HA is always crapshoot/tradeoff between cost and risk. Throw enough $ at the problem and you'll approach 100% availability.
I know that 'more robust' is a nice thing to want, but you really need to think about what you really need. If it takes 15 minutes to switch over to a backup copy (using some magic RAID disk mirroring maybe?) and 15 minutes to restart the app and let it checkpoint it's way up to a decent operational speed again, is that good enough?
If it takes an hour, how about that?
How much time/heartache or money is it worth for you to have system downtime, and how much are you willing to expend to reduce it by 5, 15, 30 minutes?
So, there's really a continuum of availabilty you have to pick your point in. At the low end, you have no backups and recreate everything from scratch. At the high end you use Vendor X's real clustering solution and 24x7 monitoring, then have zero downtime even in a disaster. Somewhere in the middle is you.
Now I realise this an overtly commercial view of things, but if needs be replace money with effort and season to taste.
Actually, you know you're anal when 1) you have a pet peeve about lanuage usage 2) you turn it into a sig 3) you get frustrated at the 160ch limit because it means you can't put in the other valid use of 'effect' as a verb...:)
Add a pinch of "anyone can use a web browser", Stir in a dash of "we need lots of training for a 'proper' GUI", Season with "GUI application support costs us money" Then Bake until golden brown.
My problem is that I have yet to come across a browser-based interface for anything more complex than 'One-click shopping' that doesn't make me want to beat the designer over the head with a bat.
Anyone used Domino.doc - IBM's pre-eminent document management system? That'll show you pain. How about webmail on any mailbox with more than about 20 messages (no, I'm not talking about deleting all the spam, I'm talking about finding your expense claim or that document you were sent five weeks ago)?
Except for all but the simplest of tasks, I really can't see the 'browser based' model as a good solution. That said, when you're in marketing, everything is a simple task. "Sell a product", "provision a mobile phone", "roll out a new server". See how easy it is, it's only a few words!
I can't agree more here. The only difficulty with this approach is that in a service industry the number of requests typically far outweigh the number of times or emails you can take to your boss.
If you make sure that everything you are given is allocated a priority, though, then you'll be getting well ahead in the game. The key thing is to define in black and white what those priorities mean
Once you have agreement on a set of PUBLISHED priority definitions, almost nobody will argue with you when you tell them that their request will be performed AFTER some other request. What's more, if they complain you can simply direct them to your manager for an exception (raise the priority based on an ad-hoc decision).
For example: Critical = More than one employee/system unable to perform their primary business tasks. No workaround is available.
Very High = One employee/system is unable to perform their primary business tasks; OR More than one employee/system unable to perform their primary business tasks but a workaround is available.
High = One employee/system is unable to perform their primary business tasks but a workaround is available; OR More than one employee/system unable to perform their day-to-day business tasks and no workaround is available.
Medium = Employees or systems are unable to perform their day-to-day business tasks.
I'm a big fan of the SFX, and yes, there are landmark scenes, but surely the 'influence' of Computer science on movies goes well beyond what you see on screen.
For example:
- Computer based editing suites for storing and editing sequences together.
- Motion capture cameras for unprecedented reality in later compositing.
- How about Word Processors for on-the-fly script revisions?
In summary, Technical and Social issues are inextricably linked, and what you're really looking for is a group of people to take on a governmental role for your website(s).
Definitely not worth whining about to anyone other than your provider's customer service dept
Start by requesting official 'notification' of the change under s34 of the ToS (ie: get it documented somewhere). If they won't document it for you, then document the conversation, with the manager's name yourself in a letter.
You then have the right to quit or change providers under the same s34 of your ToS (which gives you the right to terminate following any amendment which is unacceptable to you).
Not only that, but if they are forced to notify all subscribers then you may get a bit of a backlash happening too.
Of course,the alternative of a letter of complaint to them explaining exactly how you believe that your upload was 'fair and equitable use' of their network might get some results too.
Remember, if you're ever worried about the legality of things it's time to start collecting proof in the form of correspondance. Compare the ToS you agreed to with the current one. Any changes? Document their official responses. If they're as badly managed as you hope then they will contradict themselves at some point.
SMS>Since IANAL, and 4 this U need AL, U should get AL.<\summary>
Seriously though, you're asking about the details of copyrights, sub-licensing and contract law on a site, which you have to admit, has shown it's opinion on this matter pretty strongly before.
If you are being pressured into buying an SCO license, how about you ask those that are pressuring you what exactly you are licensing? what rights does it grant you and how may those rights be revoked?
Isn't this is the point that most people who are against all 'IP' style rights make. Why protect some things and not all things? Where do we draw the line?
I guess the Reasonable Man will tell you what is Obvious...
In this case I would assume that they could not claim rights to the word (as posted above it's the Japanese word for Green), but to the presentation/logo of the word/mark.
For example, if you check out one of Australia's largest banks, The Commonwealth Bank you will see that the 'mm' characters are glued together in the word. The word commonwealth can't be copyrighted, but the logo when the 'mm' is glued together like that can.
Actually, this is the first SCO article in a while which has made me think of the case in a new light... refreshing for a rehashed story like this.
It seems that SCO are saying that the issue is not actually about copyrighted code being in Linux at all. The issue is about IBM putting it there in contravention of their contract to "keep it secret, keep it safe".
However, I understand that IBM's linux teams and the AIX teams were pretty seperate for that specific reason - no cross pollination. So, SCO is saying that algorithms, solutions and ideas are the problem, not code.
<irony>Luckily this area of legal rights on ideas, concepts and algorithms is really clear in the US legal system.</irony>
clips from the article: This lawsuit is about breach of contract and other tort claims. It is not about copyright infringement.
SCO-Caldera being able to prove that IBM-developed AIX code... are derived works under the Unix licenses is the critical and key issue to SCO proving that IBM breached the Unix license agreements....the Unix license prohibits IBM from disclosing Unix Software Product code, methods, secrets, and so forth to third parties.Simply put, if SCO-Caldera can prove that IBM-developed AIX code... are derivate works and therefore part of the Unix Software Product and that IBM disclosed the code, methods, secrets and for them to the Linux developers, then SCO wins its IBM lawsuit.
The embedded space varies widely, ranging from 400Mhz PDAs with 64MB RAM to 50MHz mobile phones with 1 MB RAM.
It constantly surprises me how powerful the systems are that are defined as 'embedded'. After all, the minimum spec for DOOM is a 386 processor operating at a minimum of 33MHz and for Quake it's an Intel Pentium(R) 75 MHz processor or better.
Guy can move rocks from A to B by rocking and/or rolling.
Guy doesn't really know how the rocks were lifted 60ft into the air, but suggests 'shimmying' may have been part of it.
Next week - Guy explains that he can fly to the moon using only pogo sticks. "See, how bouncy they are," he says, "now just bounce a little higher and I'm there."
What a pointless, informationless article that was.
Re:Responsible Service of Alcohol
on
Another Beer Please
·
· Score: 2, Funny
Obviously it looks at your credit card's RFID and cross matches it with the last time you bought curry or McD's at 2am, comparing with how many you had that night!
What you really should be asking, is does your SO get access to the logs to see how much you REALLY had to drink and where?
My point is, you can assume that every hour on the hour of every day it is midnight somewhere. Additionally you have some weird timezones on the half hour too. The timezone your machine is in is GMT (UTC). It does not have timezones.
What sort of thing is it that needs to be done every half hour of every day on the one unix box?
I assume that you have a bunch of users, each of whom is assumed to be in a given timezone. You need to do something for each user at midnight in their timezone and you're worried about the management overhead of everyone moving in and out of daylight savings.
Take a step back and think about the real problem.
For example: Assume you're a webhost and you are implementing bandwidth limits on a per day basis. - why not pro-rate them to every hour? - why not define a day as GMT? - why not ignore daylight savings? - why not define the switchover at 4 or 5am and hence ignore daylight savings? - why not review the aggregates MORE frequently (ie: every 30mins) and do your 'action' as soon as the breach occurs?
Don't get hung up on the technical solution if an easier one appears.
Note that in most juristictions daylight savings changeovers happen at 3 or 4 am rather than midnight, so there are always 365 or 366 midnights in a year. Also, governments have the power to change daylight savings etc on a whim. There is going to be no zero-maintenance solution.
One unix box. Multiple timezones. Evidently a different script needs to be run for each timezone, since otherwise you'd just run your script every half hour on the half hour...
Why not set up a set of groups with defined IDs (say maybe 1000 - 1047) for each timezone. Allocate users in the appropriate timezone to those groups
Run a master script every hour (or half hour) which su's to a dummy member of that group and runs some script. you get some protection from accidentally breaking things by running as root.
Care to tell us exactly what the obvious solutions don't do? what you're trying to do maybe?
If that were the case, I'd imagine that people would volunteer to be on call, in order to get more $.
Agreed. This is largely the reason why the 'swaps' were handled fairly amicably. It's easier to find someone who's willing to take on the job if there's a payoff and not just a drawback.
I'm salaried too (we all were). I don't mind putting in some extra hours for special occasions (ie: your datacentre move), but this was on-call support for a production billing system - 24x366. There's a big difference between those two things.
I'd say Mod this guy/gal/chicken up, but it's already at 5...
He's completely correct; in most cases the actual Cost/Benefit ratio decision is not yours to make Your boss is paid to make those choices, you are paid to tell him/her/it what compromises each choice makes.
If you're in the position of making the compromise, you will be documenting everything and discussing it with the people with the money (ie: clients, upper management or investors).
Slashdot Mirror
HA is always crapshoot/tradeoff between cost and risk. Throw enough $ at the problem and you'll approach 100% availability.
I know that 'more robust' is a nice thing to want, but you really need to think about what you really need. If it takes 15 minutes to switch over to a backup copy (using some magic RAID disk mirroring maybe?) and 15 minutes to restart the app and let it checkpoint it's way up to a decent operational speed again, is that good enough?
If it takes an hour, how about that?
How much time/heartache or money is it worth for you to have system downtime, and how much are you willing to expend to reduce it by 5, 15, 30 minutes?
So, there's really a continuum of availabilty you have to pick your point in. At the low end, you have no backups and recreate everything from scratch. At the high end you use Vendor X's real clustering solution and 24x7 monitoring, then have zero downtime even in a disaster. Somewhere in the middle is you.
Now I realise this an overtly commercial view of things, but if needs be replace money with effort and season to taste.
Actually, you know you're anal when :)
1) you have a pet peeve about lanuage usage
2) you turn it into a sig
3) you get frustrated at the 160ch limit because it means you can't put in the other valid use of 'effect' as a verb...
+1 Troll to the parent :)
As you said, "marketroids"...
Add a pinch of "anyone can use a web browser",
Stir in a dash of "we need lots of training for a 'proper' GUI",
Season with "GUI application support costs us money"
Then Bake until golden brown.
My problem is that I have yet to come across a browser-based interface for anything more complex than 'One-click shopping' that doesn't make me want to beat the designer over the head with a bat.
Anyone used Domino.doc - IBM's pre-eminent document management system? That'll show you pain. How about webmail on any mailbox with more than about 20 messages (no, I'm not talking about deleting all the spam, I'm talking about finding your expense claim or that document you were sent five weeks ago)?
Except for all but the simplest of tasks, I really can't see the 'browser based' model as a good solution. That said, when you're in marketing, everything is a simple task. "Sell a product", "provision a mobile phone", "roll out a new server". See how easy it is, it's only a few words!
I can't agree more here. The only difficulty with this approach is that in a service industry the number of requests typically far outweigh the number of times or emails you can take to your boss.
If you make sure that everything you are given is allocated a priority, though, then you'll be getting well ahead in the game. The key thing is to define in black and white what those priorities mean
Once you have agreement on a set of PUBLISHED priority definitions, almost nobody will argue with you when you tell them that their request will be performed AFTER some other request. What's more, if they complain you can simply direct them to your manager for an exception (raise the priority based on an ad-hoc decision).
For example:
Critical = More than one employee/system unable to perform their primary business tasks. No workaround is available.
Very High = One employee/system is unable to perform their primary business tasks; OR More than one employee/system unable to perform their primary business tasks but a workaround is available.
High = One employee/system is unable to perform their primary business tasks but a workaround is available; OR More than one employee/system unable to perform their day-to-day business tasks and no workaround is available.
Medium = Employees or systems are unable to perform their day-to-day business tasks.
etc.
I'm a big fan of the SFX, and yes, there are landmark scenes, but surely the 'influence' of Computer science on movies goes well beyond what you see on screen.
For example:
- Computer based editing suites for storing and editing sequences together.
- Motion capture cameras for unprecedented reality in later compositing.
- How about Word Processors for on-the-fly script revisions?
Actually, I just this morning read this article by Clay Shirky which covers your points exactly.
In summary, Technical and Social issues are inextricably linked, and what you're really looking for is a group of people to take on a governmental role for your website(s).
...but few people use it that way, and slashcode crops me to 160ch anyway :)
My point exactly. If it was fair, equitable and legal then proving so or explaining why should be a good enough reason to frce them to apologise.
:)
Otherwise bad freakin' luck!
Definitely not worth whining about to anyone other than your provider's customer service dept
Start by requesting official 'notification' of the change under s34 of the ToS (ie: get it documented somewhere). If they won't document it for you, then document the conversation, with the manager's name yourself in a letter.
You then have the right to quit or change providers under the same s34 of your ToS (which gives you the right to terminate following any amendment which is unacceptable to you).
Not only that, but if they are forced to notify all subscribers then you may get a bit of a backlash happening too.
Of course,the alternative of a letter of complaint to them explaining exactly how you believe that your upload was 'fair and equitable use' of their network might get some results too.
Remember, if you're ever worried about the legality of things it's time to start collecting proof in the form of correspondance. Compare the ToS you agreed to with the current one. Any changes? Document their official responses. If they're as badly managed as you hope then they will contradict themselves at some point.
SMS>Since IANAL, and 4 this U need AL, U should get AL.<\summary>
Seriously though, you're asking about the details of copyrights, sub-licensing and contract law on a site, which you have to admit, has shown it's opinion on this matter pretty strongly before.
If you are being pressured into buying an SCO license, how about you ask those that are pressuring you what exactly you are licensing? what rights does it grant you and how may those rights be revoked?
ie: Ask a lawyer. Don't ask slashdot.
Why not map that name to a dud address too?
I assume you ownzor the DNS that client PCs will use!
Isn't this is the point that most people who are against all 'IP' style rights make. Why protect some things and not all things? Where do we draw the line?
I guess the Reasonable Man will tell you what is Obvious...
The TM can be for the logo, symbol, wording etc.
In this case I would assume that they could not claim rights to the word (as posted above it's the Japanese word for Green), but to the presentation/logo of the word/mark.
For example, if you check out one of Australia's largest banks, The Commonwealth Bank you will see that the 'mm' characters are glued together in the word. The word commonwealth can't be copyrighted, but the logo when the 'mm' is glued together like that can.
Yeah, I know I was actually being allanic. (It's not ironic, but you'd think it was if you were a dozy canadian bint)
Realised as soon as I posted.
OpenGL is, as the name suggests, a Graphics Language, not a sound, network, input, pay-to-play platform.
Yes, MS did something right for games developers but OpenGL is a different kettle of Trolls.
Actually, this is the first SCO article in a while which has made me think of the case in a new light... refreshing for a rehashed story like this.
... are derived works under the Unix licenses is the critical and key issue to SCO proving that IBM breached the Unix license agreements. ...the Unix license prohibits IBM from disclosing Unix Software Product code, methods, secrets, and so forth to third parties.Simply put, if SCO-Caldera can prove that IBM-developed AIX code ... are derivate works and therefore part of the Unix Software Product and that IBM disclosed the code, methods, secrets and for them to the Linux developers, then SCO wins its IBM lawsuit.
It seems that SCO are saying that the issue is not actually about copyrighted code being in Linux at all. The issue is about IBM putting it there in contravention of their contract to "keep it secret, keep it safe".
However, I understand that IBM's linux teams and the AIX teams were pretty seperate for that specific reason - no cross pollination. So, SCO is saying that algorithms, solutions and ideas are the problem, not code.
<irony>Luckily this area of legal rights on ideas, concepts and algorithms is really clear in the US legal system.</irony>
clips from the article:
This lawsuit is about breach of contract and other tort claims. It is not about copyright infringement.
SCO-Caldera being able to prove that IBM-developed AIX code
The embedded space varies widely, ranging from 400Mhz PDAs with 64MB RAM to 50MHz mobile phones with 1 MB RAM.
It constantly surprises me how powerful the systems are that are defined as 'embedded'. After all, the minimum spec for DOOM is a
386 processor operating at a minimum of 33MHz and for Quake it's an Intel Pentium(R) 75 MHz processor or better.
That's now in 'embedded systems' sizing easily.
Guy can move rocks from A to B by rocking and/or rolling.
Guy doesn't really know how the rocks were lifted 60ft into the air, but suggests 'shimmying' may have been part of it.
Next week - Guy explains that he can fly to the moon using only pogo sticks. "See, how bouncy they are," he says, "now just bounce a little higher and I'm there."
What a pointless, informationless article that was.
Obviously it looks at your credit card's RFID and cross matches it with the last time you bought curry or McD's at 2am, comparing with how many you had that night!
What you really should be asking, is does your SO get access to the logs to see how much you REALLY had to drink and where?
My point is, you can assume that every hour on the hour of every day it is midnight somewhere. Additionally you have some weird timezones on the half hour too. The timezone your machine is in is GMT (UTC). It does not have timezones.
What sort of thing is it that needs to be done every half hour of every day on the one unix box?
I assume that you have a bunch of users, each of whom is assumed to be in a given timezone. You need to do something for each user at midnight in their timezone and you're worried about the management overhead of everyone moving in and out of daylight savings.
Take a step back and think about the real problem.
For example: Assume you're a webhost and you are implementing bandwidth limits on a per day basis.
- why not pro-rate them to every hour?
- why not define a day as GMT?
- why not ignore daylight savings?
- why not define the switchover at 4 or 5am and hence ignore daylight savings?
- why not review the aggregates MORE frequently (ie: every 30mins) and do your 'action' as soon as the breach occurs?
Don't get hung up on the technical solution if an easier one appears.
Note that in most juristictions daylight savings changeovers happen at 3 or 4 am rather than midnight, so there are always 365 or 366 midnights in a year. Also, governments have the power to change daylight savings etc on a whim. There is going to be no zero-maintenance solution.
One unix box. Multiple timezones. Evidently a different script needs to be run for each timezone, since otherwise you'd just run your script every half hour on the half hour...
Why not set up a set of groups with defined IDs (say maybe 1000 - 1047) for each timezone. Allocate users in the appropriate timezone to those groups
Run a master script every hour (or half hour) which su's to a dummy member of that group and runs some script. you get some protection from accidentally breaking things by running as root.
Care to tell us exactly what the obvious solutions don't do? what you're trying to do maybe?
If that were the case, I'd imagine that people would volunteer to be on call, in order to get more $.
Agreed. This is largely the reason why the 'swaps' were handled fairly amicably. It's easier to find someone who's willing to take on the job if there's a payoff and not just a drawback.
I'm salaried too (we all were). I don't mind putting in some extra hours for special occasions (ie: your datacentre move), but this was on-call support for a production billing system - 24x366. There's a big difference between those two things.
I'd say Mod this guy/gal/chicken up, but it's already at 5...
He's completely correct; in most cases the actual Cost/Benefit ratio decision is not yours to make Your boss is paid to make those choices, you are paid to tell him/her/it what compromises each choice makes.
If you're in the position of making the compromise, you will be documenting everything and discussing it with the people with the money (ie: clients, upper management or investors).
If you're genuinely as useful as you claim, management would certainly be receptive
It's glib, but anytime you think you're indispensable, stick your hand in a bucket of water and when you pull it out see how big a hole it leaves.
Remember that when you walk in to the office.
Just because HR have a policy doesn't mean they won't make an exception if your boss is supportive of it
OTOH - why don't you apply for the new hire positions if they're worth more money?????