SUSE Linux Enterprise Server 10 – Your Linux is Ready
July 24th, 2006 by Jeff Jaffe
Servers and desktops
On April 3, I launched the Novell CTO blog with an enthusiastic discussion of the rapid technological progress made by the Linux desktop. The desktop is one breakthrough. Beyond that, our SUSE Linux Enterprise 10 launch is about a common code base for desktop and server. The near term market interest remains around the server. And thus a key focus is to take the server to a new spot – the data center.
So why the earlier focus on the desktop?
OK. I set myself up. If the action is on the server – why all the attention to the desktop. Here are three key reasons:
- Disruption. Linux on the server is well established. Linux on the desktop is incipient. From a technology viewpoint, the disruption is more powerful on the desktop.
- Top of mind. The desktop attracts attention. Pretty simple. It is what everyone sees every day. We use it. It is visual. The psychological effect of widespread Linux usage on the desktop will become the most powerful accelerant in the move to Open Source technology.
- It solves the most strategic need. The most strategic need facing corporate I/T managers is lack of choice and vendor lock-in. The Linux desktop addresses this head-on.
Many people have spoken to me about my focus on the desktop. One group tells me I am too modest. How dare I call the Linux desktop “good enough”? Boast even more about the innovative graphics and search features! Look at all the incredible press reviews that say it is better than Vista.
Others say that I’m too optimistic. Don’t I realize that Microsoft is too powerful on the desktop? A waste of time.
Some of Novell’s most loyal NetWare customers also bemoan that we are not doing enough for them. I appreciate their concerns. We are addressing their needs with offerings such as the Open Workgroup Suite (check out my blog postings from June 26 and July 10).
So I’ve received my share of criticism from all sides. I’m comfortable with my middle position: significant enterprise Linux desktop pilots in 2006, followed by mass deployments in 2007.
Servers
But the action is on the server. Where Linux has had the most impact to date. Where mission critical applications take place. Testing and certifications are key. Quality and reliability must be fantastic. So how have we been doing with our Linux servers?
In truth, despite some impressive progress, Linux has a long way to go. Here are some data points. According to IDC, in 2005, Linux operating environment revenue was less than 3% of total industry operating environment revenue.
Growth rates have been modest. True, the projected growth rate of Linux (24%) is very impressive in a market that overall has much smaller growth. But none of us would call 24% “hypergrowth”. How many new technology areas have experienced growth rates in the 30-50% range. Especially when Linux has huge cost advantages in a market where it only has 3% share. Let’s face it. We have more work to do!
What is going on? What is missing?
Server workloads are very demanding. This is true about all servers. But it is particularly true as you get to the data center. That is where mission critical applications are housed. That’s where customers bet their businesses. The fact of the matter is that critical characteristics of server operating systems have not yet matured.
This is not to say that Linux does not exist in data centers. There are I/T shops with the maturity and skills to take a decent Linux solution, surround with enough people, and make it industrial strength. But have we moved the vast majority of mission critical applications? Just think about the 3% figure.
Characteristics of data center servers
Let’s look at some of those characteristics.
- Testing, reliability and high availability. Data center servers simply cannot fail. A complete test strategy is a critical piece of ensuring high availability.
- Virtualization support. To support the consolidation of workloads on fewer servers, virtualization support is required.
- Scale and clustering. Mission critical applications need to scale to very large workloads. Companies cannot rewrite their applications as they grow. The underlying system must allow growth. A specific piece of scaling is to be able to run on a variety of clustered system architectures: including SMP (symmetric multi-processor), NUMA (non-uniform memory access), and loosely-coupled systems.
- Performance. Many applications have demanding response time needs. A general purpose operating system must have overhead to deal with the wide range of applications environments; but it still must be tunable to provide immediate turnaround when needed.
- Security. With many threat models challenging the security of I/T installations, security mechanisms are required to ensure that nothing can damage mission critical applications.
- Manageability. Sometimes problems do occur. A rich set of tools is required so that the data center can be monitored. I/T managers must have their fingers on the pulse of the servers and must be able to fix problems as they occur.
- Tuned to a variety of underlying architectures and integration. To allow maximum flexibility and portability, it is highly desirable that the operating system work across multiple processor architectures. Testing with these hardware architectures is needed to provide a better out-of-the-box experience.
Following the UNIX adoption curve
As we move Linux to the Data Center, we have the benefit of a clear roadmap. All of these key features were introduced over time in the UNIX world and became the obvious set of requirements to be addressed. We have the opportunity to address them much faster than the UNIX pace. We don’t need to follow blind alleys – we only need to enhance Linux to be at the level of other UNIX systems. We have the power of the community to address these features.
Of course, we are doing more than “catching up”. We are also taking it to the next level. Our approaches to security, virtualization, and manageability – among others – are fundamentally new. We are advantaged not only by the speed of the Open Source community – but also by its innovation.
What are the examples? What are the proof points? I’ll discuss these in the next posting.