The obvious point is that the Solaris network stack isn't as good as either, one of the problems being that it makes a lot of tradeoffs for it's traditional large customers. Sure, they might well appreciate their specific hack (like a recent Solris to Linux user I spoke to who was unhappy because Solaris "respected" the write() message bounderies down to the TCP packet level, and Linux didn't)... but they also like things that go fast in the general case.
Bullshit. The S10 tcp/ip stack is fantastic. Are
you talking single stream performance or aggregate throughput because fireengine scales almost perfectly. And your write() message annecdote just shows how little you know about tcp. Solaris will
also aggregate writes if the cost of this is amortized by the lower networks costs (packet count, rtt)
IMHO sun is not a friend of free software or even plain open source.
You are obviously a poor troll, an idiot or 10 years old (these
are not mutually exclusive).
http://www.sunsource.net/
Sun has made some very significant contributions over the years which are pretty fundamental to the way a lot of us use computers in our everyday lives. Get a clue before you post.
Hmm. Cynicism is fine to a point (we all remember MS's NT
mantra IITNR (its in the next release)). The problem, as others have
stated, is the sheer amount of legal checking and code that has
to be unencumbered. Sun made an estimate of how long this would
take when calculating a release date. This has turned out to be
more work than estimated and thus the projected release date has
slipped. I am confident that it will be out before the end of Q2
or early Q3 at the latest. I think the benefit of the doubt should be
given in this case, especially considering Sun's track record of
contributions to open source and open standards, rather than
assuming something nefarious is going on.
Isn't it better to wait a bit longer but get the quality right? (something that the software industry as a whole ALWAYS seems to get wrong).
Interesting? Since when did posting baseless
speculation become interesting on Slashdot? (actually, don't answer that). Niagara WILL suck at
floating point. This is no secret and never has
been. Most server apps don't touch the fpu. Single
thread performance will also suck. However, Niagara
will outperform a 4 way Xeon/Opteron box on throughput oriented workloads (i.e. those simulated
by such benchmarks as SAP, specweb, specint_rate,
specjbb2000). This is all for 60 watts in a 2u
rack or blade. Thats the value proposition (which is pretty compelling for enterprises like google, ebay, mobile phone companies (sms management)).
UltrasparcIV is actually not that bad, particularly on real workloads. The problem is Oracle's licensing costs.
I am a bit late to the party but what the hell.
On bigger SPARCS Sun also have dynamic system
domains which allow you to split the machine
up physically into a subset of fault tolerant
entities. Unlike LPARS/uPARS on P-series you
also get electrical isolation. This coupled
with zones gives you virtually as much
granularity as you need. Unlike IBM's virtualization, Solaris Zones also work on single
CPU X86 machines. Oh, and it's free.
Mostly agree. You are restricted to running the
same kernel image. However, the goals for zones
were: utilization, security and reduced admin
costs. It's hard to say that zones doesn't meet
these goals and meet them well. When you couple
zones (which come with the OS - not licensed
separately like VMWARE or upars) with dynamic
system domains which give you physical separation
(including electrical isolation within the chassis -
something pretty unique)
and redundancy then you have a very compelling
solution. Running more OS'es gives you more admin
costs and, in some cases, more licensing costs
for that extra flexibility.
I think they should of allowed the fopen to have file handles 0-254 and open/close etc. to use 255 upwards.
The commond workaround for this is to open 255 files
first and manage these yourself for stdio calls.
Successive calls (i.e. open(2)) will get fd's over
255. It's a pain but it works and doesn't break
the ABI. If you need more than 255 fopen()
descriptors then you are screwed.
Another great one is:
Woman: Mr Churchill, your drunk!
Churchill: Yes madam, but in the morning I will
be sober but you will still be ugly.
Re:Comparing UML to N1 Grid Containers? Ridiculous
on
Sun-isms Debunked
·
· Score: 1
Fortunately, they don't have to settle for inefficient 8 or 16 way machines, they can have 8 or 16 blades in about the same space.
Yeah, which they can buy from Sun running Solaris
OR you will be able to get a 2U 32way machine from
Sun within 18months.
Of course, I'm not saying that virtualization is useless. Quite to the contrary: it does have some important specialized uses. But Linux supports a wider spectrum of virtualization and partitioning options than Solaris.
No it doesn't. Sun supports dynamic system
domains in hardware which provide separate (
and electrically isolated - something that IBM
still can't do on their UNIX servers) execution
environments for resource utilization and
resiliency. What does linux do here?
Solaris10 has containers which are distinct,
isolated execution environments. They are free
with the OS (no licensing), you can have as
many as you like (even on a uniproc) as long as
you have the resources to support them. Single, master OS image to faciliate maintenance, backup and patching. Zones coupled with DSD's is a pretty compelling solution. The only other vendor that can even come close is IBM with Micropartitions however, these are separately licensed from IBM, have to run a distinct OS image, you can only have 10 per cpu and have a much higher performance penalty the more you have. What does linux have?
The problem is that Sun is overselling virtualization and selling virtualization to people who don't need it, and they are overhyping what they have. When a Sun salesperson tries to sell you a big Solaris server plus virtualization, hold on to your wallet.
This is so far off base it's not even funny.
Zones don't cost a penny. DSD's come free with
the hardware that supports it. You can get a
v20Z for peanuts and have a 1000 zones on it
if you want. Isolated webhosting (each zone
can be independently restarted) which costs
nothing more. Sounds good to me.
Re:Comparing UML to N1 Grid Containers? Ridiculous
on
Sun-isms Debunked
·
· Score: 1
Thats bullshit and will be even more evident when
8 and 16 way opteron boxes are out in CY05. Most
ORGANIZATIONS consider real estate, power
consumption and cooling when specifying an
installation. Having application virtualization
built in, for free, can't be anything other than
a good thing. Especially when the OS is free and
the support cheaper than the competition.
How do you know?
The source is NOT 'largely the same in some ways'.
The source is approx 98% exactly the same. The
only bits that are different are the platform
specific parts. However, these are also different
for the various sparc platforms so the implementation of Solaris, is in fact, exactly the
same for SPARC and for x86. Now, the build tools
are different (compilers). Solaris 10 X86 will be
the first release where the x86 version is given the same investment and development parity as it's
SPARC cousin. This will reflect in it's performance.
Sun techs would get all uppity about it being UNSUPPORTED and such
OK. So your GNU tools are normally supported by ???
(i.e. whats the difference between running them on
a linux distro or solaris). And, GNU tools, whilst
not covered under a support contract, are given
'best effort' support which is better than nothing.
System managers want to observe what's going on inside their kernel about as much as they want to see what's going on inside their bowels. That stuff just has to work, and it has to work automatically and without being noticed. If people ever have to muck around with dtrace or tuning kernel parameters, there is something seriously wrong with Solaris.
You ARE kidding right? If not that is one of the most shortsighted and ignorant things on slashdot (some sort of record in itself). I guess you still believe that there is only a need for 5 computers in the world?. Live in ignorance if you want but don't post about it.`
Erm did I miss something? Sun is not anti-linux, it's not anti OSS. It's a corporation which COMPETES with other companies. All Eric did was consider the apparently disparate philosophies behind Linux and Solaris development. Nowhere does he claim to be a Linux expert the same way most people who post here,
judging by the content of their posts, could claim to be Sun or Solaris experts. Do RedHat not have to convince people to shell out money, do they not have 'marketing machinery'?
Slashdot Mirror
Erm, you mean legacy services. Maybe you should read about SMF (hell, perhaps even use it) before posting?
If it was written to the DDI/DKI properly then it is guaranteed to work. I am always amazed that most slashdotter don't appreciate this....
Also, with the release of OpenSolaris, Dtrace is now becomes more useful (and it was very, very useful before).
Or just use smf on Solaris/OpenSolaris
Hey, stop posting facts. This post does not label Sun and SCO evil so will be removed.
Rearrange ther following words: Line Sinker Hook And
The definition of vapourware in this story was declared in the OP. Your definition of vapourware is, like your post, irrelevent.
Hmm. Cynicism is fine to a point (we all remember MS's NT mantra IITNR (its in the next release)). The problem, as others have stated, is the sheer amount of legal checking and code that has to be unencumbered. Sun made an estimate of how long this would take when calculating a release date. This has turned out to be more work than estimated and thus the projected release date has slipped. I am confident that it will be out before the end of Q2 or early Q3 at the latest. I think the benefit of the doubt should be given in this case, especially considering Sun's track record of contributions to open source and open standards, rather than assuming something nefarious is going on. Isn't it better to wait a bit longer but get the quality right? (something that the software industry as a whole ALWAYS seems to get wrong).
http://dictionary.reference.com/search?q=irony
Interesting? Since when did posting baseless speculation become interesting on Slashdot? (actually, don't answer that). Niagara WILL suck at floating point. This is no secret and never has been. Most server apps don't touch the fpu. Single thread performance will also suck. However, Niagara will outperform a 4 way Xeon/Opteron box on throughput oriented workloads (i.e. those simulated by such benchmarks as SAP, specweb, specint_rate, specjbb2000). This is all for 60 watts in a 2u rack or blade. Thats the value proposition (which is pretty compelling for enterprises like google, ebay, mobile phone companies (sms management)). UltrasparcIV is actually not that bad, particularly on real workloads. The problem is Oracle's licensing costs.
I am a bit late to the party but what the hell. On bigger SPARCS Sun also have dynamic system domains which allow you to split the machine up physically into a subset of fault tolerant entities. Unlike LPARS/uPARS on P-series you also get electrical isolation. This coupled with zones gives you virtually as much granularity as you need. Unlike IBM's virtualization, Solaris Zones also work on single CPU X86 machines. Oh, and it's free.
Mostly agree. You are restricted to running the same kernel image. However, the goals for zones were: utilization, security and reduced admin costs. It's hard to say that zones doesn't meet these goals and meet them well. When you couple zones (which come with the OS - not licensed separately like VMWARE or upars) with dynamic system domains which give you physical separation (including electrical isolation within the chassis - something pretty unique) and redundancy then you have a very compelling solution. Running more OS'es gives you more admin costs and, in some cases, more licensing costs for that extra flexibility.
You shouldn't underestimate patent indemnification and guaranted binary compatibility
Another great one is: Woman: Mr Churchill, your drunk! Churchill: Yes madam, but in the morning I will be sober but you will still be ugly.
Thats bullshit and will be even more evident when 8 and 16 way opteron boxes are out in CY05. Most ORGANIZATIONS consider real estate, power consumption and cooling when specifying an installation. Having application virtualization built in, for free, can't be anything other than a good thing. Especially when the OS is free and the support cheaper than the competition.
How do you know? The source is NOT 'largely the same in some ways'. The source is approx 98% exactly the same. The only bits that are different are the platform specific parts. However, these are also different for the various sparc platforms so the implementation of Solaris, is in fact, exactly the same for SPARC and for x86. Now, the build tools are different (compilers). Solaris 10 X86 will be the first release where the x86 version is given the same investment and development parity as it's SPARC cousin. This will reflect in it's performance.
Solaris x86 = Solaris SPARC. The source tree is the same.
Erm did I miss something? Sun is not anti-linux, it's not anti OSS. It's a corporation which COMPETES with other companies. All Eric did was consider the apparently disparate philosophies behind Linux and Solaris development. Nowhere does he claim to be a Linux expert the same way most people who post here, judging by the content of their posts, could claim to be Sun or Solaris experts. Do RedHat not have to convince people to shell out money, do they not have 'marketing machinery'?