Apple should add dual PSU to the G5 Xserve. IMO that was a big oversight.
Not really a majority of X86 1U Servers don't have Redundent PS either.
Apple need not rush the Xserve. Servers are for running 24/7 with stability. You've lost any speed advantage the G5 offers if there are thermal issues which cause problems.
I wouldn't be suprised to see the Xserve be the first Macs to use 90nm 970s. IBM would only need to fab a fraction of 970s to cover the Xserves needs.
Panther Server should be tuned for the Xserve as well.
[B]xServe should support at least 16GB RAM. 2GB RAM is joke in server world.
At least 16GB eh? Hmm. A Sun system that specs in around what a dual G5 w/ 16GB RAM will set you back about $30k. I agree that 2GB is light, but >16GB? I think that's a bit overkill for a $3k server, no? 8GB like the G5 will do quite nicely.
Quote:
Where are SCSI hard disks??? SCSI is the thing in the server world.
Yes, teh bomb! You don't own a server, do you? SCSI is great if you have high transaction servers. Apple's customers don't buy high transaction servers. Therefore, Apple doesn't do high transaction servers.
Apple's customers buy high bandwidth and large storage servers. Apple needs to sell high bandwidth and large storage servers. My 240GB Xserve would have cost 3x as much if it was loaded up with SCSI drives. SATA will make the Xserve better than it is now.
I was well aware of the loudness that would be acceptable in an xServe. I guess I was referring to a lower MHz processor (1.0, 1.2GHz G5). This is the internal xServe:
There are no external fans other than on the power supply, and the two internals are mainly for the (up to) 4 hard drives. You certainly wouldn't need that in a PowerBook. One HD one CPU and one fan. Anyone think that could happen inside a PowerBook soon? I sure do. Mobo designing with the additional HT and various other components could be the one thing holding up a G5 PowerBook. Seeing as how a 1.2GHz PPC970 consumes 19w according to IBM and the 1.0GHz G4 7455 consumes 20w according to Motorola, it doesn't seem like CPU heat is the main issue.
My guess would be that Apple will go through a die shrink before putting a 970 into a laptop. I am pretty sure that the real bottleneck is getting the new MOBO ready and that this is what is holding back a 970 powerbook. Hopefully the mobo and die shrink happen at the same time- it would be good to get a 1.8GHz G5 in a powerbook.
<ramble>
I was aware of the 19W heat dissipation, but I think that Apple wants to put a faster 970 into a laptop. 1.2 looks pretty slow compared to 2GHz. Of course, it is possible that 1.2 (600MHz bus) would make for an easier mobo design.
My general thoughts are that Apple isn't going to come out with a new laptop mobo (17 incher) and then eight months later come out with yet another new mobo. The G4 desktop DDR hacktastic mutant mobos came out with the XServe and then migrated into the Pro machines where they were recently replaced by the new 970 mobos. Counting shipping time of G5's, then these hacktastic DDR mobos were on the market for a year. If the laptops cycle through their similar DDR mobos, then we should have a G5 laptop announced in January and available in March? That would be cool. Who needs a RIO G4? If we don't get a new G4 15 powerbook in a month or two, then we know that something really big is coming.
Originally posted by Mark- Card Carrying FanaticRealist
And that's a good point!
The xServe is a nice first effort, in terms of price and packaging, but it cannot be taken seriously in a real line-of-business or 24x7 function until it has concepts like redundant PSUs or bullet-proof remote management.
I might be mistaken, but it depends on what you are using your servers for. Running your coroprate database is a 24/7 operation anc needs all the redundancy you can get (heck, you need redundant servers, not just power supplies). Running a render farm is a different issue. You need the master rendering program to be fault tolerant in case one of its slaves decides to go down. I usually see a G5 as a render farm kind of server, kind of a folding@home rack of death.
I might be mistaken, but it depends on what you are using your servers for. Running your coroprate database is a 24/7 operation anc needs all the redundancy you can get (heck, you need redundant servers, not just power supplies). Running a render farm is a different issue. You need the master rendering program to be fault tolerant in case one of its slaves decides to go down. I usually see a G5 as a render farm kind of server, kind of a folding@home rack of death.
And I agree with a fair amount of that
However, in order to grow and to continue being seen as a relevant platform, Apple has to look beyond the CGI market for G5 and understand that for Oracle's and Sybase's investment to have a realistic chance of generating return, xServe must grow from a single product to a range of products, some of which should have N+1 or load-balanced redundancy for network and environmental components alongside comprehensive redundancy for storage and hot-swap replacement for memory.
If you look at the IBM range, there is an unbelievable degree of granularity in terms of server options, whereas in the HP Proliant range there probably is not enough.
I'm proposing something in the middle: the current xServe will become a 970-based design and become the baby of the range, a 4-way launched in 2004 will have ship with a single 970, redundant PSU, dual gigabit, 4 PCI-X slots and the potential for 16GB of RAM with 2GB as standard, whilst an 8-way launched in 2005 will ship 2-way as standard, hot-swap PSU and memory, 6 PCI-X, capable of 32GB but shipping with 2GB as standard.
I also think it should be possible to join up to two 4-ways or up to four 8-ways via a NUMA-type mechanism to create a larger system unconstrained by some of the limitations of conventional larger SMP configurations. Please note that I am not confusing the concept of a NUMA system with the concept of clustering.
I can see the potential in the G5 to effectively outperform and underprice Sun Microsystems' server products. I also believe that - given the cost of Madison and the fact that Itanium will have difficulty becoming a mass-market product (whilst Xeon and P4 continue to breathe) - Apple has the opportunity to leverage the economies of scale offered by using 970 across the board to build marketshare on cost of ownership.
When Xeon and P4 cease to exist, MS and Intel lose a significant advantage as customers will be forced on to a forced upgrade path if their applications cannot operate at tolerable speed under 32-bit emulation.
Apple needs to have a mature and complete product line with a high-quality ISV portfolio and a killer set of developer tools for internal developers by that time. This is the fight for the next generation of computing - like the shift to 32-bit minicomputers in the late-seventies, or the shift from character to GUI - and Apple can capitalise by having the best products combined with the best capital cost profile, best management tools and lowest TCO.
I'm not saying that the market will suddenly do a polarity shift and Apple will have 95% of the market, I'm saying that Apple has the opportunity to get back to a 30%-35% share which - if there wasn't already a monopoly power in the marketplace - would be considered de facto market leadership.
Let's see do I want a $3200 chip or a $3200 Computer that is a whole lot better?
I guess if I was a business man in an Intel world I would want to do what everyone else was telling me to do and buy the 3200 doller chip.
If It was my own company and cared about what I was doing, I would go for the G5 computer coming out in August. The G5 comes with more open code than Windows will ever be. And that is proving to be the better option for stability, the belief is that when people volunteer their time to work, they are more passionate and therefore more able to write better code, and that's what I believe. Apple.
Comments
Apple should add dual PSU to the G5 Xserve. IMO that was a big oversight.
Not really a majority of X86 1U Servers don't have Redundent PS either.
Apple need not rush the Xserve. Servers are for running 24/7 with stability. You've lost any speed advantage the G5 offers if there are thermal issues which cause problems.
I wouldn't be suprised to see the Xserve be the first Macs to use 90nm 970s. IBM would only need to fab a fraction of 970s to cover the Xserves needs.
Panther Server should be tuned for the Xserve as well.
Where are SCSI hard disks??? SCSI is the thing in the server world.
Apple should release new Xserves when new 10.3 is ready.
Originally posted by Stratosfear
[B]xServe should support at least 16GB RAM. 2GB RAM is joke in server world.
At least 16GB eh? Hmm. A Sun system that specs in around what a dual G5 w/ 16GB RAM will set you back about $30k. I agree that 2GB is light, but >16GB? I think that's a bit overkill for a $3k server, no? 8GB like the G5 will do quite nicely.
Where are SCSI hard disks??? SCSI is the thing in the server world.
Yes, teh bomb! You don't own a server, do you? SCSI is great if you have high transaction servers. Apple's customers don't buy high transaction servers. Therefore, Apple doesn't do high transaction servers.
Apple's customers buy high bandwidth and large storage servers. Apple needs to sell high bandwidth and large storage servers. My 240GB Xserve would have cost 3x as much if it was loaded up with SCSI drives. SATA will make the Xserve better than it is now.
Originally posted by Rhumgod
I was well aware of the loudness that would be acceptable in an xServe. I guess I was referring to a lower MHz processor (1.0, 1.2GHz G5). This is the internal xServe:
There are no external fans other than on the power supply, and the two internals are mainly for the (up to) 4 hard drives. You certainly wouldn't need that in a PowerBook. One HD one CPU and one fan. Anyone think that could happen inside a PowerBook soon? I sure do. Mobo designing with the additional HT and various other components could be the one thing holding up a G5 PowerBook. Seeing as how a 1.2GHz PPC970 consumes 19w according to IBM and the 1.0GHz G4 7455 consumes 20w according to Motorola, it doesn't seem like CPU heat is the main issue.
My guess would be that Apple will go through a die shrink before putting a 970 into a laptop. I am pretty sure that the real bottleneck is getting the new MOBO ready and that this is what is holding back a 970 powerbook. Hopefully the mobo and die shrink happen at the same time- it would be good to get a 1.8GHz G5 in a powerbook.
<ramble>
I was aware of the 19W heat dissipation, but I think that Apple wants to put a faster 970 into a laptop. 1.2 looks pretty slow compared to 2GHz. Of course, it is possible that 1.2 (600MHz bus) would make for an easier mobo design.
My general thoughts are that Apple isn't going to come out with a new laptop mobo (17 incher) and then eight months later come out with yet another new mobo. The G4 desktop DDR hacktastic mutant mobos came out with the XServe and then migrated into the Pro machines where they were recently replaced by the new 970 mobos. Counting shipping time of G5's, then these hacktastic DDR mobos were on the market for a year. If the laptops cycle through their similar DDR mobos, then we should have a G5 laptop announced in January and available in March? That would be cool. Who needs a RIO G4? If we don't get a new G4 15 powerbook in a month or two, then we know that something really big is coming.
</ramble>
Originally posted by Mark- Card Carrying FanaticRealist
And that's a good point!
The xServe is a nice first effort, in terms of price and packaging, but it cannot be taken seriously in a real line-of-business or 24x7 function until it has concepts like redundant PSUs or bullet-proof remote management.
I might be mistaken, but it depends on what you are using your servers for. Running your coroprate database is a 24/7 operation anc needs all the redundancy you can get (heck, you need redundant servers, not just power supplies). Running a render farm is a different issue. You need the master rendering program to be fault tolerant in case one of its slaves decides to go down. I usually see a G5 as a render farm kind of server, kind of a folding@home rack of death.
Originally posted by Yevgeny
I might be mistaken, but it depends on what you are using your servers for. Running your coroprate database is a 24/7 operation anc needs all the redundancy you can get (heck, you need redundant servers, not just power supplies). Running a render farm is a different issue. You need the master rendering program to be fault tolerant in case one of its slaves decides to go down. I usually see a G5 as a render farm kind of server, kind of a folding@home rack of death.
And I agree with a fair amount of that
However, in order to grow and to continue being seen as a relevant platform, Apple has to look beyond the CGI market for G5 and understand that for Oracle's and Sybase's investment to have a realistic chance of generating return, xServe must grow from a single product to a range of products, some of which should have N+1 or load-balanced redundancy for network and environmental components alongside comprehensive redundancy for storage and hot-swap replacement for memory.
If you look at the IBM range, there is an unbelievable degree of granularity in terms of server options, whereas in the HP Proliant range there probably is not enough.
I'm proposing something in the middle: the current xServe will become a 970-based design and become the baby of the range, a 4-way launched in 2004 will have ship with a single 970, redundant PSU, dual gigabit, 4 PCI-X slots and the potential for 16GB of RAM with 2GB as standard, whilst an 8-way launched in 2005 will ship 2-way as standard, hot-swap PSU and memory, 6 PCI-X, capable of 32GB but shipping with 2GB as standard.
I also think it should be possible to join up to two 4-ways or up to four 8-ways via a NUMA-type mechanism to create a larger system unconstrained by some of the limitations of conventional larger SMP configurations. Please note that I am not confusing the concept of a NUMA system with the concept of clustering.
I can see the potential in the G5 to effectively outperform and underprice Sun Microsystems' server products. I also believe that - given the cost of Madison and the fact that Itanium will have difficulty becoming a mass-market product (whilst Xeon and P4 continue to breathe) - Apple has the opportunity to leverage the economies of scale offered by using 970 across the board to build marketshare on cost of ownership.
When Xeon and P4 cease to exist, MS and Intel lose a significant advantage as customers will be forced on to a forced upgrade path if their applications cannot operate at tolerable speed under 32-bit emulation.
Apple needs to have a mature and complete product line with a high-quality ISV portfolio and a killer set of developer tools for internal developers by that time. This is the fight for the next generation of computing - like the shift to 32-bit minicomputers in the late-seventies, or the shift from character to GUI - and Apple can capitalise by having the best products combined with the best capital cost profile, best management tools and lowest TCO.
I'm not saying that the market will suddenly do a polarity shift and Apple will have 95% of the market, I'm saying that Apple has the opportunity to get back to a 30%-35% share which - if there wasn't already a monopoly power in the marketplace - would be considered de facto market leadership.
Let's see do I want a $3200 chip or a $3200 Computer that is a whole lot better?
I guess if I was a business man in an Intel world I would want to do what everyone else was telling me to do and buy the 3200 doller chip.
If It was my own company and cared about what I was doing, I would go for the G5 computer coming out in August. The G5 comes with more open code than Windows will ever be. And that is proving to be the better option for stability, the belief is that when people volunteer their time to work, they are more passionate and therefore more able to write better code, and that's what I believe. Apple.