While many have adopted Linux deep into the data center for mission critical IT infrastructure functions and for mission critical applications and data, there are still many watching and waiting. This article focuses on the technical, market, and collaboration readiness of Linux in a competitive world to assist the reader on the timing and location of Linux-based software stacks in the data center.
What does hardening of an OS mean?
When one thinks about the hardening of an Operating System for Data Center mission-critical applications, the goal or comparison is made to a mainframe (a.k.a. mainframe disciplines). A mainframe typically is a scale-up hardware and software stack focused on a market of hundreds/thousands of servers (i.e., RISC and Itanium Servers fit this target). Other types of servers tend to focus on a market of thousands/millions of servers (i.e., Xeon and Opteron servers fit this target) in a scale-out style. The business models of ISVs can be quite different when looking at these two markets which can effect deployment techniques (e.g., the popular one application/one server has deviated from the concept of mainframe disciplines for some years).
A mainframe embodies several values:
- Reliability, Availability, Serviceability and Security (RASS) with availability often being the most sought-after priority
- Quality (both intrinsic reliability and the follow-up testing)
- Performance – Price/Performance (e.g., excellent $/tps in consolidated scale-up applications; excellent TCO over life of asset including such things as power consumption, maintenance, etc.)
- Support: high customer satisfaction is expected in that problems are well understood by the vendor and handled quickly with a sense of urgency.
Since “hardening” is more than the kernel, this subject must eventually include thoughts higher up in the software stack. One must fully understand what .orgs and/or .coms are doing to fill hardening gaps in the Linux ecosystem software stack. The positive influence of foundations, alliances and consortiums such as The Linux Foundation, Green Grid, OW2, Open Solutions Alliance, Apache Software Foundation, Eclipse Foundation and many others must also be considered. Server and Storage Virtualizaton has long been a valued feature of Mainframes to keep purchase and operational costs to a manageable level. Such features have come to the Linux world and will continue to mature. This supports the “greening” of data centers for economic viability, environmental impact, and social responsibility.
What is the current state of Linux?
Most agree that the future for server operating systems is a duopoly made up of Windows and Linux. This is particularly true for new application deployments by end-users, ISVs, IHVs, etc. This is less true for legacy applications.
The innovators and early adopters of Linux servers for enterprise applications were drawn to Linux (e.g., in 2002) by using much less expensive hardware (x86 then x64 style from Intel and AMD) as compared to Unix/proprietary RISC hardware.
Many ISV legacy solutions based on UNIX have been recently ported to Linux. As is often the case with first time offers, some ISVs have done the minimum to get a Linux version into their sales catalogs in an attempt to catchup with the market popularity of Linux. As market demand has grown and time has passed, these applications have or are in the process of being optimized for the Linux-based software stack.
Server OEM’s tend to offer server platforms with Unix, Windows, and/or Linux with the marketing message of “choice” for buyers to consider. Their investment in Linux R&D has grown over the last seven years raising the functional and hardened waterline for all.
Starting in 2004, the Linux-based software stack exploded with OSS-based solutions (from middle to top) coupled with the already mature set of proprietary offers from ISVs who now play in the Linux market. This provides a lot of choice for buyers interested in servers supporting the Linux ecosystem. Making the right choice in selecting software stack components becomes important. Bull Services has been successful in providing consulting services to assist customers in making choices appropriate to their needs. In reality, most users will deploy a hybrid software stack in the Linux market (combination of Proprietary and OSS). More and more alliances such as http://www.opensolutionsalliance.org/ will appear in an attempt to emphasize integrated software solution stacks making the buying decision easier.
Using the Technology Adoption Life Cycle picture (see Figures 1 and 2), Linux adoption in the Data Center has crossed the “chasm” and is deep into the early majority segment. Accelerated adoption will occur in 2008 when the “late majority” start adopting in larger numbers. A key reason for this acceleration is the improved awareness by this segment of the market through successful messages and case studies from key ISVs (e.g., Oracle, SAP, etc.), OSS-based businesses and communities (PostgreSQL, EnterpriseDB, MySQL, SugarCRM, etc.), public policy, hardware OEMs and semiconductor companies (see www.redhatonintel.com for an interesting example of increasing market awareness and size of the Linux ecosystem through collaboration). The growing trend of SaaS (Software as a Service) results in extreme data centers that tend to be Linux-based (e.g., Google) which accelerates hardening efforts. There are many examples of where Linux-based servers have moved deeper into data center beyond web servers and application servers. The strong technical, marketing, and support push from RDBMS, Business Intelligence and ERP vendors, has resulted in Linux being adopted at the heart of many enterprise data centers. Based upon the appropriate Linux-based software stacks, there is no data center workload that Linux can not do or has not been deployed with. It has crossed the chasm.
 |
 |
Figure 1 |
Figure 2 |
Linux has become a magnet for innovation which will drive it to “good-enough” parity to Unix and threatening to surpass innovation in Windows from 2008 through 2010. Figure 3 reflects the maturing of Operating Systems from 2004 through 2010.
Figure 3
What is the contribution of The Linux Foundation to this topic?
Figure 4 highlights where The Linux Foundation (TLF) came from and when.
Figure 4
TLF has fostered collaboration through leaders coming from the eco-system (Vendors, Developers, and Users). During 2007 there has been a significant improvement in communication per the objective highlighted in Figure 5.
Figure 5
To stay on top of The Linux Foundation activities, see their website at http://www.linux-foundation.org/en/Main_Page.
A common criticism from those that compete against the Linux eco-system is that there is no roadmap. The Linux User Advisory Council from the old OSDL days had made a similar criticism. The OSDL/DCL Steering Committee took this under advisement back in 2005. The analogy of a “weather forecast” was proposed since that is a less committal term than roadmap. The Linux Technical Advisory Board has taken over ownership of the weather forecast and it can be found at http://www.linux-foundation.org/en/Linux_Platform_Weather_Forecast. Jonathan Corbet gave an excellent presentation on the weather forecast during the June 2007 Linux Collaboratio Summit which can be found at http://lwn.net/talks/lfcs2007/. As you will see from this presentation there is a lot of development work that is planned for submittal in the near future. There is good reason for the term “roadmap” being avoided. The historical term roadmap as used by IT suppliers implies commitment even though the predictability has often been inaccurate. Just imagine making commitments on an Operating System kernel (e.g., 2.6.22 for July 2007) that since June 2006 and kernel 2.6.17, has accepted patches from 2100 developers for a total of 30,100 changesets, and over 2 million lines of code changed with the largest contributor providing just 2.3% of the changesets. Also consider the work coming from a variety of organizations most of which are large recognizable international companies (see Figure 6 - Bull visibily contributes code and test efforts directly to the communities and also through IBM as part of a Cooperation Framework between the two companies). One might wonder how proprietary OS vendors can compete over the long-term with the breath, depth, passion and quality of the Linux development community.
Figure 6
The weather forecast is therefore an attempt to track ongoing developments in the Linux development community that have a good chance of appearing in a mainline kernel and/or major distros sometime in the near future. The "chief meteorologist" is Jonathan Corbet, Executive Editor at LWN.net. The kernel release program was changed in 2005 to be fast with new kernels every 2 to 3 months that are predictable (expect 2.6.23 around October, 2007). This faster release cycle is portrayed in Figure 7.
Figure 7
With priority input from the Vendor Advisory Council, User Advisory Council and Technical Advisory Board, development work is more closely in tune with market needs therefore accelerating adoption. Bull is an active participant in the Vendor Advisory Coucil. For highlights of TLF’s first ever Collaboration Summit see http://linux-foundation.org/weblogs/
Conclusion
The question for CIOs and CTOs (users and suppliers) in 2007 and 2008 is changing from “should we be using Linux-based software stacks” to “when and where should we use Linux-based software stacks?” As a pioneer contributor, integrator and infrastructure solution provider, Bull is there for you as an architect of an open world. Enjoy the liberty you get with the Linux ecosystem!
|
