May Issue of the Taos Newsletter: Linux

Linux – What’s the buzz about?
Jim Ravelli – Director of IT Infrastructure
Brocade Communications

Introduction

It seems you can’t pick up a business magazine or journal lately without reading about Linux – it’s bordering on a phenomenon and is becoming increasingly important in our professional lives. Linux is a key issue for almost everyone involved in IT. In recent IT surveys, Linux ranked in the top 5 issues that CIO’s are focused on. For example, if you type the word “Linux” into Google, you’ll get almost 100 million hits. It’s the “O/S” that many of the twenty-something sys-admins entering the workforce grew up on and it’s all they know.

Many CIO’s already consider Linux more reliable, flexible, and cheaper than proprietary alternatives. Linux is driving a “revolution” in the data center by:

• Driving costs down
• Driving productivity up
• Embedding itself in commercial products (TIVO, Linksys wireless routers, PDAs, etc)
• Driving the adoption of new technologies such as grid computing and blade servers

It is clearly not something IT professionals can ignore but, quite the contrary, will need to embrace. This article looks at how we got here, why, and what our strategy is for Linux as it becomes a greater component of our infrastructure.

A brief history of Linux – how did we get here?

Linux started in 1991 as a project by Linus Torvalds (a Finnish college student at the time) to write his own operating system. Linus posted his progress on-line and invited comments from others. Sys Admins not being the shy type, he got plenty of help. This “open” review of his source code continued for a few years and was largely the realm of the hobbyist.

Enterprises did not get into the act of using this “free O/S” for a while. Their adoption was largely driven by 3 things:

1. It was free and ran on a low cost Intel platform, thus driving down costs.
2. It was viewed as a more flexible, secure, and cheaper alternative to Microsoft.
3. IBM and others decided to invest lots and lots and lots of money into it.

In fact, IBM has invested well over a billion dollars into Linux so far. Many other companies have been created to capitalize on Linux, and my expectation is that the market has lots of maturing (i.e. consolidating) to do, so expect changes – almost constantly.

Trends in the marketplace

According to a recent Gartner group survey, over 76% of companies (not exclusively technology companies) are either in tactical deployment or early adoption of Linux. In addition, governments such as Spain, New Zealand, and China are adopting Linux rather than pay huge licensing fees to Microsoft. China, in fact, is creating it’s own version of Linux called “Red Flag”.

What’s interesting to note is that while Linux is viewed as a major threat to Microsoft, the vast majority of Linux implementations come at the expense of Unix, not Windows. Windows deployments have held mostly steady, while Unix deployments have dropped as Linux deployments rise. As a result, most of the major Unix vendors have significant Linux offerings and, in many cases, are touting their expertise and leadership in this area. A recent survey (yes, Gartner again) showed the preferred vendors for Linux are:

• IBM – 29%
• Red Hat – 20%
• HP – 17%
• None – 6%
• SUSE – 5%
• Combination – 23%

Noticeably late to the party is the one company that has the most to lose with the adoption of Linux – SUN, which has just now started to make some noise in the marketplace

Because of how Linux entered the marketplace and its initially slow adoption by risk-adverse data center managers, Linux was largely deployed at the edge of the enterprise. Deployments of mail, web, and directory servers are common. As Linux has matured and moved closer to the mission critical portion of the enterprise, implementations of departmental application servers or “specialized” servers/appliances have become more common. If you look at how people are planning to deploy Linux this year, it breaks down like this (Recent Gartner group survey):

• Line of business/departmental applications – 30%
• Infrastructure/network edge – 27%
• Web Servers – 19%
• Data Center servers – 11%
• Compute Clusters – 4%
• Other – 9%

The last hurdle for Linux is to make itself “enterprise” grade. CIO’s are reluctant to install mission critical, “bet the business” applications on Linux primarily because of still-evolving standards and lack of “commercially” available tools and management software. This is changing quickly as Linux matures, but will likely take a few more years (and more early adopters) before Linux is the first choice for Enterprise grade applications.

Brocade’s strategy

At Brocade, we have deployed Linux in significant amounts since January, 2000. Linux now makes up 40% of our installed servers, and that number continues to grow. And like others, it comes at the expense of SUN. We use Linux today for a variety of things:

• Infrastructure – DNS, Mailservers, Backup servers, LDAP, SAMBA, Monitoring, Kickstart
• Engineering build farms
• NFS servers (SAN attached to storage arrays)
• Desktops – primarily in the engineering organization
• Embedded in our products

One of the biggest success stories that we have had with Linux is with our ASIC development build farm. We built this in late 2000/early 2001. Our farm consists of:

• 75 IBM X series machines, identically configured
• Running LSF (Load Sharing Facility)

We have seen huge benefits with this farm in building our next generation ASIC’s. The benefits centered around the following areas:

• Cost - $225,000 vs $1,000,000 in equivalent SUN hardware. Reduced licensing costs -since jobs turned so quickly, the contention for licenses was eliminated (no need to purchase more)
• Speed – Build time improvements of 50 - 80%
• Reliability – 100% uptime since any one failure does not affect any of the other machines. We just pull the offending machine out, replace it, and keep going.

In addition, scaling this environment is as easy as just adding more hardware - No downtime required.

What’s next for us?

We’re continuing to look at ways we can leverage this platform. Two of the opportunities that we have are 1) Linux on the desktop, and 2) GFS (Global/Clustered file system).

The Linux desktop environment is rapidly maturing and provides the potential to reduce dependence on, and subsequently cost of, equivalent Microsoft products. Many of the software products that duplicate a windows desktop have reached a level of maturity and reliability that they warrant a look. File management, email, and the office suite are now worthy of use with certain users versus wholesale “rip and replace” for the enterprise.

Clustered file systems represent another interesting opportunity. We are looking at them as a potential replacement for some applications use of NFS. The ability to scale application server performance by the adding machines, yet have access to the same file system has lots potential to improve productivity and reduce our costs. While this effort is in the early stages, it is one we think could be of great benefit in a year or so.

Risks and Challenges

While Linux is a maturing product(s) with lots of vendors offering competing products, it is not without its issues and challenges. Some of the issues that we have seen are:

• Lack of applications – some products just aren’t there yet
• You can quickly grow number of servers causing data center issues (electrical, cabling, heat)
• Many Linux “distributions” – who do you choose, why, and for what?
• Issue of self-installation on the desktop – granting of “root” privilege causing security concerns.

Other concerns center around what could be considered maturation issues and are not exclusive to Linux itself. As Linux matures, the marketplace will make decisions on these, but as you move forward please be aware of them:

1. Issues of commercialization. There are literally hundreds of Linux packages out there supported by all types of companies. Issues of application support, licensing, vendor consolidation are still being worked. The number of vendors in the marketplace today will change, and who’s staying, going, or combining forces is actively being decided. Additionally, Linux is lacking mission critical applications and associated support that will allow it’s full entry as an “enterprise grade” operating system.
   
2. On the Desktop. As Linux gains traction on the desktop, it starts to increasingly impact the folks in Redmond. They will certainly act to protect their turf. My hope is that they do it by creating better and more secure products and not look to “legal” remedies to slow Linux down (See next item)
   
3. Legal issues. As almost everyone is aware, SCO is suing IBM and many others for what it views as protection of its intellectual property rights. This case has not been decided in a court of law, only in the court of public opinion (Linux won). The final result of this case, assuming SCO prevails, could have huge impact on the adoption of Linux. If SCO does win, it could cost companies millions in licensing fees. As of yet, the threat of this legal action has not slowed the adoption of Linux at all.
   
4. Security exploits. As more and more machines run this operating system, expect that security problems such as viruses and worms will become more common. It’s logical to assume that there are folks out there writing them as we speak and whatever you are doing to protect your windows environment, you’ll need to do the same for your Linux systems at some point.

In summary, Linux is a full-fledged revolution for business. It is helping to drive down the cost of computing, increase the productivity of our computing assets, and be infinitely flexible to meet our needs. Ultimately, time will tell how successful this operating system was in becoming the “next big thing”, but if early indications continue, Linux is here to stay and in a big way.

Jim Ravelli joined Brocade communications in January 2000 as the Director of IT Infrastructure. Jim built the group that is responsible for architecture, strategy and process development, and support services for all facets of the computing infrastructure. Jim often speaks to customers on behalf of Brocade on the application and use of SAN technology.

Prior to Brocade, Jim was the President and Founder of Archinetix, a company specializing in infrastructure and network consulting. Archinetix customers included Intel, Synopsys, Portal Software, and Network Appliance. Prior to Archinetix, Jim spent 6 years at Synopsys starting as the IT manager for their Beaverton Oregon site. He was promoted to Director of Network and Computing Services where he was responsible for the worldwide computing infrastructure of Synopsys. Jim has held several other positions with companies such as Procter and Gamble, James River Corporation, and Mentor Graphics. Jim holds a Bachelors degree in Engineering from Arizona State University.