No. With application pools, you restart the application pool instead of IIS. Yeah dickhead, you know what I meant. MS didn't make IIS more stable, they gave more process isolation capability and the ability for a watcher process to orphan a thread and start a new one when one did become unstable. They didn't fix anything.
I remember very well IIS5 being taken down completely by misbehaving apps, which in turn would take down every single website hosted on the server. Then you had IIS5 configured wrong. IIS5 had process isolation as well, but it was all or nothing, you couldn't logically combine sites into common threads so to isolate everything on a server it took much more memory which limited its capacity, but it was possible. If you had misbehaving apps you could isolate them from IIS and it wouldn't affect the rest of the server. I still have a few dozen IIS5 servers floating around out there.
After suffering for years with crappy apps on IIS5, and moving them over to IIS6, I laugh at anyone who claims IIS6 is not more stable than IIS5. Please. We all "suffered" through it, and we still suffer. I'll say it again, IIS6 is not more stable, it just has self-healing abilities. It still crashes/hangs and orphans threads constantly. You just probably don't notice it anymore - but check your event logs, it's there. It's even less noticeable if you run.NET apps, hell you can have processes restart themselves every minute if you want, then IIS6 would be really "stable". But if you run any classic ASP apps it's more noticeable when a thread dogs and all your sessions disappear.
Well, I'd say IIS6 hangs just as much as 5, but with the application pool controls you have more options for IIS to restart itself automatically. Not really a "solution" to the problem, more of a band-aid. Also you have more control over process isolation than you did in 5, but again overall I wouldn't say it's more reliable.
The figure shows that failures do not increase when the average temperature increases. In fact, there is a clear trend showing that lower temperatures are associated with higher failure rates. Only at very high temperatures is there a slight reversal of this trend.
Sorry for repeating myself, I just thought you'd like to know that you're doing your drives more harm than good.
I just re-read the PDF, specifically check out section 3.4:
The figure shows that failures do not increase when the average temperature increases. In fact, there is a clear trend showing that lower
temperatures are associated with higher failure rates. Only at very high temperatures is there a slight reversal of this trend.
Also this particular study was over just a 9-month period of time, not 5 years as I had indicated.
HEAT? put a fan across the drives make a GIANT difference in drive longevity. You may be interested in reading this article and for even more information this paper. Apparently Google collects performance/environmental/failure data against their entire computing infrastructure and has over 5 years worth of this data. In this analysis (of likely hundreds of thousands of drives of different types / manufacturers over a long period) they found little to no correlation between heat and failure rate of the drives.
Power use, well that one you have a choice. Low power and slow. high power and fast. please pick one. To better qualify that statement, "slow" and "fast" are only relative to each other, not necessarily a user's experience on a particular application. A high percentage of the time, a "slow" machine will suffice just fine (as per your example).
Really? I thought the maximum speed limit* in most of the US was 65 (I just checked - found out most of the US it's 70, the same as here, though significant portions are 65 or 75). Yes, that is the posted "speed limit", but in the US it's extremely common to go 20+ mph over the posted limit. I certainly do.
If someone manages to not see something they're going only that much faster than, they shouldn't be on the road in the first place. When he said "run over" he didn't mean because somebody won't see the car, but because other traffic will tailgate you severely if you're "only" going the speed limit. Again, I certainly do. If you're in a little electric car going maxed out at 62mpg and a Tahoe or something rolls up on you doing 80, it's going to be intimidating.
aren't trucks in the US limited to less than 62mph already? Wow, this one made me LOL. No, trucks aren't limited to that speed. On many highways trucks go faster than anything else (85+ not uncommon). The US is a big place and long-haul truckers have schedules to keep.
you have to remember that people's speedometers do not always show accurate speeds - most are off by 5% or so You are 100% correct. In fact, I'd go so far as to say speedometers almost never show accurate speeds. Even tire inflation and tire wear will affect the speedometer - a worn tire vs. a new tire can be almost an inch different in diameter, so 3" different per rotation. I think 5% off is a low estimate, and of course the actual numerical error will grow with speed. I know my wife's vehicle when it shows 80mph is significantly slower than when my primary vehicle shows 80mph. I keep meaning to time off a few mile markers or get out the GPS to see what the actual error is.
No, wired optical mice have no batteries. Somehow, further up in the thread, the subject of optical mice and wireless mice got merged... I believe your parent poster simply changed subjects (hence the paragraph break).
Good post. The correct term for what you're referring to is "polar moment of inertia". Here is a graphic to visualize - the two objects are the same size and same weight and even have the same center of gravity, but the bottom one has a lower PMI and would be easier to spin.
While we're all nit-picking, you don't necessarily want to "put all the heavy stuff in the center of the car". The actual point of rotation of a vehicle is mid-way between the rear tires (assuming a front-steer vehicle), and that is the ideal spot to concentrate the weight. The further from that spot, the less weight you want. That of course makes your Ferrari / Corvette argument even stronger.
I've been using FAT32 as my shared drive for Linux / Windows, and the only problem with it is the 4GB file size limit. Not the end of the world but enough of a bother that I'm thinking of moving to EXT2.
the fact that the people who actually get paid to maintain Linux, benchmark it and "improve" it only care about its performance in server farms and *not* on the individual desktop You didn't RTFA, did you? Linus specifically spends a paragraph talking about how that is exactly not the case, and that SD was too focused on 3d gaming performance and not general desktop performance like CFS.
In the real world, to run a reasonably reliable application requires a modern rackmount server with remote out-of-band management, redundant power supplies and RAID. If your application itself isn't distributed / load balanced, then yes.
The most common failure modes for computers are hard disk and power supply failures, and this protects you from both. Well, most common hardware failure modes maybe. And bad memory and bad NICs aren't far behind (you do team your NICs, right?).
In fact, many apps will rarely go above 10% utilization (you do monitor your servers with SNMP, right?). Not if you know how to properly utilize your servers. I always hear this as a primary reason of using virtualization (underutilized servers) but in reality if you have a popular application or work for a growing company this is not the case. And what does SNMP have to do with utilization? Why can't you monitor server utilization using any number of other methods?
These kinds of servers are available off the shelf from any major vendor (Dell, HP, IBM, etc) and will run you $2000 or so. Dual power supplies, RAID, and the lights-out management license for $2,000? I don't think so. Especially not with dual processors and decent memory (4GB), and dual power supplies automatically put you into a 2U server. More like $4-5k minimum.
For one software project I'm working on, the vendor recommends 5 servers: one for oracle, two for crystal reports, and two application servers. You don't have to follow vendor recommendations (though they'll try to pull that "not supported configuration" crap on you - don't let them). No reason not to run db and web/app servers on the same physical box and OS. As user load increases and the server runs out of power (since you are monitoring it with SNMP), just start to break it up - move the db off first, then break up the apps into a load balanced setup, etc. This is extremely common and doesn't waste capital up front.
Hardware consolidation with VMware can lead to very big savings in hardware, colocation, power, cooling, and admin costs. Hardware consolidation without VMware (as described above) can lead to even larger savings - VM licensing costs, OS licensing costs and admin (complexity) costs.
And if you get the Vmotion software from VMware, you can move a running virtual machince from one server to another, while it is running, without skipping a beat. Definitely cool in theory, but in practice it is hardly "without skipping a beat" - there is a severe performance penalty. Might not be noticed on your 3% utilization servers though.
The disk score measures disk bandwidth (in Mega Bytes per Second). The conversion to an index number is set up in a way that all modern disks will score at least 2.0.
There is a USB logo on one side of the plug which normally goes "up". Anyway there are only two possibilities with USB, whereas with PS/2 there are 4 core directions if you don't know which way the port is oriented, and sometimes the ports are off by 10 or 20 degrees so it ends up going in at some funny angle and you have to rotate the plug until it inserts.
To clarify, Outlook Express is a good POP3 client. Thunderbird is the best IMAP client I've used. Web based interfaces vary greatly in quality, but can be decently usable. I too am a T-bird fanboy, but I've been moving to web based everything because I'm on too many different computers throughout any given week and I don't want to have to go through the trouble of re-configuring each mail client / training spam filters / etc. all separately.
Firefox is giving URLs with INVALID CHARACTERS to Windows, and Windows is treating them as best it can, which can be exploited.
Nonetheless, Firefox can fix their problem (which they will) but it doesn't make Windows any less vulnerable to the problem from other applications. And, as already pointed out, Microsoft needs to fix the problem because by their own definition (receiving application needs to verify input before doing something) they are responsible.
and Windows is treating them as best it can
No, Windows is not treating the invalid input as best it can. If it were, there would be no problem here.
The problem really does need to be fixed on both ends. Defense-in-depth and all of that.
Mozilla's XUL/XULRunner - you know, the thing Microsoft copied from to make their own XAML in the first place.
What about WCF?
Do you know what WCF even does? Did you seriously just ask if there's a linux equivalent to web services and message queuing? Or is this the part where you go on a rant about how revolutionary.NET technologies are compared with the other available frameworks?
and are able to setup communications channels with a simple config file?
OK now I know you're trolling. You just gave Microsoft credit over OSS for having a simple config file.
Since you haven't used MS in 8 years you probably don't know that security has greatly improved. XPSP2, Vista and Server 2003 are all very secure, as is IIS6 and Sql Server 2005.
This is relatively accurate, though I'd qualify that by saying that they can be very secure but sometimes only after a little bit of work (XPSP2 is still pretty vulnerable out of the box, can't wait for SP3 if they ever get around to it).
Exactly, if I had mod points right now you'd get 'em. I hate to generalize too much, but in this case Email/IM/PM/SMS/whatever are all just ways of sending text in real-time or near real-time from some client/device to another client/device that will await the other user to read it and respond if necessary. You can even look at old style forums as performing the same task. Every SMS capable phone has an email address to email text messages to it. Some IM networks have "offline" capabilities that store messages received when your client isn't logged in. GMail displays your GTalk conversions the same as email. Private messages from social networks can SMS or email notify you. It's all the same thing, but Email and SMS are the only ones that are a defined, universal protocol and not tied to some specific company.
You provided two links to an Opera QA tester's blog? Not exactly an unbiased source of information... is that your blog?
Interesting reading anyway. My favorite line is "They [Mozilla] decide to tweak the spec to solve the problem..." - since when is Mozilla in charge of the W3C? They changed the spec all on their own? Wow.
I actually read through the linked bugzilla report, and it's not at all that they had a bug and were too lazy to fix it. They had a decent discussion about it, and even on the blog's site you linked to the author admitted "To be fair, I do think that aligning a spec to the web can be a good solution, as long as it ensures interoperability and covers the use cases. And the DOM2 Events spec itself is vague. I guess the problem is a spec that attempts to be so language-generic that it avoids referring to even the all-important JavaScript window object, it isn't clearly specified where the window object fits into the event capture/bubble model." (emphasis mine). Isn't that what I said in my last response? Specs are changed to clarify ambiguities; rarely, if ever, are they actually changed to a new direction altogether.
On the other link, again I read the bugzilla report that the blogger linked to. The developers have a discussion again on the interpretation of what the W3C specification says and whether or not Mozilla's implementation is correct or not - apparently they were under the initial understanding that they were (again, ambiguous specs). They also talk about other browsers' implementations and how many sites a change would affect for good or bad. And yes, the DOM is being changed apparently, but not because of that bug report or the Mozilla developers, but because their interpretation of the previous DOM spec made more sense than other interpretations.
You're right, we are pretty far off topic here and could go on forever. Thanks for the links though.
Opera has a content blocker (a fancy name for an ad blocker), not a site blocker....You would have noticed the name and the behaviour if you actually tried to use Opera.
As I went to great lengths to explain, I have and do use Opera. Yes, the button in preferences is called "Blocked Content". Then the title of the list box on the window it opens up says "All blocked sites". You say tomato, I say tomato - wait that doesn't work in text. Anyway, point being it's by no means an "ad blocker", it blocks URLs. Nothing else. It doesn't handle regular expressions, it doesn't know about object and iframe tags, it can't import/export lists, it just blocks URLs. Like I said, site blocker.
As for Firefox and Google, back when Gmail appeared... blah blah blah... - only two browsers are coded for, the rest are usually ignored.
I don't care. Right now, Firefox runs everything Google makes and Opera fails miserably on many things that I need. Opera's problem seems to be centered around javascript, not rendering accuracy.
a bug was submitted for Gecko not being able to properly render "display: inline-block"
I would come back with an Opera bug that's been around for just as long, but, oh yeah - Opera doesn't publicly list submitted bugs and upcoming features so nobody has any way of knowing. And the Fx bug you speak of doesn't prevent me from using sites I need day in and day out.
it's not a nice ecosystem browsers have to play in
Agreed, it's not. I'm all for standards, and Mozilla does render to standards pretty well and they are getting better with every release, as I'm sure Opera is. (Let's not bring ACID2 into this discussion - no, it does not show how well a browser supports standards, it tests a handful of features the authors thought were important and CSS failure modes). Many times the "specifications" are ambiguous when it comes to actual implementation, and the specs aren't "rewritten" as you say, but clarified as to what the specifications were supposed to mean.
addEventListener remains so completely broken in Gecko that KHTML/Webkit and Presto have been strong-armed into violating the specs
Huh? "Different than Opera" "broken". Here are the specs I believe you speak of. You are supposed to be able to stack multiple event listeners to call different functions. That's why there is also "removeEventListener". Opera overrides the old one with the new one (not in the specs), Mozilla preserves both and calls them in order. Yes, that's how it's supposed to work, according to the specs. Nobody strong-armed anybody, Apple fixed it in Webkit because, again, that's how it's supposed to work. That must be one of the many Opera bugs you spoke of, maybe they'll get around to fixing it sometime.
You're right that there is a cult following, but it's to basically any browser that isn't IE. Opera people can't keep their noses out of conversions that don't involve them, and if you want to see Safari fanbois at their best rewind about a month or so to when Safari for Windows was released.
Extensions really do empower Firefox, and you got it wrong to setup your straw man argument. Some extensions are badly written and take up too much memory or CPU time, and this is not Mozilla's fault. Sort of like how Microsoft ends up taking the blame for crappy Windows programs they didn't write. So yes, both "extensions make Firefox great" and "extensions can slow down Firefox" are true, they're not mutually exclusive. It's under the user's control and up to them to figure out the right balance of features and speed.
Firefox doesn't even come with an ad blocker.
What? No browser comes with an ad blocker. Opera has a site blocker that comes empty; it's basically like making hosts file entries. In some browsers you can download a custom user CSS file to accomplish simple ad blocking (doesn't work that well). Fx does it with extensions, and better than any other browser I've seen. But no browser ships with an "ad blocker" (no browser being IE5/6/7, Opera, Safari, Firefox, Seamonkey, Konqueror, K-Meleon, Epiphany, Nautilus, Amaya, Dillo - some of the "IE shell" browsers like Avant and Maxthon do, but they're more like extensions for IE and not browsers in themselves).
I used Fx when it didn't have tab reordering or session saving and yes, those things are cool, but certainly not deal breakers. I still don't even use the session saving features (it's one of the first things I turn off in Opera), and my Fx doesn't crash on me.
So, when speaking of cross-platform browsers (Windows / Mac / Linux), you're really down to Opera and Firefox. I use both, and in fact recently tried to use Opera as my main browser because it is fast and comes out-of-the-box with things that I add to Fx via extensions (e.g. mouse gestures). I dual-boot Vista and Ubuntu on my main home computer, and I'm on numerous other computers both at home and at work, so I've been moving to web-based apps as much as possible (web mail, Google Reader, Google Docs & Spreadsheets, del.icio.us, etc.), and Opera isn't supported on many (most?) of these systems. For some reason Google isn't bothering to make many of their apps work with Opera (Docs & Spreadsheets is completely broken, GReader/iGoogle Home Page don't work well). I thought it was just a google issue and tried to work around it, then I found that both my personal and work web mail systems didn't work right in Opera either (yes, I even tried the IE emulation on all of these). So I went back to Fx. I can also use the same Firefox profile when I dual boot into Linux or Windows which means even less configuration.
So, yes, in fact Firefox really is "That Good A Browser".
it certainly runs on phones that Gecko could never even dream of touching
Apparently you've never heard of Minimo.
"Minimo is a free, open-source web browser for Windows Mobile (Also called Windows CE, Pocket PC 2000, 2002, 2003, and 2003se) and is based on the Mozilla [Gecko] codebase." Not sure why you would think that Opera's rendering engine code base would somehow be significantly smaller than Gecko.
The reason why is because the Woodcrest Xeon is the only 3GHz Xeon that Intel made, and for some reason they decided to standardize this test on "3.0 GHz". Since everybody knows that AMD outperforms Intel on a per-GHz basis, it does lead one to wonder why they chose that particular metric, but honestly no matter what metric they chose people would complain.
For one, AMD and Intel don't release their new chips on the same date, so one side can always complain "that's not our newest stuff" or "yeah, but just wait until our next generation". If you wait for same generation, same CPU frequency chips from both manufacturers before you do a benchmark, you're going to be waiting a while - it'll never happen. And if you pick a "performance class" to set your benchmark on, somebody will complain "yeah but XXXX's chip is.5GHz slower/faster than XXXX's". It's a lose-lose situation for the tester.
Also above there is a discussion about chipsets / power supplies / etc. Again nearly impossible to standardize on this stuff as well. Obviously there is no motherboard that is identical in every regard except the processor that it accepts. Another thread talks about the memory controller for Intel being off-chip vs. on-chip for AMD - so right there you have to go beyond the CPU and include more platform to make a "fair" comparison. Even if they standardized on a power supply, people can argue that the system that pulls less power doesn't need the larger power supply and could save more power (less loss to inefficiencies) on a smaller unit. So do you run the recommended unit for the server or run the same, possibly wrong power supply for both?
My overall point being that in for somebody to do any kind of test like this, they need to setup some base rules. I don't know why people complain so much - they provide all the criteria they chose and did a comparison based on that. If that doesn't answer a question you had, do it yourself or go to another benchmark. Don't complain that the test is invalid because your chip of choice didn't win. For this benchmark, power consumption for 3.0 GHz servers under "real world" conditions (not idle, not pinned, running various applications from databases to web servers), AMD won. Get over it.
It's $20-$30 per server per year, so over a 5 year lifespan of a rack of say 40 servers, that equates to a $5000 difference in lifetime cost, not including the cooling costs (add another 30-50%). Also if they pull less power then less electrical infrastructure is needed (think larger data centers) which reduces costs on UPSs, generators, PDUs, etc. In your environment it may not matter much but it becomes more important with scale. If this particular benchmark is of no use to you, then move on to one with more value.
Slashdot Mirror
Well, I'd say IIS6 hangs just as much as 5, but with the application pool controls you have more options for IIS to restart itself automatically. Not really a "solution" to the problem, more of a band-aid. Also you have more control over process isolation than you did in 5, but again overall I wouldn't say it's more reliable.
I doubt anybody has, since, you know, IIS7 isn't available yet on a server OS (well, a stable one that's not beta).
No, wired optical mice have no batteries. Somehow, further up in the thread, the subject of optical mice and wireless mice got merged... I believe your parent poster simply changed subjects (hence the paragraph break).
Good post. The correct term for what you're referring to is "polar moment of inertia". Here is a graphic to visualize - the two objects are the same size and same weight and even have the same center of gravity, but the bottom one has a lower PMI and would be easier to spin.
While we're all nit-picking, you don't necessarily want to "put all the heavy stuff in the center of the car". The actual point of rotation of a vehicle is mid-way between the rear tires (assuming a front-steer vehicle), and that is the ideal spot to concentrate the weight. The further from that spot, the less weight you want. That of course makes your Ferrari / Corvette argument even stronger.
I've been using FAT32 as my shared drive for Linux / Windows, and the only problem with it is the 4GB file size limit. Not the end of the world but enough of a bother that I'm thinking of moving to EXT2.
The most common failure modes for computers are hard disk and power supply failures, and this protects you from both. Well, most common hardware failure modes maybe. And bad memory and bad NICs aren't far behind (you do team your NICs, right?).
In fact, many apps will rarely go above 10% utilization (you do monitor your servers with SNMP, right?). Not if you know how to properly utilize your servers. I always hear this as a primary reason of using virtualization (underutilized servers) but in reality if you have a popular application or work for a growing company this is not the case. And what does SNMP have to do with utilization? Why can't you monitor server utilization using any number of other methods?
These kinds of servers are available off the shelf from any major vendor (Dell, HP, IBM, etc) and will run you $2000 or so. Dual power supplies, RAID, and the lights-out management license for $2,000? I don't think so. Especially not with dual processors and decent memory (4GB), and dual power supplies automatically put you into a 2U server. More like $4-5k minimum.
For one software project I'm working on, the vendor recommends 5 servers: one for oracle, two for crystal reports, and two application servers. You don't have to follow vendor recommendations (though they'll try to pull that "not supported configuration" crap on you - don't let them). No reason not to run db and web/app servers on the same physical box and OS. As user load increases and the server runs out of power (since you are monitoring it with SNMP), just start to break it up - move the db off first, then break up the apps into a load balanced setup, etc. This is extremely common and doesn't waste capital up front.
Hardware consolidation with VMware can lead to very big savings in hardware, colocation, power, cooling, and admin costs. Hardware consolidation without VMware (as described above) can lead to even larger savings - VM licensing costs, OS licensing costs and admin (complexity) costs.
And if you get the Vmotion software from VMware, you can move a running virtual machince from one server to another, while it is running, without skipping a beat. Definitely cool in theory, but in practice it is hardly "without skipping a beat" - there is a severe performance penalty. Might not be noticed on your 3% utilization servers though.
There is a USB logo on one side of the plug which normally goes "up". Anyway there are only two possibilities with USB, whereas with PS/2 there are 4 core directions if you don't know which way the port is oriented, and sometimes the ports are off by 10 or 20 degrees so it ends up going in at some funny angle and you have to rotate the plug until it inserts.
To clarify, Outlook Express is a good POP3 client. Thunderbird is the best IMAP client I've used. Web based interfaces vary greatly in quality, but can be decently usable. I too am a T-bird fanboy, but I've been moving to web based everything because I'm on too many different computers throughout any given week and I don't want to have to go through the trouble of re-configuring each mail client / training spam filters / etc. all separately.
The problem really does need to be fixed on both ends. Defense-in-depth and all of that.
It would be better if Microsoft called it Services & Technologies for Unix, or STFU.
Exactly, if I had mod points right now you'd get 'em. I hate to generalize too much, but in this case Email/IM/PM/SMS/whatever are all just ways of sending text in real-time or near real-time from some client/device to another client/device that will await the other user to read it and respond if necessary. You can even look at old style forums as performing the same task. Every SMS capable phone has an email address to email text messages to it. Some IM networks have "offline" capabilities that store messages received when your client isn't logged in. GMail displays your GTalk conversions the same as email. Private messages from social networks can SMS or email notify you. It's all the same thing, but Email and SMS are the only ones that are a defined, universal protocol and not tied to some specific company.
You provided two links to an Opera QA tester's blog? Not exactly an unbiased source of information... is that your blog?
Interesting reading anyway. My favorite line is "They [Mozilla] decide to tweak the spec to solve the problem..." - since when is Mozilla in charge of the W3C? They changed the spec all on their own? Wow.
I actually read through the linked bugzilla report, and it's not at all that they had a bug and were too lazy to fix it. They had a decent discussion about it, and even on the blog's site you linked to the author admitted "To be fair, I do think that aligning a spec to the web can be a good solution, as long as it ensures interoperability and covers the use cases. And the DOM2 Events spec itself is vague. I guess the problem is a spec that attempts to be so language-generic that it avoids referring to even the all-important JavaScript window object, it isn't clearly specified where the window object fits into the event capture/bubble model." (emphasis mine). Isn't that what I said in my last response? Specs are changed to clarify ambiguities; rarely, if ever, are they actually changed to a new direction altogether.
On the other link, again I read the bugzilla report that the blogger linked to. The developers have a discussion again on the interpretation of what the W3C specification says and whether or not Mozilla's implementation is correct or not - apparently they were under the initial understanding that they were (again, ambiguous specs). They also talk about other browsers' implementations and how many sites a change would affect for good or bad. And yes, the DOM is being changed apparently, but not because of that bug report or the Mozilla developers, but because their interpretation of the previous DOM spec made more sense than other interpretations.
You're right, we are pretty far off topic here and could go on forever. Thanks for the links though.
You're right that there is a cult following, but it's to basically any browser that isn't IE. Opera people can't keep their noses out of conversions that don't involve them, and if you want to see Safari fanbois at their best rewind about a month or so to when Safari for Windows was released.
Extensions really do empower Firefox, and you got it wrong to setup your straw man argument. Some extensions are badly written and take up too much memory or CPU time, and this is not Mozilla's fault. Sort of like how Microsoft ends up taking the blame for crappy Windows programs they didn't write. So yes, both "extensions make Firefox great" and "extensions can slow down Firefox" are true, they're not mutually exclusive. It's under the user's control and up to them to figure out the right balance of features and speed.
What? No browser comes with an ad blocker. Opera has a site blocker that comes empty; it's basically like making hosts file entries. In some browsers you can download a custom user CSS file to accomplish simple ad blocking (doesn't work that well). Fx does it with extensions, and better than any other browser I've seen. But no browser ships with an "ad blocker" (no browser being IE5/6/7, Opera, Safari, Firefox, Seamonkey, Konqueror, K-Meleon, Epiphany, Nautilus, Amaya, Dillo - some of the "IE shell" browsers like Avant and Maxthon do, but they're more like extensions for IE and not browsers in themselves).
I used Fx when it didn't have tab reordering or session saving and yes, those things are cool, but certainly not deal breakers. I still don't even use the session saving features (it's one of the first things I turn off in Opera), and my Fx doesn't crash on me.
So, when speaking of cross-platform browsers (Windows / Mac / Linux), you're really down to Opera and Firefox. I use both, and in fact recently tried to use Opera as my main browser because it is fast and comes out-of-the-box with things that I add to Fx via extensions (e.g. mouse gestures). I dual-boot Vista and Ubuntu on my main home computer, and I'm on numerous other computers both at home and at work, so I've been moving to web-based apps as much as possible (web mail, Google Reader, Google Docs & Spreadsheets, del.icio.us, etc.), and Opera isn't supported on many (most?) of these systems. For some reason Google isn't bothering to make many of their apps work with Opera (Docs & Spreadsheets is completely broken, GReader/iGoogle Home Page don't work well). I thought it was just a google issue and tried to work around it, then I found that both my personal and work web mail systems didn't work right in Opera either (yes, I even tried the IE emulation on all of these). So I went back to Fx. I can also use the same Firefox profile when I dual boot into Linux or Windows which means even less configuration.
So, yes, in fact Firefox really is "That Good A Browser".
The reason why is because the Woodcrest Xeon is the only 3GHz Xeon that Intel made, and for some reason they decided to standardize this test on "3.0 GHz". Since everybody knows that AMD outperforms Intel on a per-GHz basis, it does lead one to wonder why they chose that particular metric, but honestly no matter what metric they chose people would complain.
.5GHz slower/faster than XXXX's". It's a lose-lose situation for the tester.
For one, AMD and Intel don't release their new chips on the same date, so one side can always complain "that's not our newest stuff" or "yeah, but just wait until our next generation". If you wait for same generation, same CPU frequency chips from both manufacturers before you do a benchmark, you're going to be waiting a while - it'll never happen. And if you pick a "performance class" to set your benchmark on, somebody will complain "yeah but XXXX's chip is
Also above there is a discussion about chipsets / power supplies / etc. Again nearly impossible to standardize on this stuff as well. Obviously there is no motherboard that is identical in every regard except the processor that it accepts. Another thread talks about the memory controller for Intel being off-chip vs. on-chip for AMD - so right there you have to go beyond the CPU and include more platform to make a "fair" comparison. Even if they standardized on a power supply, people can argue that the system that pulls less power doesn't need the larger power supply and could save more power (less loss to inefficiencies) on a smaller unit. So do you run the recommended unit for the server or run the same, possibly wrong power supply for both?
My overall point being that in for somebody to do any kind of test like this, they need to setup some base rules. I don't know why people complain so much - they provide all the criteria they chose and did a comparison based on that. If that doesn't answer a question you had, do it yourself or go to another benchmark. Don't complain that the test is invalid because your chip of choice didn't win. For this benchmark, power consumption for 3.0 GHz servers under "real world" conditions (not idle, not pinned, running various applications from databases to web servers), AMD won. Get over it.
It's $20-$30 per server per year, so over a 5 year lifespan of a rack of say 40 servers, that equates to a $5000 difference in lifetime cost, not including the cooling costs (add another 30-50%). Also if they pull less power then less electrical infrastructure is needed (think larger data centers) which reduces costs on UPSs, generators, PDUs, etc. In your environment it may not matter much but it becomes more important with scale. If this particular benchmark is of no use to you, then move on to one with more value.