Fair enough, but you can't blame Sun for being expensive, then criticise when low cost machines come in. Also, the 280R's also never been 10 times the price of an Intel equivalent. Equally, there's no Intel box with the equivalent internal bandwidth or 64 bit capability which'll run your enterprise app quite as well.
The 210s and 240s have great features - each actually has 4 built in Gb ethernet ports!
>Purchase SUN and immediately halt all new >development on Solaris.
And piss off your entire customer base, your staff, freeze any competitive advantage you have and give the compeitition time to catch up, watch the Solaris developers leave in disgust, stop anyone buying any new Sun kit as Solaris is a deadend product.
>Shift all core-engineering at SUN to shipping a >version of OS-X that runs on existing SUN >hardware and contains the scalability >enhancements that allow SUN to be so successful >in the supercomputer marketplace.
ie reinvent the wheel replacing your battle tested, well respected OS with a different one, ported to a new architecture. At the same time try and persuade your customers and ISVs that porting all their code to a new unproven untested platform that will take years to match the old Solaris is a good idea. Still keep spending money on Solaris anyway supporting exisitng customers.
>Using the SUN $5billion warchest, go on a massive >crusuade to convert existing SUN shops to OS-X on >SPARC and PPC, citing all SUN "innovations" being >introduced into the OS-X kernel and core >libraries.
And try and find an answer to customers who say 'what on earth is the point of anything you're doing - why do I want a new OS that's got no experience of enterprise computing? I've already got an OS and don't need a new one, even if it's prettier to look at.'.
This is the stupidest comment I've read on Slashdot recently, second only to the guy who said that 32 single cpu PIVs would run Oracle better than a 32 cpu 6800.
Nothing about your strategy works. The only possible thing that would work would be Sun and Apple as one company possessing a viable desktop alternative.
Sorry, Slashdotters, but this perception that the equivalent Intel box is 10-15% of the price is utterly ludicrous. Have you seen the prices of the 210s, 240s and up?
You'll need to go back and do the config against the newly released v210, then factor in the cost of the software that comes for free with the v210. Should turn the tables nicely on the Dell box.
Partitioning with hardware fault isolation lacking on Sun? What docs show this? Since Sun leads with its dynamic system domains, your comment sounds unjustified.
Not sure about SGI - but they are irrelevant in business computing anyway, but IBM and HP do not scale as well as Sun. They do not have the linear scalability of UltraSPARC and do not have 106+ cpus in one system. The 128 cpu Superdome is two systems connected together by a highspeed interconnect. Fujitsu have a 128 cpu box however, which Solaris can take full use of. HPUX and AIX can't.
As for Itanium, heat is an extremely important factor as it means that you're going to have to build larger systems that consume more power so that you can cool them. If that means your 4 way takes up twice as many RUs as the competition, as well as sucking up a huge amount more power, that's a factor. What OS were you wanting to run anyway? There's no proven track record for Itanium systems and hardly any apps.
Sources such as Gartner and IDC are fine for me regarding Sun's market share. Sun's revenues also increased - the loss was due to costs of acquisitions and Sun actually made an operating profit. As for Linux market share, you're right it is hard to measure. But since it runs on generally on 1-2 way x86 systems, you track the revenue of how those Intel based systems are selling to get an idea of the hardware market.
Solaris is indeed an important differentiator, but the hardware's also extremely important. How many Intel systems scale as well? None. SPARC's never been the leader in cpu benchmarks, that's true, but Sun have consistently produced balanced systems that perform well when running applications rather than cpu benchmarks. If just fast cpus were important, we'd all be using Alphas.
Sun's market share in Unix systems grew last year, where HP and IBM sank. HP and IBM certainly do not 'wipe' Sun at this level, either technically or financially. The low end Sun systems are competitively priced and the storage is excellent, so what's the problem exactly?
I'm confused as to why everyone on Slashdot concentrates on irrelevant cpu benchmarks and is so down on Sun at the moment.
That's part of the reason why it's important to have some fat in your diet. Fat sates you, sending signals to the brain that you have eaten. This is part of the way a low carb diet works.
It's not version 1.0. There was a previous version which was around under the iPlanet and then Sun ONE brand and is used by various customers. It was version 3 then, to fit in with the fact it was an add-on to Portal Server 3.0. We're now at Portal Server 6.0, hence the numbering match.
There's a large number of reasons to move to 8/9, not least the ability to run on the new UIII systems from 8 upwards, have a journalling, high performance file system in 9 and various other goodies. Sun.com has a good list.
For what it's worth, things went from 2.6 to 7 and up. SunOS is the operating system, Solaris is the 'operating environment', but you're right, in the end it is just numbers. It can't be easyt in the marketing department - for all the people who complain that the new numbering is 'hype', which is a little extreme a criticism, you'd have another lot complaining the old numbering made Sun look 'old', had it been kept.
Don't forget too, that the 32 node figure doesn't conform to Oracle's recommendations either. Oracle recommend using at least 4 cpu x86 boxes. What happens when you need to run a batch job that needs more than 4cpus to complete in time? You're screwed. There's also the cost of the SAN that you'd need. Also, the bandwidth you'd need to have 32 single cpu boxes communicating would be horrendous - you wouldn't get anything near linear scalability. The admin for the whole set up would be pretty intensive as well.
In the Sun world, if price were the key player, you'd be looking at an HA cluster of something like v480s, v880s or v1280s, all very competitively priced. They could use direct attached storage, so no need for the expense of a SAN. If you wanted RAC, you could do that too, with no need for the expense of a SAN. Even if you had things attached via a SAN, RAC with large SMP boxes offers much better performance than lots of smaller nodes. RAC puts significant extra load on cpus usage over normal Oracle, so it's still good to have a reasonable number of cpus in each node.
There's a belief on Slashdot that all computing problems can be solved with a lot of small PCs. It's simply not the case.
He was right in a way - look at how web based interfaces are now the norm. To a great extent, the vision of centralised apps and services has come true. It's just the method used to 'view' things didn't quite work out. Unless you're talking about Sun Rays, WinTerms and the like.
"On the PC platform, it's amazing that Sun actually recommends WINDOWS rather than a UNIX OS (like Linux.) They've given up on the PC platform - they let Microsoft own the entry-level systems."
I'm sorry but that's utter nonsense. What do you think the LX50 is about? What about the Sun Ray product line, what about the Mad Hatter plans?
Red Hat with Sun Linux branding. Free ChiliSoft Active Server Pages software, you could stick the cdrom in the LX50 and have it boot and automatically install everything. It's good stuff.
This new desktop will be Linux based. The Sun Rays are Solaris based, but any OS stuff that runs on Solaris runs happily on them. Sun are actively developing the Sun ONE stack for Linux and support numerous open source projects - Gnome for example. What more are you expecting?
The IT problems of this world cannot be cured by the panacea of a Linux cluster, which appears only to exist in the world of Slashdot. Chances are this guy is talking about a large database from Oracle or Sybase. Noone is doing this in Linux, however much Oracle's marketing and sales guys would like you to believe that. Equally important is that noone can implement and support this as at a decent price either.
True, but for the vast majority of IT managers and staff too, the source code being available to them is utterly irrelevant. The fact it is avaialble is good for Red Hat, who will be supporting these 'products', but most sys admins/IT guys have no interest whatsoever in ploughing through source code to fix something they paid money for in the first place.
Excellent comment. Great to read something sensible on Slashdot about Sun rather than all the 'it didn't ship with a web server' or 'Solaris is slow' crap.
Slashdot Mirror
We did have Outlook Express for Solaris!
Fair enough, but you can't blame Sun for being expensive, then criticise when low cost machines come in. Also, the 280R's also never been 10 times the price of an Intel equivalent. Equally, there's no Intel box with the equivalent internal bandwidth or 64 bit capability which'll run your enterprise app quite as well.
The 210s and 240s have great features - each actually has 4 built in Gb ethernet ports!
Oh come on, it's crazy!
>Purchase SUN and immediately halt all new >development on Solaris.
And piss off your entire customer base, your staff, freeze any competitive advantage you have and give the compeitition time to catch up, watch the Solaris developers leave in disgust, stop anyone buying any new Sun kit as Solaris is a deadend product.
>Shift all core-engineering at SUN to shipping a >version of OS-X that runs on existing SUN >hardware and contains the scalability >enhancements that allow SUN to be so successful >in the supercomputer marketplace.
ie reinvent the wheel replacing your battle tested, well respected OS with a different one, ported to a new architecture. At the same time try and persuade your customers and ISVs that porting all their code to a new unproven untested platform that will take years to match the old Solaris is a good idea. Still keep spending money on Solaris anyway supporting exisitng customers.
>Using the SUN $5billion warchest, go on a massive >crusuade to convert existing SUN shops to OS-X on >SPARC and PPC, citing all SUN "innovations" being >introduced into the OS-X kernel and core >libraries.
And try and find an answer to customers who say 'what on earth is the point of anything you're doing - why do I want a new OS that's got no experience of enterprise computing? I've already got an OS and don't need a new one, even if it's prettier to look at.'.
This is the stupidest comment I've read on Slashdot recently, second only to the guy who said that 32 single cpu PIVs would run Oracle better than a 32 cpu 6800.
Nothing about your strategy works. The only possible thing that would work would be Sun and Apple as one company possessing a viable desktop alternative.
You forgot the important bit - go out of business after a couple of months after strategy proves uttlery stupid.
Sorry, Slashdotters, but this perception that the equivalent Intel box is 10-15% of the price is utterly ludicrous. Have you seen the prices of the 210s, 240s and up?
You'll need to go back and do the config against the newly released v210, then factor in the cost of the software that comes for free with the v210. Should turn the tables nicely on the Dell box.
Partitioning with hardware fault isolation lacking on Sun? What docs show this? Since Sun leads with its dynamic system domains, your comment sounds unjustified.
Not sure about SGI - but they are irrelevant in business computing anyway, but IBM and HP do not scale as well as Sun. They do not have the linear scalability of UltraSPARC and do not have 106+ cpus in one system. The 128 cpu Superdome is two systems connected together by a highspeed interconnect. Fujitsu have a 128 cpu box however, which Solaris can take full use of. HPUX and AIX can't.
As for Itanium, heat is an extremely important factor as it means that you're going to have to build larger systems that consume more power so that you can cool them. If that means your 4 way takes up twice as many RUs as the competition, as well as sucking up a huge amount more power, that's a factor. What OS were you wanting to run anyway? There's no proven track record for Itanium systems and hardly any apps.
Sources such as Gartner and IDC are fine for me regarding Sun's market share. Sun's revenues also increased - the loss was due to costs of acquisitions and Sun actually made an operating profit. As for Linux market share, you're right it is hard to measure. But since it runs on generally on 1-2 way x86 systems, you track the revenue of how those Intel based systems are selling to get an idea of the hardware market.
Nice post.
Now cue all the posts informing us that all you need is a few single cpu Linux boxes and all the world's applications suddently scale ad infinitum.
Solaris is indeed an important differentiator, but the hardware's also extremely important. How many Intel systems scale as well? None. SPARC's never been the leader in cpu benchmarks, that's true, but Sun have consistently produced balanced systems that perform well when running applications rather than cpu benchmarks. If just fast cpus were important, we'd all be using Alphas.
Sun's market share in Unix systems grew last year, where HP and IBM sank. HP and IBM certainly do not 'wipe' Sun at this level, either technically or financially. The low end Sun systems are competitively priced and the storage is excellent, so what's the problem exactly?
I'm confused as to why everyone on Slashdot concentrates on irrelevant cpu benchmarks and is so down on Sun at the moment.
That's part of the reason why it's important to have some fat in your diet. Fat sates you, sending signals to the brain that you have eaten. This is part of the way a low carb diet works.
It's not version 1.0. There was a previous version which was around under the iPlanet and then Sun ONE brand and is used by various customers. It was version 3 then, to fit in with the fact it was an add-on to Portal Server 3.0. We're now at Portal Server 6.0, hence the numbering match.
p /h ome_portal_icp.html
http://wwws.sun.com/software/products/portal_ic
There's a large number of reasons to move to 8/9, not least the ability to run on the new UIII systems from 8 upwards, have a journalling, high performance file system in 9 and various other goodies. Sun.com has a good list.
For what it's worth, things went from 2.6 to 7 and up. SunOS is the operating system, Solaris is the 'operating environment', but you're right, in the end it is just numbers. It can't be easyt in the marketing department - for all the people who complain that the new numbering is 'hype', which is a little extreme a criticism, you'd have another lot complaining the old numbering made Sun look 'old', had it been kept.
Great, except you can't even buy Opteron systems yet.
Don't forget too, that the 32 node figure doesn't conform to Oracle's recommendations either. Oracle recommend using at least 4 cpu x86 boxes. What happens when you need to run a batch job that needs more than 4cpus to complete in time? You're screwed. There's also the cost of the SAN that you'd need. Also, the bandwidth you'd need to have 32 single cpu boxes communicating would be horrendous - you wouldn't get anything near linear scalability. The admin for the whole set up would be pretty intensive as well.
In the Sun world, if price were the key player, you'd be looking at an HA cluster of something like v480s, v880s or v1280s, all very competitively priced. They could use direct attached storage, so no need for the expense of a SAN. If you wanted RAC, you could do that too, with no need for the expense of a SAN. Even if you had things attached via a SAN, RAC with large SMP boxes offers much better performance than lots of smaller nodes. RAC puts significant extra load on cpus usage over normal Oracle, so it's still good to have a reasonable number of cpus in each node.
There's a belief on Slashdot that all computing problems can be solved with a lot of small PCs. It's simply not the case.
Bit confused here. The Ultra 5 uses and UII and stopped at 400Mhz, so it's hardly surprising your 1.6Ghz PC compiled things quicker.
Also a fair point, how did the app run under heavy load (if relevant)? You're not just buying a Sun box, you're getting Solaris too.
He was right in a way - look at how web based interfaces are now the norm. To a great extent, the vision of centralised apps and services has come true. It's just the method used to 'view' things didn't quite work out. Unless you're talking about Sun Rays, WinTerms and the like.
"On the PC platform, it's amazing that Sun actually recommends WINDOWS rather than a UNIX OS (like Linux.) They've given up on the PC platform - they let Microsoft own the entry-level systems."
I'm sorry but that's utter nonsense. What do you think the LX50 is about? What about the Sun Ray product line, what about the Mad Hatter plans?
Red Hat with Sun Linux branding. Free ChiliSoft Active Server Pages software, you could stick the cdrom in the LX50 and have it boot and automatically install everything. It's good stuff.
This new desktop will be Linux based. The Sun Rays are Solaris based, but any OS stuff that runs on Solaris runs happily on them. Sun are actively developing the Sun ONE stack for Linux and support numerous open source projects - Gnome for example. What more are you expecting?
The IT problems of this world cannot be cured by the panacea of a Linux cluster, which appears only to exist in the world of Slashdot. Chances are this guy is talking about a large database from Oracle or Sybase. Noone is doing this in Linux, however much Oracle's marketing and sales guys would like you to believe that. Equally important is that noone can implement and support this as at a decent price either.
True, but for the vast majority of IT managers and staff too, the source code being available to them is utterly irrelevant. The fact it is avaialble is good for Red Hat, who will be supporting these 'products', but most sys admins/IT guys have no interest whatsoever in ploughing through source code to fix something they paid money for in the first place.
You can get a 64bit v100 server from Sun for under a $1000 dollars...
Excellent comment. Great to read something sensible on Slashdot about Sun rather than all the 'it didn't ship with a web server' or 'Solaris is slow' crap.
You really need to brush up on the market for Unix kit out there. Sun's volume kit sells extremely well.