President's Perspective

Semi-Counterpoint

Perhaps you have read certain newspapers and magazine articles where the editor poses a subject or issue and various opinions are rendered. In some of these, there is a counterpoint column with a contrasting view. This is not to suggest my article today is counter to the balance of this edition of TrendSetter.

My company develops solutions for a variety of platforms and operating environments. These solutions vary widely from solution stack to application environment, and from client to client. We have implemented more than a few UNIX based (primarily AIX) solutions over the years, so when Datatrend talks about Linux, some people may be surprised (some may even scratch their head since they just came out of a Datatrend AIX session). Regardless, people invariably get around a virtual campfire and engage in discussions contrasting UNIX and Linux.

Linux continues to gain momentum. It is viable in many areas. Having Linux as part of an overall information technology strategy is a likely, oftentimes advantageous, scenario, built upon solid logic and sound business principles; however, Linux is not to be construed as everything for everyone for every solution stack or application.

Linux is gaining so much momentum (for good reasons) so quickly, that some organizations may be making mistakes. Assessment is critical in determining what platforms to develop in or base applications upon and what corresponding infrastructure is best suited to the given application design. Some people are attempting to move to Linux where it may not be the best course to take.

The open source community is an interesting realm. On one hand, the explosion of open source tools, the sharing, the collaboration and the accessibility to the tools, scripts, work pieces and modeling conventions yield very positive results. Developers can design, develop, test and deploy applications efficiently, quickly, and at lower cost in many cases. It is goodness in these regards.

Linux continues to evolve with more and more functionality and more sophisticated security enablement. High performance computing is racing to Linux based upon the open source community's plethora of tools, shareware and the specific types of collaboration that yield the performance tuning tips needed, on the fly, during the development phase of a project.

However, there are some issues. Vertical scaling is one issue and it is a big issue. AIX, as an example, is awesome at vertical scaling. With AIX, you can exploit the POWER (System p5) architecture, which is particularly important as you move into the enterprise class configurations of 16- and 32-way (and beyond) server solutions. Linux, by comparison, is not well suited, today, to be applied to more than an 8-way configuration. This forces one to cluster a proliferation of many Intel servers, which can run counter to some IT managers' desires to consolidate. Blades are a nice packaging concept and are meant for certain types of workloads, but do not have the higher level systems management features of the enterprise class Intel servers. Linux on the mainframe can be deployed with the advantage of that platform's virtual machines capability, but at the highest server facilitation cost, and, most of the time, at substantially less performance than most Intel and POWER based solution architectures that would be configured as equivalently as possible.

It is difficult to compare apples to apples in viewing Linux on mainframe versus POWER versus Intel. However, from our experience, in most application scenarios, POWER runs Linux better than Intel. Truth is, POWER is a widely-regarded performance leader, far above anything else. POWER holds most of the world records in most application designs and at every price point. In most cases, a POWER implementation of Linux is a better approach than using the mainframe.

The mainframe culture still has a strong upper hand in the adoption battle, in terms of best practices, change control and virtualization. UNIX (AIX specifically) is catching up to the systems management features the mainframe has, and with good total solution packaging, and competes extremely well with those "big fat pipes" (Where is the mainframe in TPCC, SpecWeb and all the benchmark wars?).

I do not want to mislead. The given enterprise today should strongly consider having Linux as part of an overall strategy. Also, your Linux space is likely to grow. Again, it is about assessing the application, the business initiative being supported, your internal skills, what level of support you want, total cost of ownership, and whether server proliferation or consolidation come into play. And don't forget the RAS thing!

The open source community is a wonderful thing and more 'democratic' than the arena of UNIX vendors. However, with that democracy comes quite a few debates that lead to a creepy crawl pace in evolving Linux to fill some holes that AIX and other flavors of UNIX (but especially AIX!) plugged long ago (in the areas of scaling, management, and performance). One client told me last year, "Linux delivers what UNIX promised a long time ago" (speaking to application portability). Certainly, Linux is very portable. However, certain varieties of UNIX, AIX in particular, are not like RPG. Datatrend migrates customers from Solaris and UX to AIX all over the place ... UNIX to UNIX is not that tough. But by Linux standards, one can argue there is a significant delta in ease. I yield to this point, but I am in the migration business, so my pain is less than most. However, Linux is about as good as it gets in portability, and this is a strong consideration. When you do NOT need certain features (as described in the next article), and vertical scaling is not priority one, Linux could very well be your best choice.

Another consideration ... Linux is really not free. If you want anything other than self-maintained support, you probably need SuSE or Red Hat somewhere in your plans. If you secure Linux based applications from ISVs, you probably opt for billable support.

One more note about AIX: Stay tuned ...AIX and POWER continue to evolve. What is coming later in summer is very significant. Datatrend can assist clients in executing a nondisclosure agreement that IBM will require to give you a sneak peak of the new features coming that programmers/developers will appreciate. I cannot say more, but you can have a review of these futures when you desire; please give us a shout.

Linux on Intel is a known standard with fantastic footprint, track record, and many customers happy with the result. Not as well known is the fact that Linux on POWER is not only viable, but offers advantages and features not in place elsewhere. According to IBM, there are over 1600 applications test validated to Linux on POWER from over 600 ISVs, of which 800 applications have been certified in the last 12 months. This is happening for a reason.

If you are not looking at Linux today, you could be making a mistake ... it is that compelling. If you are thinking you should move just about everything to Linux, you probably are not being realistic or are in a rare space where vertical scaling and advanced functions are not required. It is all about assessment and putting things in proper perspective and in their proper place. To Linux or not to Linux is not the question. How much, where, and for what is more like it.

Linux: an overview

From college project to global significance in just over a decade

"Linux" stands for Linus' Minix. How is it possible that a single man (Linus Torvalds) creating his own version of something most people have never heard of (Minix) can affect the future of all of us?

As the old saying goes - timing is everything! Linux came along when the world needed it (see sidebar "Linux history"). Although relatively small in itself, it has become the badge for much bigger phenomena - so big that many parts of our lives are already, or will soon be affected.

Linux as the tip of the iceberg

We talk about installing "Linux". But the real Linux is just the kernel of the operating system. When you install "Linux" you get a lot more - as the diagram to the right shows.

Even at the simple level of the code you install on a server, only between 4% and 8% is actually Linux. So, you see that like the proverbial iceberg, most of what we call "Linux" and certainly most of the value that you get from "Linux" is hidden below the surface.

Open standards: the real battleground

The most fundamental difference in the Linux world - both philosophically and practically - is adherence to open standards. This is where "Linux" is changing the the face of IT for all of us.

In this hotly debated area we are certainly not going to take sides. But think of it like a political debate; however passionately you support one party or the other, our job as managers is to work with those who won the last election.

The history of Linux Ancient history: From the beginning of IT many people, especially developers, saw the value of having industry standards for programming languages, operating system interfaces and so on. 1983: Richard Stallman et al had a vision that embraced not only open standards but also open source code and new approaches to licensing and copyright. The GNU project was born. The 80s: Against all odds, the open source movement gained momentum and produced most of the elements of a practical OS - based mostly on a UNIX heritage. 1991: Linus Torvalds needed a better vehicle for teaching OS design at the University of Helsinki. So, he wrote Freax. The SysAdmin called it Linux and the latter name stuck. The 90s: GNU+Linux became poplar in the open source community. But the "distros" (distributions) available were still mostly the domain of hobbyists and engineers. The 00s: The prowess of GNU/Linux made the commercial possibilities obvious and distros appeared that were designed for the general server and desktop markets. Today: Linux is established in thousands of applications covering a huge range of function (see the sidebar "Already using Linux?" in the next article). It has strong backing from IT giants IBM, Sun, HP and Novell. It's ultimate recognition came a few years ago when Microsoft began anti-Linux PR.

In this analogy IT has just lived through a period of Microsoft/Intel hegemony; whether we liked it or not, our job was to work with it. As the numbers in the next article show, this period is over; even for the few who wanted the neo-monopolies to continue, the job now is to work with the new leaders in getting the best return for our organizations. If this challenge is approached objectively, each of us is going to find that the world of Linux will provide superior returns in an increasing number of cases.

So, what should we do about it?

Quite simply, we should be exploiting the potential value of "Linux" both on our servers and on the desktop. As the quotation marks remind us, that means exploiting the value of the whole world of possibilities for which Linux is the flag bearer:

• Organizations with no direct experience with Linux or open source should aggressively look for opportunities - recognizing that some pressure is always needed to overcome resistance to change.

• Organizations whose experience is limited to Linux on servers should take a look at the potential effect on desktops over the next two years.

Later in this newsletter you will find an article to address each of these areas and supply some reasons why so many people are part of the movement to the world of Linux.

... on Linux projects and help available.

Market share: some undisputed numbers

Why would anyone dispute OS market share?

Why? - because the stakes are enormously high. The richest man in the world got that way by selling a few hundred million copies of Windows, followed by a few hundred million copies of Office. Any suggestion that Linux and Open Office are going to make fundamental changes to that franchise are met with strenuous challenges.

We have no desire to increase controversy. So, we eschew the latest disputed "estimates" and stick with some old numbers that are well established. They will suffice to show that Linux should be a serious alternative for all of us.

Undisputed fact 1: Linux is growing fastest

Contesting the content of this chart would be mere quibbling with the details. Everyone agrees with the basics of these IDC measures and projections; Linux is the only OS growing faster than the market - therefore, it is going to continue to take market share.

Undisputed fact 2: Windows continues to lead

Windows will continue to have an enviable share of the OS market for both desktops and servers (the latter is shown in the chart). But don't be fooled by the chart into thinking that Microsoft is "maintaining" its market position. Microsoft attained this market share by being a price leader; now Linux is the price leader. Already we have observed discounting of Windows to fend off the Linux competition and the future projections shown here assume that Microsoft will further reduce prices to stop its share from dropping too much. In other words, the "Linux effect" saves money for people who stay with Windows! But it saves even more money for those who make the change.

Undisputed fact 3: There are many reasons to select Linux

Among the tens of thousands of managers switching to Linux there are many reasons for the decision. While initial cost is the most common, the other top reasons make interesting reading. Overthrowing the current king of the hill demands more than attaining parity; study after study and satisfied user after satisfied user report that the new kid in town is superior in a wide range of ways. A common consensus is that, in many areas, Linux is what Windows should have been - or at least could have been. We examine some of those ways in more detail in the next three articles.

The IBM Linux portal provides a lot of information on Linux in the IBM world - click here.

For an IBM Executive Report that includes coverage of the survey data shown in this article click here.

... on how Linux could affect your future 

Case study: BroadSoft

A telling story

The breadth of the Linux phenomenon is such that it would appear difficult to find one case study that would have relevance to most situations. However, the BroadSoft story will resonate with many who are in the early stages of adopting Linux, because it illustrates a number of key messages, including:

• The use of Linux servers for applications with the ultra-critical reliability needs.

• The ease of migration of existing applications.

• The attractive price/performance available with the right Linux/hardware combinations.

• The value of a good partner in making the move to Linux.

Background

Since 1998, BroadSoft has been supplying solutions to telecomm carriers for both wireline and wireless services. They are one of the leaders in providing solutions to support the next generation of telecomm services with, as you might expect, an emphasis on hosted VoIP.

The 365/24 nature of telecomms requires that every aspect of their solution, including the hardware/OS platform should be "carrier grade". Like many in the industry their base platform was UNIX running on specially hardened servers.

In 2005 BroadSoft introduced new versions of their products for a Linux/IBM platform. IBM has an Integrated Platform for Telecommunications, which integrates carrier-grade Linux on BladeCenter T (a BladeCenter designed to match the special form factors used in telecommunication's racks).

Why introduce Linux?

Bob Weidenfeller, BroadSoft's VP of Engineering picks the two blockbuster reasons for making the move:

• "Some of our customers were asking for it."

• "We can get 30% to 40% more capacity per dollar on the Linux/Blade platform."

With those benefits, the term "slam-dunk" comes to mind!

But, of course, Bob would not even have considered the platform if it was not capable of providing the reliability that their applications and their reputation demand.

How difficult was the migration?

BroadSoft's application was written mostly in Java - so by definition it was portable. There were some specialized codecs (for sound and other waveform processing) that had to be converted with lower-level coding, but this was minimal. That's why Bob reports that "Porting the applications was not a significant issue."

So it was easy?

Far from it. As a purely technical challenge it was not that tough. But as a business decision it was vital to get it right first time. Bob emphasizes how important it is to have the right partners in making such strategic changes; "We worked with the IBM Telecommunications Group in Raleigh (NC). Having a partner who could address all the issues from the strategic commitment to the smallest technical details was vital to our decision to proceed."

100% Linux from now on?

Definitely not! Bob is quick to point out that UNIX will continue to have unique benefits for many customers. "I expect to see us offer the two platforms alongside each other for a very long time."

Summary

A Linux case study that shows:

• Ultra-high reliability for a most demanding application.

• Easy porting because of open standards.

• Impressive price/performance increases.

• The value of IBM as a partner committed to Linux

'nuff said!

Datatrend thanks BroadSoft and Bob Weidenfeller in particular for sharing their experience with our customers.

Linux benefits: Servers

This is an extract of the article "Why customers are flocking to Linux" by Jeffrey Smith, IBM Software Group Vice President.

Flexibility

Linux uncouples the operating system decision from the hardware decision. This flexibility lets customers choose what hardware is best from price and performance standpoints, consolidate operations and better control costs.

Security

From its very origins, Linux was built for security. It shares a common design philosophy with the UNIX operating system and, unlike some other operating systems where security was added later, security was an integral part of Linux from the start.

Reliability

The reliability of Linux is a result of its modular architecture and design, its maturity and the testing and refinement done by thousands of customers around the world during its development cycle. The result is an operating system that you can truly trust.

BusinessWeek online pointed out another of the main reasons that Linux is highly reliable and stable. Its article 'Linux Inc.', reminds us "begun as a meritocracy, Linux continues to operate that way. In a world where everybody can look at every bit of code that is submitted, only the A+ stuff gets in." This is a key reason for Linux reliability, and it's why leading companies in finance, government, retail and other industries are trusting business-critical applications to Linux.

Total value

Early concerns from customers about Linux were often focused on learning and cost curves for competent administrators and Linux's lack of formal support and documentation—both factors claimed to contribute to an operating system that is costly to operate. However, the entrance of enterprise Linux distributions from companies such as Novell and RedHat, combined with other support options from vendors such as IBM, is providing the formal support, albeit at a price, that raw Linux lacked. Also, the administrator training consideration is lessened by the fact that Linux is built on stable market and UNIX practices-skills that are easy to find in today's IT marketplace. This availability of needed skills is being augmented by the arrival of Linux-skilled university graduates in to the job market.

Future value

The numbers are out there. As the BusinessWeek online article points out, IDC predicts Linux server market share based on unit sales will rise from 24 percent to 33 percent in 2007. A Forrester Research Inc. survey states that 52 percent of business users said they are replacing Windows servers with Linux. According to IDC findings presented to the Open Source Development Labs, the overall Linux marketplace revenues for server and PC hardware and packaged software on Linux, which includes IBM middleware software, is expected to reach $35.7 billion by 2008. In fact, packaged software is the fastest growing market segment within the Linux marketplace in terms of revenue, growing 44 percent annually to over $14 billion in 2008. More fuel for hot disputes? Perhaps, but whether these forecasts prove prescient or not, Linux success did not come because of insightful predictions. Real customers with real IT and business challenges are powering Linux adoption based on real value. Put simply, Linux works.

This is an extract of the article "Why customers are flocking to Linux" by Jeffrey Smith, IBM Software Group Vice President. To read the complete article on IBM's Web site click here.

... on how Linux could affect your server strategy

Linux benefits: Desktop

The current situation

Adoption of Linux for desktops has been slower than for servers. Superficially this seems strange. After all, if Linux can be trusted to run mission critical servers, surely it is good enough to run a desktop? Plus there are many more desktops than servers, so the potential for savings is enormous.

One reason for slow adoption is, of course, applications. In particular, dependence on the ubiquitous Microsoft Office applications tends to limit your choice of desktop environments. But this is where Linux plus open source applications could provide the first real alternative in a decade - at the same time as providing huge savings for your IT budget.

Desktop revolution: an illustration

It is easier to see just how significant the change will be by using an illustrative case study. We have chosen a very modest example on purpose. Read through this illustration and, if you recognize some of your own problems, adjust the size to see how it could affect your organization. Let's call this mythical company LNB. Here is a brief description:

• LNB is a small company with 100 desktop machines

• They use Windows XP on all desktops

• Office 2003 on 10 desktops (the "power" users who get value from new features)

• Office 2000 on 90 desktops (general users who use 1% of the features of Office)

Like many companies, LNB used to stay current with the latest versions of Windows and Office. However, the two factors have changed that:

• The administrative costs of upgrading and patching Windows on the desktop got out of hand. Even the decision to move to Service Pack 2 on Windows XP is now treated with great trepidation.

• The per-user upgrade cost for each version of Office has caused them to have a sense-of-humor failure. Especially when they consider that most of their users need very little of the Office function and, therefore, get no value from the version upgrades. This has resulted in their upgrading a few power users to the latest versions but staying back-leveled for the other 90%.

The next two years for LNB's desktops

LNB is well aware that they will be offered upgrades to Windows Vista and Office 12 around the end of the year. Based on previous experience they have no intention of moving to either until the bugs and security exposures are found by others. Currently, their default plan would be:

• Move to Vista in 2007 or 2008 when it is stable and before support for Windows XP is withdrawn.

• Move their power users to Office 12 in 2007.

• For their other Office users they are not sure what to do. Having 90% of their employees on 8-year old software with no support doesn't seem like a good plan for 2007; but the cost of a wholesale Office 12 upgrade is indefensible.

A Linux/open source alternative

The cost of Microsoft upgrades has already pushed LNB into a de facto strategy of having two levels of desktop configuration. One for the power users and one for the other 90% of general users. Taking this a little further, why not use Linux and Open Office for the general-user configuration? This would substantially reduce the cost of OS upgrades and Office upgrades; the reductions could be in the region of $100 and $250 per desktop respectively. Even though LNB has only 90 general users, this is a total of $31,500. But acquisition cost is not the only potential benefit:

• They can upgrade their Office users right away to newer, more powerful software. A small example, Open Office has pdf processing built in.

• End-user retraining may actually be less than with a transition from Office 2000 to Office 12!

• Total administration costs may be similar to maintaining the two levels of Microsoft configurations - some claim it is less.

• Security exposures could be reduced by having Linux on the desks of the general users - who are not only the most numerous, but also the least able to deal with the complexities of viruses, OS vulnerabilities and so on.

• Difficult and expensive stair-step upgrades could become a thing of the past - replaced by inexpensive upgrades that can be chosen and planned based on their merit rather than being driven by a vendor with threats of removing support.

• Access to the vast array of open-source software at little or no cost could enable many new extensions to Open Office as well as completely new applications.

Realistic illustration?

The LNB story is just that. However, it raises some of the desktop software issues that most organizations face today. The double whammy of Vista and Office 12 are just around the corner. They present a compelling reason for most people to take a new look at how we provide basic, everyday applications to the majority of our users.  The simplified approach suggested for LNB may not be the optimal solution for many organizations - but then neither is continuing with the OS and office software approach of the last decade. Next time you update the strategy for your desktop platforms, expect Linux to play a significant role.

... on how Linux could play in your desktop strategy.

TechTip: Researching Linux

By Mindy Jurgenson, Marketing Manager, Datatrend Technologies

Migrating to Linux from a Windows or Solaris platform takes research and planning. After the decision is made, partners like Datatrend are able to help companies design and implement the migration; however, companies often choose to stay with an underperforming or outdated system rather than commit to begin the planning. Linux has become a stable, reliable option, but getting the research to accept these claims is often seen as daunting and arduous. In this edition of the Tech Tip, I am relieving the burden of the research by presenting you with links to valuable sources of information on the web.

Contact us | Visit Datatrend website

All trademarks, registered trademarks and service marks are the property of their respective owners.
IBM, the IBM logo and other referenced IBM products and services are trademarks or registered trademarks of the International Business Machines Corporation in the United States, other countries, or both. All rights reserved.

UNIX is a registered trademark of The Open Group. Windows is property of Microsoft Corporation. BroadSoft and the BroadSoft logo are trademarks or service marks of BroadSoft. Inc.

Linux is a registered trademark of Linus Torvalds.

Brought to you by Datatrend Technologies Inc.

6815 Meadowridge Court, Alpharetta, GA 30005