, proprietary Unix (Solaris, HP-UX, AIX) will recede to high-end, low-unit-volume, legacy-platform status by 2005/06, displaced by OSs designed for Intel economics: Windows and Linux. Linux will rapidly mature and gain momentum as an ISV reference platform, moving beyond high-volume Web, technical computing, and appliance server environments into mainstream application and DBMS server roles by 2004/05. Linux server growth will initially be at the expense of Unix (2003/04), but will eventually vie for dominance with Windows (2005/06).

Through 2008, Linux service differentiation will focus on solutions that are based on proven referenceable and repeatable offerings. These must be consistent in multiple channels and geographies. IT organizations (ITOs) should question Linux server vendors’ abilty to support complex projects, including layered products and services. Vendor differentiation will be down to offerings that include a mixture of commercial, mainstream, and open source layered software. ITOs must focus services for Linux and open source in the areas of high availability (HA) clustering, database (DBMS), and OS and Unix/Windows migrations.

We have updated our Linux server scorecard for 2H03 (see Figure 1). This scorecard can be compared to 1H03. With this continuing maturity, ITOs need to scrutinize individual server vendors for their Linux service capabilities with several criteria in mind.

Basic Support Services. Such services are the foundation a vendor must supply, focusing on first-level support and client contact around the time of Linux OS installation. These can be fulfilled either directly by the vendor or indirectly through a channel partner. Adaptive resource management tools can help the Linux installation as well as hardware platform and management. Further services include the following:

• Installation support: Many technical support issues are initiated at the time of Linux installation. These include hardware driver, configuration, and setup issues. The resolution may involve escalation within the hardware provider or Red Hat/SUSE, often delaying problem resolution.
• First- and second-level support: Installation then extends to full technical support. This must leverage first-level (contact/call center) and second-level support (escalations and proofing). Here, the engineering relationship should be proven for third-level support between the Linux platform vendor and the Linux distribution company (e.g., Red Hat, Novell/SUSE).
• Regional coverage: Coverage must be consistent/localized in a native language and time zone (for contact and availability). Premium support packages may be available to increase local support contracts and content.

Core Support Services. Pre- and post-Linux installation services are key to the Linux platform vendor’s assessment. Some of these services require planning and the involvement of a vendor’s professional services organization or an associated third party (e.g., integrator/reseller). Areas of core support services include the following:

• Unix migration: Server vendors with legacy Unix offerings (e.g., Sun-Solaris, IBM-AIX, HP-UX) must leverage their core services onto Linux platforms and offerings. This leverage sets a strong foundation for moving layered service software (e.g., Oracle, BEA, IBM) to the Linux platform. This benefits services for other open source software (e.g., JBoss App Server, MySQL DBMS). App software can also be considered to move from Unix to Linux.
• Windows migration: Moving from Windows to Linux is more of an issue than Unix to Linux, because there will be some migration from Microsoft software to an alternative commercial or open source product. Areas of concern for migration will include directory, application server, DBMS server, and systems management. The Linux service vendor must provide these skills across the entire stack on top of Windows, not just the OS itself.
• Lights-out support: To improve Lintel server sysadmin ratios and reduce remote-site costs, lights-out support and remote administration are effective extensions to adaptive resource management tools. This will include hardware as well as Linux OS alerts.
• Education services: Server sysadmin ratios are key cost metrics in total-costof-ownership and ROI calculations. Moving to Linux from Unix or adding servers may require additional training from the hardware vendor or third-party training organizations.

Layered Support Services. As a measure of maturity, the selection of Linux service vendors is not just down to basic or core services but also additional services of related layered software. This will include third-party software services (e.g., Oracle, BEA, SAP, PeopleSoft). However, it also includes additional services from the vendors for Linux or open source software products that cover the following:

• DBMS services: Linux has started on the path of scalability for DBMS services. Currently, Linux DBMS performance is attainable on two to four processors maximum, due to I/O limitations in the 2.4 kernel. We expect this to double to eight processors with the 2.6 kernel (introduction in 2004). Until then, any serious Linux DBMS deployments will be either small instances or leading-edge scale-out clustering (e.g., Oracle9i RAC, IBM DB2 ICE). This scale-out and scale-up model requires increased services from the Linux platform vendor.
• HA services: High availability and the use of Linux clustering are becoming prevalent in the use of database and application failover. This, in turn, helps mature Linux as a DBMS platform. ITOs need to assess the vendor’s capabilities in supporting clustering solutions and services for products such as Oracle9i RAC, IBM DB2 ICE, and Veritas.
• Open source services: Open source service software is behind Linux in maturity. Currently, migration to Linux with commercial software is fueling maturity and momentum. By 2007, open source software will be a viable choice for all software services (e.g., DBMS, directory, clustering, app server). This will demand more service resources from Linux platform providers.

Having understood these relevant metrics, an ITO can then assess the capabilities of Linux services vendors (see Figure 2). We position the vendors for Linux services as follows:

• IBM: IBM is a leader across all Linux basic, core, and layered services. By 2005, as Linux layered services mature, we expect IBM to be able to maintain this strong Linux position and work with Novell to promote SUSE and layered products as an alternative to both Red Hat and Microsoft.
• HP: HP is strong across all Linux services but slightly weakened (compared to IBM) by the lack of some layered and core Linux services - largely due to its focus on Microsoft. As Itanium gains volume with both HP-UX and Linux, we believe HP services will cross-pollinate and help fill the gaps versus IBM.
• Dell: Dell’s model bodes well for price/performance and first-level support for the Linux platform but less with complex projects. Dell is strengthening its relationship with the Linux players (e.g., Oracle, Red Hat, Novell/SUSE), which help plug its Linux services shortfalls across the board.
• Unisys: Unisys is a good services company for high-end Wintel platforms and offerings. Unisys needs to leverage these skills to sell Linux on its high-end platforms. This, in turn, helps strengthen Linux database deployments as an alternative to Unix RISC.
• Fujitsu: Fujitsu has the opportunity to compete with a service model similar to that of HP and do so in strong geographies such as Germany, the UK, and US. However, elsewhere Fujitsu must invest locally in its Linux service capabilities or leverage an integrator/channel partner.
• Sun: Sun’s Linux services take strength from a strong Solaris services base. However, its Lintel platform position is still unfolding. This is a good step in Sun’s solution/services recovery, but Sun needs to leverage layered services and software on more heterogeneous multivendor platforms.

ITOs need to track both Red Hat and Novell/SUSE, which currently depend on hardware OEMs for Linux services. As Linux matures, we expect Novell and Red Hat to invest in service capabilities in a layered approach similar to that of Microsoft and Windows.

Business Impact: Common Linux and Unix services across a portfolio diminish costs by reducing variety and risk.

Bottom Line: Linux services differentiation is a balance of basic, core, and layered services as well as relationships with Red Hat and Novell/SUSE. ITOs should examine Linux service gaps across the portfolio and fill in accordingly.

META Group originally published this article on 7 January 2004.