From Brian Cox, the director of Software Product Management and Marketing in HP Business Critical Systems...
I frequently am asked the question by customers which operating system should they use for their mission-critical environment. My HP field colleagues, who see customers on a daily basis, also get this same inquiry. I will attempt to provide my perspective on this important question. However, like most things in life, the answer is not “one size fits all”.
First, a bit of background on me. I manage the HP team responsible for the product management and marketing for the system software on the Integrity brand of servers. The Integrity servers have been designed for customers’ most demanding applications, and my team helps to plan and to deliver the operating systems HP-UX, Linux, Windows and OpenVMS, along with closely associated software such as compilers, file systems and clustering that run on these servers. These four operating systems are my “family” at work. I love them all.
Coincidentally, I happen to have four children at home. And they are all different, which is a good analogy for how I view the four operating systems. Like my children, each has developed its own set of strengths, set of followers, and is at a different stage of maturity.
Without stretching the family analogy too far, let me tell you about my two oldest sons, Jonathan and Alex, ages 16 and 13, respectively. Both are highly competitive basketball players. In fact, my 16-year-old was MVP of his Junior Varsity High School basketball team last year. They play one-on-one basketball against each other all the time. However, my 16-year-old Jonathan always beats my 13-year-old Alex. Jonathan is simply bigger, stronger, and has more years of skilled coaching and practice drills. Similarly, when I am asked which is better, HP-UX or Linux, I view my response as I view comparisons to my sons. I love them both, but for the most demanding basketball games, I will usually recommend Jonathan as my starter. However, in games against other players in their early teens, I am very comfortable putting Alex into the starting line-up.
Here’s the additional rub. Though Alex is shorter at 5 feet 8 inches, his older brother Jonathan is about topped out at 5 feet 11 inches. However, based on growth charts, Alex will likely end up being about 6 feet 4 inches. Therefore, if you asked me today, who would I start in the most demanding basketball games, I would still say Jonathan. However, if you asked me that same question in five years, then I would seriously have to consider Alex as my starter.
Similarly, if you asked me to choose between HP-UX and Linux for a customer’s most demanding workload, I would typically recommend HP-UX. However, if my customers’ time horizon is five years from now, then I would seriously consider Linux (by the way, you could replace OpenVMS for HP-UX and Windows for Linux in the above comparison and I would give you a similar answer).
There will always be exceptions to these general recommendation based on a customer’s unique situation. For example, the customer might have a deep history with Linux and is able and willing to do the integration themselves that already exists within HP-UX. In that case, I would be quite confident to recommend Linux. On the other hand, if the customer needs proven 99.999% uptime, one-stop serviceability and proven scalability to millions of transactions per minute or 10s of terabytes of data warehouse size, then HP-UX would be the safer bet today.
On a feature-by-feature basis, Linux more closely matches HP-UX and the other enterprise UNIX offerings with each passing day. However, the open source community is still building out the breadth of manageability toolsets, and most importantly the proven case studies where the system regularly delivers flawlessly under the most demanding workloads. A good description of this Linux evolution is a white paper written by Ideas International that can be found at: http://h71028.www7.hp.com/enterprise/w1/en/os/hpux11i-v3-and-linux.html
As I was writing this blog, a manager on my staff named MJ was at the Gartner Data Center Conference in Las Vegas.. She kept me updated with text messages about highlights of the conference. During a session entitled “Is the Luster Off Linux?,” hosted by Gartner analyst George Weiss (its OS and server expert), George polled the audience of about 150 people and asked how many were actively moving from Unix to Linux/Windows, and just a few hands went up. This is consistent with findings I saw earlier this year from another industry analyst firm called Gabriel Consulting Group on the current state and future for UNIX. For background on these findings see www.hp.com/go/gabrielonunix.
In addition to the poll taken during the Gartner session, here were some of the key points:
* A New Reality Has Set In: Linux Is Not Inexpensive
* Big users are committed to Linux but believe they're paying too much.
* Linux should be synonymous with ease of application portability and zero switching costs, but is not.
* What should be a leader in virtualization is not yet.
* Virtualization market presence — missing in action
* Escalating contract costs and restrictions
* Large nonpaying population: Equated with lower value?
* TCO evaporating up the food chain
* One-vendor dominance, despite dozens of distributors
* "The stand-alone OS is dead." (Nat Friedman, vice president, Novell )
* "Linux is bloated" comment by Linus Torvalds, Linux creator; (A forewarning?)
* Difficulty growing beyond 1- to 2-socket servers
In summary, Gartner did not recommend that customers not move to Linux, but they did state the issues and risks. Most vulnerable are systems in the $5-25K range. They acknowledged that systems in the $25-500K range are UNIX strongholds and require stability, strong management capabilities and virtualization.
So when customers ask me, whether they should to choose between Linux and HP-UX for running most demanding workloads for deployment today, I will honestly have to recommend HP-UX in most situations. However, Linux development will continue for years to come. Thus if a customer’s time horizon is five years from now, then I think it would be time for a serious comparison, just as I will do with my two oldest sons in playing one-on-one basketball a few years from now.
In closing, and to wrap up my family analogy, I love all my children just as I love all the operating systems for which my team has responsibility. They are simply at different stages in their development.
Brian Cox
