My reply is missing the point, I know that, but I can’t resist:
I’m sorry but I really think that’s just extremely perverse to run a huge database on Windows Server.
Why did you choose Windows for this role?
@ Samuli: Why not? Seems to work well enough for Stackoverflow, which is very database dependent but still very fast and runs on a modest amount of hardware.
Or does this go against your religious beliefs?
This blog post is the exact reason why no uses MS products for anything that needs to scale past a “small” project/website/performance/etc etc.
Just remember - Microsoft owns the patent on doing this:
Dynamic SKU management
http://patft.uspto.gov/netacgi/nph-Parser?patentnumber=6442559
The problem with this is not that there are multiple versions, I want multiple versions if I am going to pay for it so I don’t have to pay for things I do not want
The problem is that there is not one single matrix of features that each has (or if there is one it it very difficult to find), and the consequences of each of these features/limits are not shown
The other examples you gave are all clear, this is what you get, you get more if you pay more. Microsoft seems to be trying to get you to buy more than you need, or you risk buying something that will not do what you want …
“… as if 17 billion bytes simultaneously cried out…”
Umm, ain’t that supposed to be 16 billion bytes?
16GB=17,179,869,184 bytes
OS X also only comes in awesome.
yeah awesome tiger, awesome leopard, awesome snow leopard ( ~ $29 upgrade from leopard and ~ $169 from tiger)
Yeah, thank God that Linux only comes in one flavor.
I hope you meant FOSS. Free and Open Source is awesome for all the advantages that gives the world. Why do we have to invent hundreds of word processors, when we can invent one and move on as a Human race to newer more interesting projects. Its those “don’t teach anyone what you have been taught” principles that are causing us to not advance.
What I like about this post, is it goes even further to explain this greed, where people sell the same product under different labels to make more money. We have seen it with intel chips, ati and nvidia cards, and notoriously windows.
I disagree with the hosting plans argument, since in that specific case there is some costs associated with the extra service. Extra support, extra hardware, extra resources. I guess your way would be to give everyone the same exact hosting plan, but that is not necessarily the right thing to do. Each situation is different, I agree that root kits, and limitation features should not come pre-installed (or post-installed) and we should look to the future rather than try to squeeze money of what has been made in the past.
“The problem is that there is not one single matrix of features that each has (or if there is one it it very difficult to find), and the consequences of each of these features/limits are not shown”
I agree that this can be annoying for an ordinary user, but I think people making business decisions should be prepared to put a reasonable amount of effort in to research before making the choice. And I’d have thought that calling something standard edition was offering a bit of a clue that it wasn’t going to be the edition with the highest features. Also, this matrix seems to be pretty clear about the memory limits.
http://www.microsoft.com/windowsserver2008/en/us/compare-specs.aspx
“Microsoft seems to be trying to get you to buy more than you need, or you risk buying something that will not do what you want …”
I though the complaint was that they were not forcing you to buy more than you need.
In any case, I believe there is a blog post somewhere (or webpage) that discusses this limitation and the registry key involved to allow your edition to see your extra ram. Hope you didn’t pay them more money…
This has nothing to do with religion.
Yes, It does seem to work surprisingly well for stackoverflow most of the time (as seen from end-user point of view).
But there are problems, as the one discussed in this very post, and many others. This is why I was curious to know the rationale behind choosing a windows server as a database for the high-traffic and desirably highly available and scalable web-application. I think it is a very unusual choice if you ask any IT professional (excluding of course microsoft’s marketing department).
Samuli, the tone of your post reflects that it has everything to do with religion. That’s the insidious nature of it: you believe it is unwise to use the platform, and, in the face of a counterexample, you act like it is an anomaly and try to write it off in a hand-waving manner. It is possible to run Windows server to serve a database or a web server. Now, do you have specific scaling problems relating to stackoverflow.com?
@Hatem Nassrat
"> OS X also only comes in awesome.
yeah awesome tiger, awesome leopard, awesome snow leopard ( ~ $29 upgrade from leopard and ~ $169 from tiger)"
Yeah, or you could instead pay $219 to upgrade from XP to Vista, and then another $219 to upgrade from Vista to Windows 7, and then pray that it will at least be on par with Tiger this time.
While I won’t comment on the practices of microsoft, I have to say that I am surprised nobody at stack overflow was aware of this limitation. Not only is it in the first chapter of any windows server book, but it comes up quite often when purchasing.
Not knowing the limitations of your OS is just as irresponsible as not provisioning the right amount of hardware for a server.
Something that appears to be missing from all of this is how do you actually go about getting from Windows Server 2008 Standard to the Enterprise edition.
Do you reinstall? Is there a straightforward authorization process you can execute from the 2008 Enterprise disc to convert your 2008 Standard in-place?
Let’s assume you have to reinstall, does it go into the same Windows directory? Will it overwrite my registry? Will all existing software from Standard edition work? Should I risk doing this on my production server? I wonder how long I’ll need to take my website offline to do this, maybe it would be easier just to buy another server rather than risk an unrecoverable site outage.
These are not the kind of questions you need to be asking yourself after a memory upgrade. We had a similar experience using PAE to get above the 4GB limit on a 32 bit system, but it was easier to plan ahead and use Enterprise Edition since these are known hardware limitations.
As many commenters have noted, FOSS software does have examples of tiered licensing. The one that jumps to mind is Red Hat Enterprise Linux, which has similar restrictions based on number of CPUs, amount of memory, etc. The difference vs. Microsoft is that if you don’t need access to the high priced support drones nor the gold stamp of “Red Hat Enterprise”, then you have options. Real options, not just love-it-or-leave-it options. You can get the exact same versions of the software for free with CentOS ( www.centos.org ), or for a little bit of porting/integration you could easily choose another distro entirely.
I’m not slamming Red Hat Enterprise. I’m glad they are convincing the pointy-haired bosses that they need to value “enterprise support”. That helps pay for the good things that Red Hat does, like supporting Fedora.
“Just don’t use Windows… Oops sorry, I forgot, you’re dumb!”
What an eloquent and persuasive argument!
Not to support Microsoft’s pricing, but segmented pricing may not be so much about extracting as much value from rich customers as to ensure that you are capturing the value people are extracting from the software. One of the frustrating things as a software publisher is how do you charge for the person who uses it occasionally (who wants to pay less) versus the person who depends on it for their business. All things being equal, they are willing to pay different amounts because the value they extract is different. Segmentation should try to capture that - although most segmentation exercises end up in extremely confusing pricelists, for the consumer and even the sales people selling it.
Excellent article - had me hooked with the weasel pictures.
Hahahahaha, what a n00b!
The problem is that marketing weasels generally don’t really understand their markets or what segmentation is really for. They are just regurgitating what their heard in MBA school cookbooks without any thought or consideration about what they are actually doing.
I can say this because 1) I have an MBA, 2) I do think about this kind of stuff for my company, and 3) it often doesn’t really make any sense to segment a market that doesn’t really have a natural cleavage point in the first place nor is it necessary to segment just because you can.
For my company, simple and painless up-scaling, up-grading of performance is a selling feature so we don’t segment on trivial performance levels like this example - instead we segment on mutually exclusive functional options and performance levels that cleanly separate. Even then you can take the code or the hardware you were using and re-purpose it without throwing anything way - it’s only an incremental cost rather than 3 steps forward and 2 steps back with the full 5 steps going out as cash. We sell software and hardware so it’s a win either way for both us and our customers to help our customers control their costs.
These aren’t just marketing weasels, but they are unthinking, mindless marketing weasels - pretty much describes most MBAs graduated from US or UK schools, sadly. It’s a large part of the reason the Fortune 1000 is in such dire jeopardy but they don’t even know it. Time to die.