On 10/10/05, Alan DuBoff <[EMAIL PROTECTED]> wrote: > Stefan, > > I wouldn't say that's completely true, there are actually some folks that > design data centers around spec numbers. Most of the folks I know do not do > that, but I do hear there are folks that do that. I 'spose this needs to be > taken into consideration, since there are some folks that do that. It would > be interesting to understand why those folks that do design data centers > based on spec numbers do that.
Agreed. For certain types of software, spec numbers are important. If the software in question will spend most of its time doing what the benchmark did. i don't know how realistic that is. > So, let's just say that Linux is faster is some operations (which it most > certainly is, even though it's slower in others). Let's say that you need to > select an OS. Do you select it on the results that point to a faster system > on Linux for a specific task? Nope (if i am making a rational decision based on observable evidence and facts). i deal with (soft) real-time software every single day of my day job. this does not only include (near) real-time complex math, it also includes (near) real-time network data tranfers (to the extent real-time is possible). i keep saying "near" because none of this is implemented in firmware for some specific processors, it's all software. For these types of systems, spec numbers become second in consideration. First and foremost are reliability and predictability plus cost of development plus cost of maintenance (too many buzzwords already). The operating system in question must have a 100% reliable threads implementation, a 100% reliable filesystem implementation, documented uptimes measured in hundreds of days, a flat response time curve under increasing load, 100% binary compatibility when applying upgrades or patches, clearly observable performance analysis tools, and a state of the art set of development tools. these are the foremost considerations. whether or not it came in second on one spec and first on another spec, that will be noted, but is secondary, and ultimately becomes the developer's (my) job: make it as fast as possible, and make it perform to requirements. spending some extra $$$ on four additional CPU's is significantly less expensive than losing several USD $MM because of a crash in the middle of the day. > I'd like to take a step back and concede that Linux is faster at some > operations. I know Sun's performance group is working to tighten up that very > gap in those areas. What does this all mean? sure it is faster on some operations, according to the benchmark. but the benchmark is just one aspect, in my experience. --Stefan -- Stefan Teleman [EMAIL PROTECTED] _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
