By: Bill Buros.
As several of us (expanding quickly into many of us) are involved in kicking off this new Think Power Linux community, Jeff Scheel has mentioned in an earlier post
that he'd like to get more "experts" listed here on Think Power Linux.
Working in the IBM Linux Technology Center
is a fascinating and interesting experience in the industry. We work daily with many of the best programmers in the world on Power systems, working on today's servers, tuning, optimizing, improving the software stacks specifically for the many POWER7
server solutions, and of course focusing on tomorrow's ideas, products, and technologies.
In that context, I hardly feel qualified as an "expert", but I will certainly admit that being in the middle of Performance on a server platform that prides itself on industry leading performance is a cool place to be. My background is years of development programming, development management, product ownership, working closely with our independent software vendors, marketing/sales/account teams, and over the last years in the performance analysis teams on Power servers. That background gives me a good perspective of working with varying perspectives and views, which allows us to easily shift from helping customers, working with business partners and software vendors, and designing the new products to solve ever more complex problems.
In a performance team, there's always a balance of nailing down performance focus items to be black and white unambiguous information, in an environment where the context of useful information usually falls back into the realm of "well, it depends". We rely on tangible benchmarks, recognizing the good and the bad of benchmarks. In the good sense, benchmarks can often provide clear and useful comparisons between the combinations of hardware, software, and interconnected systems. In the bad sense, benchmarks are generally not always reflective of real-life customer use, and in marketing collateral, can be mis-used and poorly characterized.
Our focus is in transforming the benchmark, system, server, software experiences into real-life best practices, hints, tips, and recommendations. We've discovered that it's easy to create wiki informational pages, and much harder to maintain them with current information over the long-term. For example, over the last couple of years, we went a little crazy with pages, with indexes and targeted information spanning many aspects of performance.
With the new Think Power Linux community, our team has already started to update the many performance pages and move them over to the new community, Developerworks articles, and IBM's Information Center for Linux
. Two very recent examples are:
These two articles are good examples that reflect contributions from a wide breadth of people - in my mind the real experts - like Steve Munroe, Michael Meissner, Chakarat Skawratananond, Peter Wong, Brian Hall, and many others.
In the Think Power Linux community and on Power Linux, watch for more articles, new updates, improved tools, and of course ever-improving performance. We hope you'll help us with questions, thoughtful comments, and insights
into what you're doing with Power servers, Linux, and the applications
you care about. Comment here (below), or post a message/question on the Think Power Linux Message Board