Let me recap for you (both of the below points taken from the links I provided...):
1) Proposed by the physicist Miguel Alcubierre, popularised by Star-Trek.
2) Proposed by the physicist Robert W Bussard (hence "Bussard Ramjet"), popularised by Larry Niven (the author), and even referred to by Carl Sagan on TV and in books...
Various other authors have used the same ideas. Perhaps I ought to have mentioned that I'm a physicist too... And the gentle humour regarding tense was supposed to clue you in that I wasn't suggesting we had a practical solution just yet... I wish I'd spelt "two thoughts" correctly, though.
After reading the article (yeah, I know...) tow thought spring to mind...
1) Warp drive doesn't posit a traditional "go-very-fast-through-normal-space" type of spacecraft engine - it warps[*] space-time (hence the name!) in front of and behind the spacecraft - see here for an explanation. The spacecraft itself is sitting in a bubble of normal space, possibly even at rest.
See This journal entry I posted a while back...
These days, at least in the US, I'd probably be up on federal wiretap charges or something. Back then, it was serious enough that they'd threaten to throw me out of college, but I never got any sense of there being jail-time involved...
I really don't get this. I sit in front of an LCD monitor all day, every day, and read and/or type away. I touch-type, so even if I'm typing, I'm looking at the screen. I don't get tired eyes, I don't have a problem reading for hours on end.
I went into the local bookstore and saw one of these e-ink readers (made by Sony, I think), and I thought the display was truly awful - blurry, low-contrast, and far-and-away more difficult to read.
Perhaps it has to do with the different ways that OSX and Windows (which most people are used to) put type on the screen. OSX tries to mimic print layout as much as possible. Windows attempts to line up to the pixel matrix on the screen. I'm not saying Windows is wrong in this - it's a matter of preference - but I far prefer the smooth characters on an OSX display.
I guess it's "common knowledge" that eInk is "better", but I'm not seeing it. I wonder if it's just common knowledge that eInk is better than Windows LCD displays ?
If you think you can't do *art* on a black and white screen, then you're the one who is artistically challenged, but that wouldn't surprise me since you appear to be unable to read.
The post I replied to was:
This is incorrect.
Even before Apple launches the Mac, I ran a graphical environment much like Windows under CP/M-86, it was called Gem and was used by amiong others, the British computer Apricot. It was made by Digital Research, the same company that made CP/M
... which is untrue. Which is what I stated.
FWIW, I never had a Mac back then, I had an Atari ST, which ran GEM, and which had a fantastic mono screen...
"Xerox had invested in Apple (ie, Apple had given Xerox Board members stock in exchange for access to the research performed at PARC) and had invited the Macintosh design team to view their GUI computers at the PARC research lab; these visits had been very influential on the development of the Macintosh GUI"
Gem was first launched in 1985, whereas the Mac was launched in 1984. Apple actually sued Digital Research (and won) because it was such a blatant copy of the Mac's interface.
Simon
Not the case here in CA. It does vary (seemingly depending on the supplier), but Chevron stations always price their diesel lower than all other kinds of gas...
My car is a rag-top, so of course it takes the high-octane gas. The difference between mine and my wifes (hers is the diesel) prices per gallon is ~25 to 30 cents/gallon.
No, I'm from the UK. Been living in CA for the last 5 years or so. All them there gallons are pure ole US gallons. No 'mpeerial thingamubobs around here, no siree:)
just gone out and bought a VW diesel TDI 2010 model. It's as quiet as a petrol engine when running, and only (very) slightly louder than a petrol engine while idling. The exhaust is very clean...
As for efficiency, the TDI is currently averaging 49 miles/gallon for the sportwagen, that's real honest-to-goodness driving on both freeway and city streets, and is ~7mpg higher than the official rating of the car.
I didn't quite believe it, so I did the calculation myself based on mileage and purchased fuel, and my figures came to 52 miles/gallon. If anything, the car is under-reporting the fuel economy. Not to mention that diesel is actually cheaper per gallon than unleaded.
Note that the jetta sportwagen is the identical size to the normal jetta (it's just a different top), and that the engine is only a 2.0 litre engine, smaller than the 2.5 litre base jetta engine.
In short, I don't think you could actually be any more wrong about diesel engines.
I'm running several macs, both at home and at work, and the only time I've ever run an anti-virus on any of them was at the request of my ISP last month - there was a report of a virus originating from my home IP address. I downloaded and ran the latest ClamAV, and of course there was no virus on the machine, it was a spoofed IP address...
Over the past 5 years, that's the only time I've ever run a virus check. It came up with 0 viruses. I conclude that the likelihood of me getting a virus on a mac is still small compared to my XP box, which every time I run a virus check flags *something* new as wrong/suspicious. Sometimes I can even tell if the something is innocuous or dangerous...
Slashdot likes to say that anecdotal evidence is meaningless (which of course it is), but when a sufficiently large collection of anecdotes all say the same thing, we call that consensus. The general consensus is (I believe) that Macs are a lot less likely to be infected than Windows boxes, so your 'Anyone who uses any computer (including Mac AND Linux) without anti-virus is asking for what they get' statement is in fact news to me.
Linux though, despite the crappiness of its DEs, still beats everything, for its total and complete freedom, and scriptability. You can actually *use* your computer with Linux. For what it was invented for: To *automate* *your* work.
I don't understand this. What can *you* do on your linux box to automate your work, that *I* can't do on my Mac to automate my work ?
It's not as though you can't just compile the source code on a Mac, after all. I've never had a problem with./configure --prefix=/opt && make && make install...
In fact this evening, I ran across what http://www.elektor-usa.com/>elektor refer to as a hexadoku, which is basically a large (16x16 rather than 9x9 grid) version of Soduku. Since there's a prize worth having, I knocked up a solver for any starting state using ObjC / Cocoa. It took me a couple of hours (probably less than it would take to solve the puzzle!) to write, and that's because of the excellent coding interface the Mac has.
I've written fairly complex (a video streaming client) apps in gtk2 and it's nowhere near as easy to code with as Cocoa. Nor does it look as pretty, but that's not why I like Cocoa - it's truly easy to link interfaces and code together, whilst keeping the layout separate from the code, and with no fugly UI setup code autogenerated in your app. Cocoa rocks, and (for me) that's why OSX is the best damn unix workstation I've ever used - and I've used a *lot* of them...
I hear a lot about how OSX restricts your freedom, but I don't encounter it. The developer environment is almost lick-ably good, the machines (well, my Mac Pro) are powerful, it runs pretty much any open source s/w (since it's got a BSD userland), and I can run things like Aperture or Final Cut Pro on it. I've coded on and sold businesses based around linux servers. I *like* linux, but for a desktop workstation, nothing touches the Mac. For me. As a coder and user.
OSX uses the xnu kernel (a derivative of Mach). It is not based on BSD, and only provides a BSD userland to make things easier for developers/users. Xnu is open-source.
Having said that, a huge chunk of the user-visible runtime is not open-source, and Apple maintain an actively protective stance over it. I agree with the lawsuits comment...
Don't know much about macs, do you ? There are some legitimate gripes about newbie server admins on a Mac. Backup isn't one of them.
OSX (even the desktop version) comes with Time machine. Just plug a USB (or firewire) disk it hasn't seen before into one of the ports on the back and it will prompt you as to whether you want to use it to back up via TM.
I guess you could use one of the internal disks as well, if you don't run those as RAID.
20% cheaper, and half the computational power. The minis would have 16 processor cores, and the Dell only has 8. The Nehalem isn't that much faster than the C2D - see my post below...
The discussion was comparing blades to minis on the basis of cost, not density, but if you want to talk density...
Well, the cost argument works just as well for rackmount machines. An appropriately specced 1U rackmount is about $6.5k. 8 Mac Minis is about $8k.
I've just measured the rack in my garage - it's 30" deep by 19" across, and a mac mini fits handily into a 7" square by 4 holes in the rack with space to spare. 6U is 18 holes, so I can get 4 slide-out racks of minis into 6U, and 8 minis (including PSUs and wiring) per rack in a 2-across by 4 deep configuration. That's a total of 8x4x2 CPUs (because they're core-2 duos in the mini), or 64 CPUs.
Yes. I estimated you'd get about 8 Minis per 2U. You've measured it out at about 8 Minis per 1.5U - though I suspect you'd have trouble keeping the density you're talking about properly cooled, given the lack of high-volume forced cooling in the Mini. The ones at the upper rear of the rack, in particular, would suffer.
I'm getting less and less convinced about the need for keeping computers cool, and it seems I'm not alone. Certainly the linux boxes in my garage (which isn't at all cooled) have had no problems over the Summer here in the Bay Area. Those boxes suck a *lot* more power than the minis and run a *lot* hotter.
Actually, it works out extremely well. Those rackmount servers have Nehalem based CPUs, which are roughly twice as fast as Core 2 CPUs at the same clock speed. Even being conservative, the 48 cores in 6U of rackmount servers will be quite a bit faster than the 64 cores in the Minis. The same is true of using Xserves, though the price comparison probably won't be quite as favourable.
I include part of my other post below
Sometimes good things come in small packages. The Nehalem chips *are* better chips, but core-for-core they're nowhere near 2x as fast (let alone the ridiculous 10x as fast!).
I invite you to look at the mac-mini benchmarks (which are obviously the right ones:) and the MacPro2,1 benchmarks (which use the same processor, and run slightly faster than the blades above). Here's a summary of the relative ratios for different types of computing:
Integer. 7127 : 3162. E5405 is 2.25x
Floating point: 11849 : 4927. E4505 is 2.4x
Memory: 2752 : 2650. E4505 is 1.03x
Stream: 2062 : 1912. E4505 is 1.07x
Bear in mind that the E4505 above is a dual quad core, and the mini is just a dual-core when you compare these. Also bear in mind that they're weighted averages - in some of the individual tests, the E5405 is almost 4x the speed.
Given that the minis are matching (or beating) the number of cores you're talking about, I'd say the advantage is on the mini's side, not the 1U servers. Odd but true.
The thing is, a Mac Mini is not designed to be a 24/7 server. It also lacks numerous "standard" server features like power supply redundancy, remote management capabilities, and hot-swap hard disks. A Mini comes with a 1 year warranty standard, while even a low-end Dell server will come with a 3 year warranty. A Mini is difficult and time-consuming to service. A typical rackmount is trivial to break down and nearly any arbitrary component can be replaced in a matter of minutes.
It's only anecdotal, but I have had a mini running almost constantly 24/7 for a few years now, almost constantly under load since it does the image-processing for network cameras linked up inside the house and garage. Traditional servers are designed around performance and they run hot, therefore needing all this industrial design to keep them c
Sometimes good things come in small packages. The Nehalem chips *are* better chips, but core-for-core they're nowhere near 2x as fast (let alone the ridiculous 10x as fast!).
I invite you to look at the mac-mini benchmarks (which are obviously the right ones:) and the MacPro2,1 benchmarks (which use the same processor, and run slightly faster than the blades above). Here's a summary of the relative ratios for different types of computing:
Integer. 7127 : 3162. E5405 is 2.25x
Floating point: 11849 : 4927. E4505 is 2.4x
Memory: 2752 : 2650. E4505 is 1.03x
Stream: 2062 : 1912. E4505 is 1.07x
Bear in mind that the E4505 above is a dual quad core, and the mini is just a dual-core when you compare these. Also bear in mind that they're weighted averages - in some of the individual tests, the E5405 is almost 4x the speed.
It's the law of diminishing returns, you get a lot of bang-for-the-buck at the lower end, and you pay increasingly more as you get towards the upper end.
I just went to the page where you can configure the blades, and specced out an 8-core blade:
- 2x E5405 Blade )cheapest quad-core)
- 4GB of RAM (4G is needed for Windows 2008 server)
- 2x Cheapest drive. RAID is required for OS installation (!)
- Windows SBS 2008, standard edition
Comes to $2,857 per blade, or $27,085 for 8 of them in the cheapest enclosure. Compare that to 16 minis at $15,984. Ouch.
The discussion was comparing blades to minis on the basis of cost, not density, but if you want to talk density...
I've just measured the rack in my garage - it's 30" deep by 19" across, and a mac mini fits handily into a 7" square by 4 holes in the rack with space to spare. 6U is 18 holes, so I can get 4 slide-out racks of minis into 6U, and 8 minis (including PSUs and wiring) per rack in a 2-across by 4 deep configuration. That's a total of 8x4x2 CPUs (because they're core-2 duos in the mini), or 64 CPUs.
The blades pack 8 blades into 6U. The ones we started off discussing therefore put 8 CPUs into 6U. Whoosh. If we jump to the later offering, that's 8x4=32 cores in 6U, or 32 less than the minis. Only if we go to the 8-core blades (for a total of 4299+8*1610 for the E5405 CPUs, or $17179 before you add OS and disk storage) do you get the same density of 64 CPUs in 6U.
The largest number of cores I'm aware of in standard 1U is dual quad-core (soon to be dual 6-core I suppose), so in 6U you get 6x8 = 48 CPUs. That doesn't work out too well compared to 6U of minis (64 CPUs). There are half-width/depth servers around, and you can get dual quad-core in that form-factor, so I could see the density going up to 96 CPUs in 6U, but I'm not aware of pricing on them...
I'm not advocating that you build your next computing cluster out of mac minis. I'm saying they scale (in price) reasonably well, that's all. I assume that there's a design difference between a blade server and a more consumer-orientated server (though I've been using a mini as an entertainment hub for the last 2 years without a single problem).
Certainly it'll *look* more professional to have a blade server in the rack. Whether it'll perform better for the price is more up for debate.
The Core 2 Duo @ 2.53GHz is more than than twice as fast as the 1.8GHz Celeron, so really you only need 4 of the mini-servers to beat out 8 of the blades... Suddenly the price swings significantly to the mini's... ($4299+8*379 = 7331. 4x$999 = 3996, or roughly half the price) and you've still got 16GB of RAM, not 8GB (which is what the blades have)
Yeah, you have a point on the Celerons. I can't find anything exactly equivalent to the minimac hardware in these blades, but among the pre-configured options on that blade page is a quad-core Xeon for $609 with 2GB RAM. An extra 2GB RAM is +$129, so $738 for a quad core Xeon with 4GB RAM. Using your reasoning with the celeron, one of those xeon blades == 2 minimacs. If we load up an 8 slot chassis that's $4299+8*738 = $10,203 vs 16*999 = $15,984. Similar but with 8 cores & 8GB RAM per blade and 32 minimacs: $14,027 vs $31,968. As we approach "cramp LOADS of those things in a small space" the blades are looking way cheaper now as well as more space efficient.
What you're also ignoring here is the cost of the OS. You're paying $999 because the mini comes with the server-grade OS. You need to add on the cost of Windows Server 2008 per blade (I'll let you get away with the 5-CAL licences, even though OSX is unlimited), for a further cost (ouch!) of ~$1029 per blade. That takes your costs to $(4299 + 8*(738+1029)) = $18435, which is again more than the 16 minis ( =$15,984).
If the argument is that you can use linux rather than windows, well fine. You can do that on the minis too though. If we take this route you don't need the mini-server bundle, so you can get the $599 package and the costs compare as $(4299 + 8*738) = $10,203 vs the minis at 16x599 = $9,584. You still have to buy storage for your blades as well, whereas the minis come with hard-disks.
To a certain extent, this is tongue-in-cheek. I'm not sure I'd recommend anyone buy 32 mac-minis if they wanted a computing cluster for example, but if you're trying to do it on the cheap, they're not actually that bad...
individual cables: You need an ethernet wire and a power supply. That's one more wire per mini. If you're using external storage, you'll need a firewire cable or whatever, but that's the same as with the blades
One more? You don't need any cables going in to the blades. Power, network, console, etc are supplied by the backplane.
Well, actually I was assuming 6 independent psu's would take 6 wires, and 1 network cable would take 1 wire, for a total of 7, compared to 2 (power, network)x4 for a total of 8. That's not a huge difference:) The "per mini" was wrong, and left in from text I'd changed.
organization: The minis are running OSX server, so all the LOM, remote-desktop, remote console etc. server-admin tools work fine.
You can't install an OS, upgrade an OS, access firmware, toggle power or reset them that way. You can do all of that remotely with ILO and work with CD images over the network.
Actually, yes you can. You need to do some prior set-up work (have a netinstall image available or leave the OS DVD in the drive) but LOM (literally: "Lights-Out Management") lets you reboot, power-down and (later, if necessary) power-up, and the system management tools let you install software, choose where to boot from, and relay everything across the net to a client machine. I've done it before from the OS DVD I left for that purpose in the drive of an XServe. Upgrading software doesn't need anything more than ssh or vnc ("Screen sharing" as it's called on the mac) to use the GUI tools.
more-efficient energy: I meant in terms of scale. A few large power supplies will almost certainly be more efficient than many small ones.
The Core 2 Duo @ 2.53GHz is more than than twice as fast as the 1.8GHz Celeron, so really you only need 4 of the mini-servers to beat out 8 of the blades... Suddenly the price swings significantly to the mini's... ($4299+8*379 = 7331. 4x$999 = 3996, or roughly half the price) and you've still got 16GB of RAM, not 8GB (which is what the blades have)
I was using 8's because that's the unit of the enclosure. If you want to start using more blades, the 16-way enclosure is even more expensive, relatively, with prices for just the enclosure starting at $17k. Just multiply up to suit the numbers. I think you'll easily fit 4 mini's within a 6U rack-space though - you could probably fit them on a 2U tray...
To address your other concerns:
individual cables: You need an ethernet wire and a power supply. That's one more wire per mini. If you're using external storage, you'll need a firewire cable or whatever, but that's the same as with the blades
organization: The minis are running OSX server, so all the LOM, remote-desktop, remote console etc. server-admin tools work fine.
more-efficient energy: The enclosures list up to 6x 1200W PSU's for the enclosure, or 7,200W. They'll draw less than that usually, of course, but that's their capacity. The mini's run at 14W at idle according to apple. I've seen reports of 40W when it's working hard.
hot-swap stuff: The blades don't hot-swap, they have to be powered down and up again. They don't have hot-swap disks either. Neither does the mini of course - you're supposed to plug in external disks (which can be hot-swap, of course) for both.
Overall, the mini comes out looking pretty good:) In fact I'm thinking about getting rid of the XServe I have at home and getting a mini instead...
Slashdot Mirror
Did you look under [Edit] -> [Refactor] on the menu ? Or type 'refactor' into the help search ? Simon
Let me recap for you (both of the below points taken from the links I provided...):
1) Proposed by the physicist Miguel Alcubierre, popularised by Star-Trek.
2) Proposed by the physicist Robert W Bussard (hence "Bussard Ramjet"), popularised by Larry Niven (the author), and even referred to by Carl Sagan on TV and in books...
Various other authors have used the same ideas. Perhaps I ought to have mentioned that I'm a physicist too... And the gentle humour regarding tense was supposed to clue you in that I wasn't suggesting we had a practical solution just yet... I wish I'd spelt "two thoughts" correctly, though.
Simon
After reading the article (yeah, I know...) tow thought spring to mind...
/., I expect you all to forgive me for using the present tense here [grin]
1) Warp drive doesn't posit a traditional "go-very-fast-through-normal-space" type of spacecraft engine - it warps[*] space-time (hence the name!) in front of and behind the spacecraft - see here for an explanation. The spacecraft itself is sitting in a bubble of normal space, possibly even at rest.
2) Um, ramjets, anyone ?
Seriously, any number of sci-fi authors have covered this problem in enormous detail over the last few decades
Simon
[*] And because this is
See This journal entry I posted a while back... These days, at least in the US, I'd probably be up on federal wiretap charges or something. Back then, it was serious enough that they'd threaten to throw me out of college, but I never got any sense of there being jail-time involved...
Simon
Dude. Seriously. Learn to use XCode.
I really don't get this. I sit in front of an LCD monitor all day, every day, and read and/or type away. I touch-type, so even if I'm typing, I'm looking at the screen. I don't get tired eyes, I don't have a problem reading for hours on end.
I went into the local bookstore and saw one of these e-ink readers (made by Sony, I think), and I thought the display was truly awful - blurry, low-contrast, and far-and-away more difficult to read.
Perhaps it has to do with the different ways that OSX and Windows (which most people are used to) put type on the screen. OSX tries to mimic print layout as much as possible. Windows attempts to line up to the pixel matrix on the screen. I'm not saying Windows is wrong in this - it's a matter of preference - but I far prefer the smooth characters on an OSX display.
I guess it's "common knowledge" that eInk is "better", but I'm not seeing it. I wonder if it's just common knowledge that eInk is better than Windows LCD displays ?
Simon
Oh I see, you're just an apple-hater. Fine. Carry on, but that's the end of this conversation...
Simon
The post I replied to was:
FWIW, I never had a Mac back then, I had an Atari ST, which ran GEM, and which had a fantastic mono screen...
Simon
Simon
Gem was first launched in 1985, whereas the Mac was launched in 1984. Apple actually sued Digital Research (and won) because it was such a blatant copy of the Mac's interface. Simon
Not the case here in CA. It does vary (seemingly depending on the supplier), but Chevron stations always price their diesel lower than all other kinds of gas...
My car is a rag-top, so of course it takes the high-octane gas. The difference between mine and my wifes (hers is the diesel) prices per gallon is ~25 to 30 cents/gallon.
Simon
No, I'm from the UK. Been living in CA for the last 5 years or so. All them there gallons are pure ole US gallons. No 'mpeerial thingamubobs around here, no siree :)
Simon
just gone out and bought a VW diesel TDI 2010 model. It's as quiet as a petrol engine when running, and only (very) slightly louder than a petrol engine while idling. The exhaust is very clean...
As for efficiency, the TDI is currently averaging 49 miles/gallon for the sportwagen, that's real honest-to-goodness driving on both freeway and city streets, and is ~7mpg higher than the official rating of the car.
I didn't quite believe it, so I did the calculation myself based on mileage and purchased fuel, and my figures came to 52 miles/gallon. If anything, the car is under-reporting the fuel economy. Not to mention that diesel is actually cheaper per gallon than unleaded.
Note that the jetta sportwagen is the identical size to the normal jetta (it's just a different top), and that the engine is only a 2.0 litre engine, smaller than the 2.5 litre base jetta engine.
In short, I don't think you could actually be any more wrong about diesel engines.
Simon.
I'm running several macs, both at home and at work, and the only time I've ever run an anti-virus on any of them was at the request of my ISP last month - there was a report of a virus originating from my home IP address. I downloaded and ran the latest ClamAV, and of course there was no virus on the machine, it was a spoofed IP address...
Over the past 5 years, that's the only time I've ever run a virus check. It came up with 0 viruses. I conclude that the likelihood of me getting a virus on a mac is still small compared to my XP box, which every time I run a virus check flags *something* new as wrong/suspicious. Sometimes I can even tell if the something is innocuous or dangerous...
Slashdot likes to say that anecdotal evidence is meaningless (which of course it is), but when a sufficiently large collection of anecdotes all say the same thing, we call that consensus. The general consensus is (I believe) that Macs are a lot less likely to be infected than Windows boxes, so your 'Anyone who uses any computer (including Mac AND Linux) without anti-virus is asking for what they get' statement is in fact news to me.
Simon
I don't understand this. What can *you* do on your linux box to automate your work, that *I* can't do on my Mac to automate my work ?
./configure --prefix=/opt && make && make install...
It's not as though you can't just compile the source code on a Mac, after all. I've never had a problem with
In fact this evening, I ran across what http://www.elektor-usa.com/>elektor refer to as a hexadoku, which is basically a large (16x16 rather than 9x9 grid) version of Soduku. Since there's a prize worth having, I knocked up a solver for any starting state using ObjC / Cocoa. It took me a couple of hours (probably less than it would take to solve the puzzle!) to write, and that's because of the excellent coding interface the Mac has.
I've written fairly complex (a video streaming client) apps in gtk2 and it's nowhere near as easy to code with as Cocoa. Nor does it look as pretty, but that's not why I like Cocoa - it's truly easy to link interfaces and code together, whilst keeping the layout separate from the code, and with no fugly UI setup code autogenerated in your app. Cocoa rocks, and (for me) that's why OSX is the best damn unix workstation I've ever used - and I've used a *lot* of them...
I hear a lot about how OSX restricts your freedom, but I don't encounter it. The developer environment is almost lick-ably good, the machines (well, my Mac Pro) are powerful, it runs pretty much any open source s/w (since it's got a BSD userland), and I can run things like Aperture or Final Cut Pro on it. I've coded on and sold businesses based around linux servers. I *like* linux, but for a desktop workstation, nothing touches the Mac. For me. As a coder and user.
Simon
OSX uses the xnu kernel (a derivative of Mach). It is not based on BSD, and only provides a BSD userland to make things easier for developers/users. Xnu is open-source.
Having said that, a huge chunk of the user-visible runtime is not open-source, and Apple maintain an actively protective stance over it. I agree with the lawsuits comment...
Simon.
Don't know much about macs, do you ? There are some legitimate gripes about newbie server admins on a Mac. Backup isn't one of them.
OSX (even the desktop version) comes with Time machine. Just plug a USB (or firewire) disk it hasn't seen before into one of the ports on the back and it will prompt you as to whether you want to use it to back up via TM.
I guess you could use one of the internal disks as well, if you don't run those as RAID.
Simon.
20% cheaper, and half the computational power. The minis would have 16 processor cores, and the Dell only has 8. The Nehalem isn't that much faster than the C2D - see my post below...
Simon
The discussion was comparing blades to minis on the basis of cost, not density, but if you want to talk density...
Well, the cost argument works just as well for rackmount machines. An appropriately specced 1U rackmount is about $6.5k. 8 Mac Minis is about $8k.
I've just measured the rack in my garage - it's 30" deep by 19" across, and a mac mini fits handily into a 7" square by 4 holes in the rack with space to spare. 6U is 18 holes, so I can get 4 slide-out racks of minis into 6U, and 8 minis (including PSUs and wiring) per rack in a 2-across by 4 deep configuration. That's a total of 8x4x2 CPUs (because they're core-2 duos in the mini), or 64 CPUs.
Yes. I estimated you'd get about 8 Minis per 2U. You've measured it out at about 8 Minis per 1.5U - though I suspect you'd have trouble keeping the density you're talking about properly cooled, given the lack of high-volume forced cooling in the Mini. The ones at the upper rear of the rack, in particular, would suffer.
I'm getting less and less convinced about the need for keeping computers cool, and it seems I'm not alone. Certainly the linux boxes in my garage (which isn't at all cooled) have had no problems over the Summer here in the Bay Area. Those boxes suck a *lot* more power than the minis and run a *lot* hotter.
Actually, it works out extremely well. Those rackmount servers have Nehalem based CPUs, which are roughly twice as fast as Core 2 CPUs at the same clock speed. Even being conservative, the 48 cores in 6U of rackmount servers will be quite a bit faster than the 64 cores in the Minis. The same is true of using Xserves, though the price comparison probably won't be quite as favourable.
I include part of my other post below
:) and the MacPro2,1 benchmarks (which use the same processor, and run slightly faster than the blades above). Here's a summary of the relative ratios for different types of computing:
Sometimes good things come in small packages. The Nehalem chips *are* better chips, but core-for-core they're nowhere near 2x as fast (let alone the ridiculous 10x as fast!).
I invite you to look at the mac-mini benchmarks (which are obviously the right ones
Bear in mind that the E4505 above is a dual quad core, and the mini is just a dual-core when you compare these. Also bear in mind that they're weighted averages - in some of the individual tests, the E5405 is almost 4x the speed.
Given that the minis are matching (or beating) the number of cores you're talking about, I'd say the advantage is on the mini's side, not the 1U servers. Odd but true.
The thing is, a Mac Mini is not designed to be a 24/7 server. It also lacks numerous "standard" server features like power supply redundancy, remote management capabilities, and hot-swap hard disks. A Mini comes with a 1 year warranty standard, while even a low-end Dell server will come with a 3 year warranty. A Mini is difficult and time-consuming to service. A typical rackmount is trivial to break down and nearly any arbitrary component can be replaced in a matter of minutes.
It's only anecdotal, but I have had a mini running almost constantly 24/7 for a few years now, almost constantly under load since it does the image-processing for network cameras linked up inside the house and garage. Traditional servers are designed around performance and they run hot, therefore needing all this industrial design to keep them c
Sometimes good things come in small packages. The Nehalem chips *are* better chips, but core-for-core they're nowhere near 2x as fast (let alone the ridiculous 10x as fast!).
I invite you to look at the mac-mini benchmarks (which are obviously the right ones
Bear in mind that the E4505 above is a dual quad core, and the mini is just a dual-core when you compare these. Also bear in mind that they're weighted averages - in some of the individual tests, the E5405 is almost 4x the speed.
It's the law of diminishing returns, you get a lot of bang-for-the-buck at the lower end, and you pay increasingly more as you get towards the upper end.
Simon.
I just went to the page where you can configure the blades, and specced out an 8-core blade:
- 2x E5405 Blade )cheapest quad-core)
- 4GB of RAM (4G is needed for Windows 2008 server)
- 2x Cheapest drive. RAID is required for OS installation (!)
- Windows SBS 2008, standard edition
Comes to $2,857 per blade, or $27,085 for 8 of them in the cheapest enclosure. Compare that to 16 minis at $15,984. Ouch.
Simon
The discussion was comparing blades to minis on the basis of cost, not density, but if you want to talk density...
I've just measured the rack in my garage - it's 30" deep by 19" across, and a mac mini fits handily into a 7" square by 4 holes in the rack with space to spare. 6U is 18 holes, so I can get 4 slide-out racks of minis into 6U, and 8 minis (including PSUs and wiring) per rack in a 2-across by 4 deep configuration. That's a total of 8x4x2 CPUs (because they're core-2 duos in the mini), or 64 CPUs.
The blades pack 8 blades into 6U. The ones we started off discussing therefore put 8 CPUs into 6U. Whoosh. If we jump to the later offering, that's 8x4=32 cores in 6U, or 32 less than the minis. Only if we go to the 8-core blades (for a total of 4299+8*1610 for the E5405 CPUs, or $17179 before you add OS and disk storage) do you get the same density of 64 CPUs in 6U.
The largest number of cores I'm aware of in standard 1U is dual quad-core (soon to be dual 6-core I suppose), so in 6U you get 6x8 = 48 CPUs. That doesn't work out too well compared to 6U of minis (64 CPUs). There are half-width/depth servers around, and you can get dual quad-core in that form-factor, so I could see the density going up to 96 CPUs in 6U, but I'm not aware of pricing on them...
I'm not advocating that you build your next computing cluster out of mac minis. I'm saying they scale (in price) reasonably well, that's all. I assume that there's a design difference between a blade server and a more consumer-orientated server (though I've been using a mini as an entertainment hub for the last 2 years without a single problem).
Certainly it'll *look* more professional to have a blade server in the rack. Whether it'll perform better for the price is more up for debate.
Simon
Yeah, you have a point on the Celerons. I can't find anything exactly equivalent to the minimac hardware in these blades, but among the pre-configured options on that blade page is a quad-core Xeon for $609 with 2GB RAM. An extra 2GB RAM is +$129, so $738 for a quad core Xeon with 4GB RAM. Using your reasoning with the celeron, one of those xeon blades == 2 minimacs. If we load up an 8 slot chassis that's $4299+8*738 = $10,203 vs 16*999 = $15,984. Similar but with 8 cores & 8GB RAM per blade and 32 minimacs: $14,027 vs $31,968. As we approach "cramp LOADS of those things in a small space" the blades are looking way cheaper now as well as more space efficient.
What you're also ignoring here is the cost of the OS. You're paying $999 because the mini comes with the server-grade OS. You need to add on the cost of Windows Server 2008 per blade (I'll let you get away with the 5-CAL licences, even though OSX is unlimited), for a further cost (ouch!) of ~$1029 per blade. That takes your costs to $(4299 + 8*(738+1029)) = $18435, which is again more than the 16 minis ( =$15,984).
If the argument is that you can use linux rather than windows, well fine. You can do that on the minis too though. If we take this route you don't need the mini-server bundle, so you can get the $599 package and the costs compare as $(4299 + 8*738) = $10,203 vs the minis at 16x599 = $9,584. You still have to buy storage for your blades as well, whereas the minis come with hard-disks.
To a certain extent, this is tongue-in-cheek. I'm not sure I'd recommend anyone buy 32 mac-minis if they wanted a computing cluster for example, but if you're trying to do it on the cheap, they're not actually that bad...
One more? You don't need any cables going in to the blades. Power, network, console, etc are supplied by the backplane.
Well, actually I was assuming 6 independent psu's would take 6 wires, and 1 network cable would take 1 wire, for a total of 7, compared to 2 (power, network)x4 for a total of 8. That's not a huge difference :) The "per mini" was wrong, and left in from text I'd changed.
You can't install an OS, upgrade an OS, access firmware, toggle power or reset them that way. You can do all of that remotely with ILO and work with CD images over the network.
Actually, yes you can. You need to do some prior set-up work (have a netinstall image available or leave the OS DVD in the drive) but LOM (literally: "Lights-Out Management") lets you reboot, power-down and (later, if necessary) power-up, and the system management tools let you install software, choose where to boot from, and relay everything across the net to a client machine. I've done it before from the OS DVD I left for that purpose in the drive of an XServe. Upgrading software doesn't need anything more than ssh or vnc ("Screen sharing" as it's called on the mac) to use the GUI tools.
more-efficient energy: I meant in terms of scale. A few large power supplies will almost certainly be more efficient than many small ones.
I'm
Yes, and an 8-core XServe is 1U as well, but that's also out of scope of the discussion...
Simon
Hmm. Thinking about this a little bit more...
:) In fact I'm thinking about getting rid of the XServe I have at home and getting a mini instead...
The Core 2 Duo @ 2.53GHz is more than than twice as fast as the 1.8GHz Celeron, so really you only need 4 of the mini-servers to beat out 8 of the blades... Suddenly the price swings significantly to the mini's... ($4299+8*379 = 7331. 4x$999 = 3996, or roughly half the price) and you've still got 16GB of RAM, not 8GB (which is what the blades have)
I was using 8's because that's the unit of the enclosure. If you want to start using more blades, the 16-way enclosure is even more expensive, relatively, with prices for just the enclosure starting at $17k. Just multiply up to suit the numbers. I think you'll easily fit 4 mini's within a 6U rack-space though - you could probably fit them on a 2U tray...
To address your other concerns:
individual cables: You need an ethernet wire and a power supply. That's one more wire per mini. If you're using external storage, you'll need a firewire cable or whatever, but that's the same as with the blades
organization: The minis are running OSX server, so all the LOM, remote-desktop, remote console etc. server-admin tools work fine.
more-efficient energy: The enclosures list up to 6x 1200W PSU's for the enclosure, or 7,200W. They'll draw less than that usually, of course, but that's their capacity. The mini's run at 14W at idle according to apple. I've seen reports of 40W when it's working hard.
hot-swap stuff: The blades don't hot-swap, they have to be powered down and up again. They don't have hot-swap disks either. Neither does the mini of course - you're supposed to plug in external disks (which can be hot-swap, of course) for both.
Overall, the mini comes out looking pretty good
Simon