Slashdot Mirror
A Justification for Server CALs?
bourne.again asks: "I'm a bit confused about server client access licenses (CALs). I've looked at it from every angle I can think of, but I'm still stumped. I can't think of any justification for CALs other than greed. If you think about it, requiring CALs means that it is possible to buy a copy of a Windows server OS that can run on a server, but can't actually server anything because it has no CALs. That's a bit ridiculous. The same goes for per-cpu licenses. Shouldn't it just be per machine? An extra CPU doesn't allow you the full capabilities of a second machine. It's still just one server/workstation. Can somebody enlighten me on this, please? Why should we pay for server software, and pay per client too?"
You pay for it this way because that is how the vendors choose to sell it. You can purhcase other software if this does not suit you.... Why do you pay several thousand dollars for a car, because that is the price. Want something cheaper, buy another model.
I reject your reality
what you're told. That's why. Remember way back twenty years ago when things like a Mac cost $2500? Well, profit margins ain't what they used to be, so new and inventive ways have to be thought up for profit to be made. If people really cared about getting screwed, we wouldn't all be paying "universal access charges" on our phones.
I guess someone could make the argument that it is more fair to the smaller customers. Pay $8000 for a sever license, and $50 for each of 75 clients, rather than pay $5000 for a server license. But yeah, it's probably mostly greed.
Microsoft used this strategy to underprice Novel. With Netware, if you had n servers and m clients, then your cost was proportial to n * m. With Microsoft, it was more like n + m, which is weird but cheaper.
Client CALs are an attempt to balance between extremes of licensing. At one end you have Oracle and the infamous "processing units" where every 100 MHz of chip speed present was billable (no matter if it was due to a single fast chip or multiple slow chips). At the other is a license per machine which can be abused by having a high end multi processor machines.
Microsoft (since you use the term CAL) has given most products the option of either being per processor (a decent compromise in "bang for your buck" at the high end) or per server with CALs (a decent compromise at the low end, while scaling the revenue with usage). Frankly, I don't find it odd at all, unless you want to contrast it with free software.
Sig under construction since 1998.
You don't get to dictate their licensing terms any more than they can force you to purchace something. If you don't like it, don't buy it.
Of course, the argument could be made that since Microsoft is a convicted monopoly you don't really have a choice, but enforcing anti-trust law is part of captialism too.
Just start all free/libre/open stuff to begin with, and you'll never be troubled with this again. Leave the closed guys out of the game. Deny them access to the gene pool. They don't need to be mucking up your systems at all!
A more sensible licensing scheme might take into account actual computing power, networking capacity, etc. Of course, the pricing of replicatable goods is completely arbitrary. It has nothing to do with economics that deals with the distribution of scarce resources.
Escher was the first MC and Giger invented the HR department.
Since you've pirated your desktop software, you have to pay for it somehow. So you pay for it on the server.
Yes, I know. Don't even point out the horrid illogic in that concept. I also like how w2k AS would leak CALs forever if the client was reformatted, so you had to buy licences over and over. The only fix was to delete the stuck licences and forcing a re-activation of them with *ahem* third party software
I also loved how their fix for it in SPwhatever didn't actually fix the broken licences, it only prevented further problems. I also love how you have to buy both Citrix per-client licences _AND_ M$ per-client licences. In fact, I'm full of love for the whole awesome business model of finding ways to pick up an extra few hundred dollars per node that doesn't have to show up on the ofiginal price tag. Brilliant!
Why does Server 2003 cost me per server, instead of per media? The theory is that after spending years of programmers salaries without compensation, they have to make back the money they already spent just with licensing fees. It's an art, deciding how to price things so that both the little businesses can afford it and still recoup your costs.
Server CALs are just another tool in balancing software costs proportional to the usage (customer perceived value) and ability to pay.
It may be partly greed, but remember that most companies have to use their successes to recoup the costs of software that DIDN'T sell. Remember Infocom? Looking Glass Studios? Pricing is an art. CALs are a tool.
Everyone is entitled to his own opinions, but not his own facts.
Companies price their server software this way for the same reason that airlines charge last minute (i.e. business) travelers more than leisure travelers: price discrimination. In other words, charge more to those who are willing to pay more, less to those who are willing to pay less. Sofware firms are in it for the money, after all.
"The dinosaurs died because they didn't have a space program." - Niven
...you're asking a bunch of Open Source fanatics for a justification for client access licenses? Good luck.
~~~~~~~
"You are not remembered for doing what is expected of you." - Atul Chitnis
I don't really find the licensing scheme's that odd - they are just different ways of licensing something that is in the end not a product but intellectual property. The product is the server hardware - the software is just intellectual property of something that uses the product. Sort of like text in a book is simply ideas whereas the book is a product.
The thing I find really odd is that companies can make 80+% profit margins on intellectual property ("ideas") that are mostly 10-15 years old.
One other thing that I've noticed is that the more specialized the software, the fewer copies that are sold. This software usually turns out to be more expensive and far more restrictive than more popular software. For example, at work I use a streaming video package that we spent $10k for. When you install it, you enter a license number, it contacts the company's server over the net and then activates the software. I have two problems with this.
Anyway, events like this make me really appreciate the "freedom" aspect of free software and open source software. I have little time to dick around with playing games like that with vendors. Not to mention that it leaves you completely at the mercy of the vendor.
So my point in all this is that eventually companies will learn that it's in their interest to start using software that respects flexibility rather than being node-locked, not allowing reinstalls, charging for extra CPUs & network connections, etc. But it won't make a difference until it starts affecting the bottom line of these companies.
Slashdot: Failed Car Analogies. Amateur Lawyering. Anecdote Battles.
...it is in the standard license agreement does not mean you have to accept it. These are not EULAs for off-the-[NAME OF FAVORITE RETAILER HERE]-shelf shrinkwrapped software. You'll be surprised at how much you can negotiate - especially at the end of a business quarter (the seller's - not the buyer's).
Laws affecting technology will always be bad until enough techies become lawyers.
You now understand part of why managers started to appreciate some of the benefits of open source software. The accounting is much simpler. Service agreements are usually based on much simpler metrics (i.e. how many servers do you have running our product).
$999 for unlimited CALs. No fuss, no muss, and it can act like a PDC.
If you need groupware, go with Kerio MailServer until Apple makes their own. It's a bit pricey, but worth it.
Because you're willing to. They know they can get customers who have more money to pay more. If they didn't have all these price tiers, CALs, and such, Windows for the end user would cost more and Microsoft's total profits would be less. The more they can divide up their licensing to discriminate between specific markets, the better they can optimize the pricing to maximize profit.
You don't have to buy what you don't want. If you don't need Active Directory, or you feel up to trying another LDAP solution, you can just use Linux (of *BSD) and Samba for file serving.
The truth of the matter is, in enterprises of a certain size, the licensing costs aren't much of a consideration at all. tIn fact, they're basically insignificant compared o the costs of administration. I'm not taking a side on that one way or the other. I'm just saying that if a few thousand bucks is a big deal to you, then don't use stuff you have to buy, because obviously you have different priorities.
In other words, the cost doesn't actually have to be justified. It's obviously acceptable to the market at large, whether or not you're willing to pay it.
Slashdot - where whining about luck is the new way to make the world you want.
I see people have already given you the "because people accept it" economic answer. However, there is some logic to server CALs. It goes as follows:
The value you get from software is proportional to the number of your people using it.
In the case of software that runs on the client machine, if you have N users you are generally expected to pay for N licenses, so that the amount you pay is proportional to the value you get.
If I have a server-based web content management system, and have 10 people using it, I'm getting twice as much value from it as if I only had 5 people using it--just like if it was a client-based web editing system. Basically, whether the software resides on the server or the client is a mere technical detail that doesn't really affect the value obtained, and so shouldn't affect the pricing too much.
So CALs are a way to make server software pricing more like client software pricing--i.e. proportional to value obtained.
The bit you should be asking about isn't the CAL; it's the initial server software price. The logical reason for an initial per-server fee is to cover the expected distribution and support costs when you purchase the software and try to set up the server. Unfortunately, that lump sum basically makes it disproportionately expensive to set up a server for a small number of users. That's why there has been a trend towards offering cut-down SMB versions of server software with a lower up-front cost.
Per-CPU and per-MHz licensing is an attempt to shift the expense towards those who can apparently best afford it. Or to look at it from the other side, it's another way to try and make things relatively cheaper for entry-level/low end customers. Like income tax, in other words.
One thing you didn't bring up is the difference between workstation CALs and (concurrent) user CALs. Personally, I think workstation CALs are a bit of a rip-off, and licensing should always be based on users, whether the software is server-based or client-based. If I use 5 computers in various locations, I'm not getting 5x the value of a server-based CMS--there's only one of me, and the fact that I use 5 machines instead of carrying 1 laptop everywhere should (again) be a mere detail that doesn't affect the value obtained, and so shouldn't affect the price.
Disclaimer: Opinions mine, not IBMs.
GCHQ Quantum Insert installed. If only our tongues were made of glass, how much more careful we would be when we speak
It might be a PITa to configure (I just tried too long ago) and there might be features you need that it doesn't have, but it might me worth trying.
I don't like SuSE, but it I had to admin Windows accounts, Novell SuSE doesn't sound so bad after all.
If they charged a flat price, they'd need to charge a lot more to make the same amount of revenue. So somebody using it with 5 clients would pay $10,000 as would the person using it with 500 clients.
Does that seem like a better system to you? Well, probably if you're the guy using the 500 clients.
An important point to note about Microsoft Server CAL's, as long as each user (or device) has a CAL, you can add as many servers as you want (for the base cost of a server). So if you have one server and 20 CAL's, and you need to purchase a second server, you only need to pay the ~$600 for the Server software, all the clients are covered by CAL's already.
CAL's allow you to license the software based on your number of computers rather than users. If you have 50 users who share 5 computers, then you only need to have 5 CALs. On the other hand, in development organizations 5 users can end up using 10 computers if users need to work on multiple computers. In this case the per user model would be less expensive.
I think you already understand the economics of per user licensing. It allows Microsoft to have different price points depending on the size of the organization.
With Netware, if you had n servers and m clients, then your cost was proportial to n * m.
Since when did you have to purchase both an NDS license and a NetWare [or SuSE] server license?
Novell servers have been essentially free since 4.x [circa 1995]. In fact, way before the era of pf/ipfw/ipchains/iptables and whatnot, you could get a really nice low-level switch/router/packet filter/firewall/etc simply by installing a NetWare server and then removing NDS from it.
As I recall, the way M$FT killed Novell was when they put NT 4.0 Server on the "honor system": When you installed an NT 4.0 Server, you had to enter the number of clients "on your honor" [so everyone naturally installed it with like 999 clients, even if they had only purchased a 5-user license].
I.e. if you had n servers and m clients, then for about a four year period, from maybe mid-1996 until mid-2000, Netware cost on the order of m, and NT 4.0 cost on the order of "1" [and that "1" was probably a bootleg anyway].
What's the alternative to a CAL arrangement? The vendor could charge a flat rate per server or even per company, but that rate would be a lot higher than a per-seat license in order to keep the revenue the same. Should companies that only have five employees using an application pay as much as a company with 500?
Also, CALs actually make sense if the license fees include support. Generally speaking, more users results in more support calls.
Because it's their product, and they can charge whatever the hell they want?
Anyway, it's not a bad setup - that way the large companies that use it for thousands of clients get to foot a lot more of the R&D and support costs than the small companies using it for 10 clients. I don't really see the problem here.
Cars and SUVs do cost what people are willing to pay, but how do people justify to themselves what they are willing to pay? Why, they take into consideration the existence of the parts, labor, taxes, etc, factor.
I think it makes more sense from a business point of view. You're pricing based on the number of users verses the number of servers. You know that for every employee/managers/client/whatever that you need a cal. 50 empoyees? You just need 50 CAL's and you're covered. It also lets the vendor charge more money to bigger companies then small companies. Assuming that a CAL gets assigned to a person, it's a fixed cost per employee to a company, just like the empoyee's salary. As you can afford to pay more employee's salaries (20k/year min), you should also be able to afford more CAL's(~$200-one time cost). I tend to like this model a bit more then the per processor or per installation licensing, becaue it means the cost for a small company is a lot lower. I like the M$ provides both per-processor and CAL licensing, that way you can choose which model makes the most sense (or is cheapest) to your business.
"What is the justification for server CALs?" is just another way of saying "Why do people choose to subsidize?"
The Hitchhiker's Guide to the Galaxy illustrates neatly the problem of not understanding the question.
Can we just file this one in "where the hello have you been section" and move on?
The idea is that you legitimately can charge people based on their ability to pay. A small company gets the service much cheaper than a larger company because there are fewer seats.
Economicly there's nothing wrong with it.. A larger company can afford to pay more for the same service that a smaller company can; you're just normally allowed to discrimiate pricing.
Another justification is that w/ client-installed software, you can charge per seat.. But w/ server software there is only ever one seat.. Imagine if MS office could be installed as a web service.. There would literally only ever be a single copy of web-MS-office ever sold.. One guy with a big server and a public grant would install it, and the world could use it for free...
When http 1'st become popular, companies I worked for seriously had to consider the ramifications to webifying per-seat licensed software (as only a single webguy user would ever consume a seat). Databases that charged per seat, version control systems that charged per seat, etc.. These tools had intricate designs to more than the number of licensed users to actually use a system, but have a max concurrent useage. Single-web-user violated all of that. And consequently, many of those software packages have changed their licensing model.
Commercial databases, now, for example, are charged based on the expected number of users that are likely to serve (what use to be individual login licenses). A realistic metric would be to have some vender-specific benchmark, where higher numbers mean higher throughput and thus greater concurrent usage == higher price. However, it turned out that the most politically correct way to do it was to assume that each CPU could simultaneously serve a separate person (even though it really has nothing to do w/ anything), and thus that was a reliable metric for scaled processing power.
Again, all it is is the ability to charge small companies less than larger companies. Oracle needs to recoup their investment dollars somehow; they simply don't sell enough volume to charge $39.95 per copy, so they have to look at the economic horizon.. The law of deminishing returns.. The higher the price, the lower the number of sales... If you are limited to a single price point, then there is exactly one spot at any given time that represents a maximization of profit (or minimization of loss). BUT, if you can charge more for people that are willing to pay more (essentially having multiple price-points), you dramatically increase the profitability.
Now the whole point is slowly becoming moot, as whenever there is a big industry of over-priced server software, there is a group willing to fund an open-source effort to match the functionality.
-Michael
You could either charge some average pricing from everybody or try to charge more from those with higher demand in your product and less from those with lower demand. With the fixed average price you might completely lose the sales to the potential clients with lower demand while at the same time getting less money from those with high demand, so the average price is not what you want.
The problem is to find some rule or measurement for "how much demand for my product does potential customer x have?", ideally one that would not seem too arbitrary to the customer.
CAL are one solution for this problem, of course they are just an approximation but it's not the worst imaginable.
[i have an opinion and i am not afraid to use it]
They will do it the way that generates the most revenue. Our (the market's) behavior dictates the terms. Apparently this arrangement leads to the best bottom line. Why? General social conditions and perceptions would seem to indicate that a majority of consumers accept this notion as a reliable and predictable measure of the product's value. When one is confronted with the expense of an enterprise of elaborate dynamic scale, it is logical to break that system down in terms of units, like the bricks in a wall. People or "Users" are the most expensive units in a business, so its reasonable to equate other expense in terms of the personnel. In fact, this is an arbitrary correlation, and yet, it is one that can be understood by anyone, no matter what their experience. Furthermore, it establishes the concept that each and every individual is accountable to the price of the software - no one is exempt from the liability of existence. We all buy into the notion of indebtedness, and this fuels our willingness to be sucked dry at all times. Democracies lend themselves to this notion, even if they never actually vote, and even though some animals are more equal than others. This illusion appeals to our self projected individuality as fundamentalism, even if its inaccurate, subjective, or mythical.
In the event that the economy gets tough, eventually, there can be a higher possibility that the per "head or tail" (per coin) usage as defined by each and every individuals' CPU (heads) or seat (tails) will potentially violate that licensing agreement as the population grows; Its like taxation: all of the people involved can then be held accountable on some level. Duck, duck, goose....
This is just a little bit of Darwinian insurance in the event of real scarcity. Its a shotgun aimed at a flock of geese. Someone is going to feed the hungry hunter; it's just a matter of natural selection.
That was for a DEC Rainbow in 1981. My optical mouse probably has more storage today.
Got time? Spend some of it coding or testing
Ok, so I'm exaggerating a bit... but IRL, I'd guess something like 5%-30% of the call-out frequency, depending up on the application. By that I mean that if I'm supporting an MS-Windows server (rare), I'd average a problem requiring intervention maybe once a month on average; supporting anything else with a similar workload would be circa once a year, average. Some (like the few remaining OpenServer installations) are typically harder to fix than others, but nothing breaks as often as Microsoft stuff.
For workstations, the ratio is even worse.
First of all, because you can install random crap from the Internet on MS-Windows (unless the workstation is so locked-down that the users do stuff like bring in their own laptops to get stuff done, and cause call-outs with those)), people do. On Linux, you can just point them at a package manager and let them have at it from a list of thousands of safe packages if they have a driving need to install more stuff.
Second off, viruses and spyware for Linux are essentially nonexistent.
Third off is the background instability (avoided only by the most careful and dedicated and lucky installations) of MS-Windows.
Fourth off, MS-Windows suffers from a higher incidence of "experts" willing to "fix" things for free. I prefer to pronounce it like a New Zealander: "willing to fux things for free".
Got time? Spend some of it coding or testing
That would be NOVELL.
Well, that one's good enough for most money grubbing s/w companies. Why do you need more?
.. paranoid crackpot leftover from the days of Amiga.
shouldn't that be wulling
Does a Christian soccer team even need a goalkeeper?
Microsoft's house is built, not on sand, but on Twinkies. So now you know.
In order to eat Novell's lunch, Windows NT Server was an unlimited client license server. However, it cost something like $2400 PER SERVER. A few things happened in the ensuing years, chief among them the popularity of the web server. With a single file server on a LAN, you could server maybe a few hundred users at once. But that same server serving HTTP requests over the Internet (with slower connections per client) could server a hell of a lot more clients. Microsoft saw this as a breach of faith of the unlimited license model. Additionally, data driven websites using SQL Server as a web application server were also missing out on extra licensing dollars. So Microsoft changed the model. It's the same reason they limited the "desktop" versions of Windows to 10 file server connections - to prevent you from running the workstation version as a server. BTW, Windows is a lot cheaper now for the server license, but the CAL's are now a la carte.
Lastly, even DEC has been using CPU power (and quantity) for licensing - it's been done for decades, and now you're hearing about it on "smaller iron." This is normal sell-side trickle-down economics. DEC used VUP's - VAX Unit of Performance for licensing their apps.
A site with 10,000 clients is going to be more demanding on the support team (who often work for 'free' from the customer's point of view) than a site with 10 or even 100 clients.
Copyrights are nothing to do with free markets and "what the market will bear". They are a monopoly. It's like askking "why is coke free to begin then they jack up the price?" The answer ISN'T "because that is what the market will bear".
Bloody basstards.
all your points are based on bad administration or 3rd party software not the OS. 1. No user should have admin rights to install any software they want. if you let linux users install anything, your just hoping they are too stupid to install things you don't want. 2. this is mostly because of popularity. if linux is more popular it'll happen more. also if your network has centralized virus management, a network firewall, and proper user rights you're pretty safe. (and all networks should) 3. Instability in XP usually occurs because of 3rd party drivers and software. I've used linux on my laptop and i've suffered from buggy 3rd party software/drivers as well. 4. Don't let the "experts" fix anything. I don't let anyone user know the admin password. just remove the temptation. I gave my dad who's 60 a windows 2000 pc a few years ago. I installed it with everything he needed and put him on user rights. He's had no infections and no spyware. and he can do everything he wants. If it was today i'd give him an ubuntu system since it's free and user friendly.