You really should look at Solaris Event Port Completion. It may not have everything you're looking for, but it should help. It does a good job of bridging what exists in Unix while adding a lot of functionality.
The referenced sun.com documents are quite old, and don't include anything in modern Solaris and OpenSolaris derived OSs.
Did you look at http://opensolaris.org/os/downloads/ and then follow the ON link to the page that explains in detail how to obtain either source and build, or obtain Sun's distro of OpenSolaris, Solaris Express Community Release?
I'd agree with many of your comments, but I have to disagree when it comes to saying there is nothing new about threading in Java and other modern languages. Just look at the concurrency utilities.
Regarding cache coherency, pretty much every modern, mainstream computer system guarantees cache coherency through some approach or another (NUMA, COMA designs). It's not just the domain of Intel.
And like I said in a previous post, I probably laid too much blame at the feet of C. My point was, with the post I replied to, that it's inaccurate to state threading is too complex to be of use in the kernel. Even C, with the appropriate tools, can be highly threaded and reliable. I believe I was clear about that.
Regardless, even if you get the language and code right, the big issues with scalable system design start to get much more into how memory is handled between the hardware and software than how the code is threaded.
I agree with this, and I probably laid too much blame at the feet of C. With some modern languages though, there has been some effort at reducing the complexity of writing multithreaded code.
Don't take C's poor support for threading and tools to build/debug threaded code to mean that writing threaded code isn't possible. Other platforms and languages have taken threads to great extremes for many years, and I'm not necessarily referring to anything Unix (or from Sun).
This reminds me of the story (but I don't know how true it is) that in the early days of Fortran, the quicksort algorithm was widely understood but considered to be too complicated to implement. Now 2nd year computer science students implement it as a homework project. Threads could be considered similar. Anyone who has written a servlet is implicitly writing multithreaded code and you can very easily/quickly write reliable and safe threaded code in a number of modern languages without having to get into the details C forces you into. It's the mix of pass-by-reference and pass-by-value with only a bit of syntactical sugar that creates the problems, not the concepts of parallelism.
On the other hand, I agree with you that we'll see increased parallelism driving increases in computing capabilities in the coming years. It was mathematically proven some time ago, but Amdahl's law is now officially giving way to Gustafson's law (more on John Gustafson here). Since software codes are sufficiently complex these days (even the most simple of modern programs can make use of parallelism-- just think of anything that touches a network), it's those platforms that exploit this feature which stand to deliver the best benefits to it's users.
Bummer... Sorry to hear that. My recommendation would be to first try going through the Sun Store (store.sun.com) 800 number. When you reach them, tell them you're looking for a local reseller.
Sun's model leverages partners for nearly all customers, especially for getting access to things like loaners.
So, if you were to ask most Sun folks, they'd be unhappy that _The Register_ used the moniker "Linux Killer". It just fans the flames.
Sun contributes to and supports Linux and FOSS with all of it's products. Sun always has. Here in my office there's a guy paid by Sun to work on the Linux kernel full time.
The reality is that some folks (both inside and outside Sun) see things in Solaris they like and want to use compared to the Linux kernel. Sun's fostering that.
Sun is also using Solaris to compete with Red Hat, Novell, IBM, etc. Is that unexpected? It's Red Hat that has spent the last few years trying to take marketshare from Sun.
"Compete on implementation, not on control of the specification". In other words, you should be free to write your app, then deploy it to your Unix (including Linux in that Unix definition) of choice. Make your choice based on whatever criteria works best for your needs. If it's Debian, so be it. If it's Red Hat, so be it. If it's an OpenSolaris based distribution, then so be it...
In Sun's process on Solaris, and I agree with this, all code commits are associated with a CR (change request/bugid). So if you introduce code (to the tree) on Monday with a bug, find it Tuesday, fix it Wednesday, you'll also file a bug and close it.
Major changes always have a "one pager" describing what you are looking to accomplish and why it should be done. In Sun, that has to be approved by a council.
I bet some of this practice will become part of the opensolaris community. Its good stuff IMHO. If you can't describe what you're planning to do in one page before you go create major new functionality, then perhaps it shouldn't be part of the overall project...
This is a trollish comment. Anyone who looks at the FAQ on the site and/or looks at the time/effort put into this will see Sun is trying to build upon the community around Solaris, not tear apart other communities.
Also, my understanding is you can cross pollenate ideas (that aren't patent encumbered) and/or contribute your own thoughts to both communities. Copyright (and joint copyright) of actual lines of code is where things get more complicated. It is, however, a false assertion that looking at OpenSolaris will open one's work to OpenSolaris.
It's the GPL that has 'viral' aspects. I'm not saying that the GPL is right or wrong in this area. It is different.
This whole area is a big venn diagram of copyright law, patent law, license agreements and those who are willing or not to persue litigation.
Quite right. Sun gets this quite well. Look for the articles on the Niagara processor. People always look at it as an SMP on a chip and try to compare it to hyperthreading or the stuff that IBM has done. However, what Sun is doing is 'fast switching' between threads based on stalls for memory access. This is another way of solving the problem you mention.
If the software and programming model are capable (and most software run on Solaris is) of exploiting this, you effecively trade off bandwidth (easy to obtain) for latency (very, very hard to obtain).
It's cool stuff-- I'm looking forward to seeing it released as a product.
This is quite off base in my opinion. Yes I work for Sun, but even if Sun weren't trying to push Linux into the market (which Sun is) Sun recognizes that anything that is unix is better for the market than Windows.
"Compete on implementation, not on control of the specification"
That's been one of Sun's mottos in this area for quite a while (possibly from even before Linux existed)-- and if you really think about it, it's a pretty good one. Sun has applied that kind of thinking to chips and OSs for years (there are other URLs too, those are just two examples). Before "open source" was en vogue, open systems were, and it's worthwhile to think about what the open systems market was all about.
To: peter_schwartz@gbn.com, doug_randall@gbn.com CC: rants@wiredmag.com
Peter/Doug,
I've just completed reading your article on the Wired website. I like to think I look at pretty much everything with a healthy amount of skepticism, and through those lenses, your article appears to be much more along the lines of advocacy.
In the intro, you state: "It's plentiful, clean, and"...
Calling Hydrogen plentiful because it exists in other molecular forms on the planet is simply inaccurate. By that argument, so are hydrocarbon type fuels (natural gas/oil), though admittedly the processes for making natural gas out of water and carbon dioxide are probably not well known.
I'm at least pleased to see that you admit (though don't retract that plentiful statement) further in the article that it turns out the most practical means of obtaining hydrogen in the world today is, in fact, fossil fuels. You do lay out some practical ideas for the source of the actual fuel (the Hydrogen in this article should have been more accurately addressed as a storage medium, not a fuel), and I commend you for doing so.
However, I would have to disagree with how you so quickly spend our tax dollars. It's as if you have something to gain from it!!
I would argue that the supply of renewable energy resources should be the focus of any efforts, absent of spending money on vehicles that use this 'fuel' that can not yet be mined, pumped or otherwise extracted from any natural source.
I would further argue that the point that all of our future energy needs will eventually be supplied by solar, wind, and hydro power, and that they just need to be "developed", can be tested pretty easily. The current energy usage can be pretty easily identified from information we already have on fuel usage. Then it's pretty easy to identify the maximum amount of energy we can gain from wind/hydro/solar power, and see how big the delta is.
I hope the chasm isn't too far to cross, but it's possible that it is. Far too often in the "press" I see assertions like yours without anything to back it up. You may well have the data to back it up, and you'd be doing very well with your readers if you were to cite that kind of information.
So here's to spending some pencil, paper and $10,000 testing your assertions, before we spend billions developing vehicles that have no source of "fuel".
My personal opinion? If the fuel were plentiful, the means to consume it would be developed. I'll entirely agree that it may take government involvement to get there, but we probably don't need to spend all of these billions on parallel paths. Focusing our efforts on "developing" the sources of energy is probably the best bet. Once we have the source of energy, the rest will fall into place.
I just thought I'd mention quickly that I moved to Xi's X server from XFree so I could switch to my external VGA port on my laptop.
An amazing consequence is that Mozilla is now a *very* snappy performer. Scrolling and incremental refresh are much better than they used to be. I don't know what motif libraries 4.x is linked against (lesstif?), I know it's the commercial libraries on other platforms. Anyway, it makes me wonder about the efficency of XFree or GTK.....
For others in search of speed, you may want to try it.
Sun has a program called BoB. It stands for the
best of the best. While you're in college, and just after you graduate, Sun invests in you by setting up a regimine of 70-90% training and the rest working with Sun people, shadowing engineers, etc. I'm sure there's info on the website, or you can contact the local team or the HR department. Also check out www.sunsandiego.com-- there are some engineering groups there if you're interested in hard core stuff.
And by the way, Sun is not "evil". Yes, perhaps I'm biased, but I don't think anyone that really knows the organization would think Sun is"evil".
I run linux on my Sun issued laptop-- and I know I'm not the only one. Unix excels at flexibility, so it's easy to integrate the systems. Use the right tool for the right job....
I just installed these things for the first time on friday, and was thoroughly impressed. They're working well in our industrial complex, despite poor placement on our part (haven't run any benchmarks yet, it could be dropping to a lower speed).
A couple observations: The gateway is really just a little bridge. Don't plan on getting 11Mbps out of it though-- it has a 10BaseT interface on the other side! Perhaps someone could put together a nice little embedded linux solution.:)
Also, you're back to shared media days. It's like putting everyone on a 10Mb hub. I don't know if there's any way to get it to scale (probably not if they all use one frequency) by breaking it down into more networks.
Which version are you running? In M11 they made the GUI quite a bit faster, though the mail/news components were still fairly slow. In M12, it's as fast as 4.x under Linux and Windows for me.
Some good suggestions out there, but one thing hasn't been mentioned. It's good to test the site with some kind of load generation tool (I think there's one at apache.org) when you're trying out different configurations *before* you go live. Every site is likely to be optimally tuned a bit different.
Also overlooked is possibly tuning the filesystem for caching and the like (file descriptors) and networking (maximum connections).
Possibly most of all, when I've seen performance problems, it's been due to how the code was written:).
I work with Solaris at work and Linux at home. The reason I got a job working with Solaris (on the big boxes mind you) is because of my experience with Linux.
I work for a Sun reseller, Sun supports their resellers selling and supporting machines with Linux on them. Sun is committed to delivering solutions on Solaris and for support and advanced computing reasons should continue with Solaris development.
To continue to give Unix the scalability and reliabilty with advanced reliability features like Alternate Pathing and Dynamic Reconfiguration that allow Unix systems to be used for the largest OLTP environments, we need companies like Sun. With regard to Unix computing, SGI has fuzzy plans these days, IBM has fuzzy plans these days, HP is very fuzzy these days. Sun is pretty much, last time I checked, the only company without some kind of deal with Microsoft to push NT boxes.
I honestly don't see how this is a bad thing. Some software packages might like to port to Unix, but don't want to go open source (for any one of a number of reasons). This gives them the vehicle. It's better than looking at the Unix world and saying they can't afford to port/support the myriad of OS variants.
I also don't see the anti-establishment banter helping. Support companies that support open computing: Sun's supported open computing for a long time.
Slashdot Mirror
You really should look at Solaris Event Port Completion. It may not have everything you're looking for, but it should help. It does a good job of bridging what exists in Unix while adding a lot of functionality.
The referenced sun.com documents are quite old, and don't include anything in modern Solaris and OpenSolaris derived OSs.
Did you look at http://opensolaris.org/os/downloads/ and then follow the ON link to the page that explains in detail how to obtain either source and build, or obtain Sun's distro of OpenSolaris, Solaris Express Community Release?
I'd agree with many of your comments, but I have to disagree when it comes to saying there is nothing new about threading in Java and other modern languages. Just look at the concurrency utilities.
Regarding cache coherency, pretty much every modern, mainstream computer system guarantees cache coherency through some approach or another (NUMA, COMA designs). It's not just the domain of Intel.
And like I said in a previous post, I probably laid too much blame at the feet of C. My point was, with the post I replied to, that it's inaccurate to state threading is too complex to be of use in the kernel. Even C, with the appropriate tools, can be highly threaded and reliable. I believe I was clear about that.
Regardless, even if you get the language and code right, the big issues with scalable system design start to get much more into how memory is handled between the hardware and software than how the code is threaded.
I agree with this, and I probably laid too much blame at the feet of C. With some modern languages though, there has been some effort at reducing the complexity of writing multithreaded code.
Don't take C's poor support for threading and tools to build/debug threaded code to mean that writing threaded code isn't possible. Other platforms and languages have taken threads to great extremes for many years, and I'm not necessarily referring to anything Unix (or from Sun).
This reminds me of the story (but I don't know how true it is) that in the early days of Fortran, the quicksort algorithm was widely understood but considered to be too complicated to implement. Now 2nd year computer science students implement it as a homework project. Threads could be considered similar. Anyone who has written a servlet is implicitly writing multithreaded code and you can very easily/quickly write reliable and safe threaded code in a number of modern languages without having to get into the details C forces you into. It's the mix of pass-by-reference and pass-by-value with only a bit of syntactical sugar that creates the problems, not the concepts of parallelism.
On the other hand, I agree with you that we'll see increased parallelism driving increases in computing capabilities in the coming years. It was mathematically proven some time ago, but Amdahl's law is now officially giving way to Gustafson's law (more on John Gustafson here). Since software codes are sufficiently complex these days (even the most simple of modern programs can make use of parallelism-- just think of anything that touches a network), it's those platforms that exploit this feature which stand to deliver the best benefits to it's users.
Bummer... Sorry to hear that. My recommendation would be to first try going through the Sun Store (store.sun.com) 800 number. When you reach them, tell them you're looking for a local reseller.
Sun's model leverages partners for nearly all customers, especially for getting access to things like loaners.
So, if you were to ask most Sun folks, they'd be unhappy that _The Register_ used the moniker "Linux Killer". It just fans the flames.
Sun contributes to and supports Linux and FOSS with all of it's products. Sun always has. Here in my office there's a guy paid by Sun to work on the Linux kernel full time.
The reality is that some folks (both inside and outside Sun) see things in Solaris they like and want to use compared to the Linux kernel. Sun's fostering that.
Sun is also using Solaris to compete with Red Hat, Novell, IBM, etc. Is that unexpected? It's Red Hat that has spent the last few years trying to take marketshare from Sun.
"Compete on implementation, not on control of the specification". In other words, you should be free to write your app, then deploy it to your Unix (including Linux in that Unix definition) of choice. Make your choice based on whatever criteria works best for your needs. If it's Debian, so be it. If it's Red Hat, so be it. If it's an OpenSolaris based distribution, then so be it...
In Sun's process on Solaris, and I agree with this, all code commits are associated with a CR (change request/bugid). So if you introduce code (to the tree) on Monday with a bug, find it Tuesday, fix it Wednesday, you'll also file a bug and close it.
Major changes always have a "one pager" describing what you are looking to accomplish and why it should be done. In Sun, that has to be approved by a council.
I bet some of this practice will become part of the opensolaris community. Its good stuff IMHO. If you can't describe what you're planning to do in one page before you go create major new functionality, then perhaps it shouldn't be part of the overall project...
This is a trollish comment. Anyone who looks at the FAQ on the site and/or looks at the time/effort put into this will see Sun is trying to build upon the community around Solaris, not tear apart other communities.
Also, my understanding is you can cross pollenate ideas (that aren't patent encumbered) and/or contribute your own thoughts to both communities. Copyright (and joint copyright) of actual lines of code is where things get more complicated. It is, however, a false assertion that looking at OpenSolaris will open one's work to OpenSolaris.
It's the GPL that has 'viral' aspects. I'm not saying that the GPL is right or wrong in this area. It is different.
This whole area is a big venn diagram of copyright law, patent law, license agreements and those who are willing or not to persue litigation.
Quite right. Sun gets this quite well. Look for the articles on the Niagara processor. People always look at it as an SMP on a chip and try to compare it to hyperthreading or the stuff that IBM has done. However, what Sun is doing is 'fast switching' between threads based on stalls for memory access. This is another way of solving the problem you mention.
If the software and programming model are capable (and most software run on Solaris is) of exploiting this, you effecively trade off bandwidth (easy to obtain) for latency (very, very hard to obtain).
It's cool stuff-- I'm looking forward to seeing it released as a product.
This is quite off base in my opinion. Yes I work for Sun, but even if Sun weren't trying to push Linux into the market (which Sun is) Sun recognizes that anything that is unix is better for the market than Windows.
"Compete on implementation, not on control of the specification"
That's been one of Sun's mottos in this area for quite a while (possibly from even before Linux existed)-- and if you really think about it, it's a pretty good one. Sun has applied that kind of thinking to chips and OSs for years (there are other URLs too, those are just two examples). Before "open source" was en vogue, open systems were, and it's worthwhile to think about what the open systems market was all about.
To: peter_schwartz@gbn.com, doug_randall@gbn.com
CC: rants@wiredmag.com
Peter/Doug,
I've just completed reading your article on the Wired website. I like
to think I look at pretty much everything with a healthy amount of
skepticism, and through those lenses, your article appears to be much
more along the lines of advocacy.
In the intro, you state:
"It's plentiful, clean, and"...
Calling Hydrogen plentiful because it exists in other molecular forms on
the planet is simply inaccurate. By that argument, so are hydrocarbon
type fuels (natural gas/oil), though admittedly the processes for making
natural gas out of water and carbon dioxide are probably not well known.
I'm at least pleased to see that you admit (though don't retract that
plentiful statement) further in the article that it turns out the most
practical means of obtaining hydrogen in the world today is, in fact,
fossil fuels. You do lay out some practical ideas for the source of the
actual fuel (the Hydrogen in this article should have been more
accurately addressed as a storage medium, not a fuel), and I commend you
for doing so.
However, I would have to disagree with how you so quickly spend our tax
dollars. It's as if you have something to gain from it!!
I would argue that the supply of renewable energy resources should be
the focus of any efforts, absent of spending money on vehicles that use
this 'fuel' that can not yet be mined, pumped or otherwise extracted
from any natural source.
I would further argue that the point that all of our future energy needs
will eventually be supplied by solar, wind, and hydro power, and that
they just need to be "developed", can be tested pretty easily. The
current energy usage can be pretty easily identified from information we
already have on fuel usage. Then it's pretty easy to identify the
maximum amount of energy we can gain from wind/hydro/solar power, and
see how big the delta is.
I hope the chasm isn't too far to cross, but it's possible that it is.
Far too often in the "press" I see assertions like yours without
anything to back it up. You may well have the data to back it up, and
you'd be doing very well with your readers if you were to cite that kind
of information.
So here's to spending some pencil, paper and $10,000 testing your
assertions, before we spend billions developing vehicles that have no
source of "fuel".
My personal opinion? If the fuel were plentiful, the means to consume
it would be developed. I'll entirely agree that it may take government
involvement to get there, but we probably don't need to spend all of
these billions on parallel paths. Focusing our efforts on "developing"
the sources of energy is probably the best bet. Once we have the source
of energy, the rest will fall into place.
Regards,
- Matt Ingenthron
I just thought I'd mention quickly that I moved to Xi's X server from XFree so I could switch to my external VGA port on my laptop.
An amazing consequence is that Mozilla is now a *very* snappy performer. Scrolling and incremental refresh are much better than they used to be. I don't know what motif libraries 4.x is linked against (lesstif?), I know it's the commercial libraries on other platforms. Anyway, it makes me wonder about the efficency of XFree or GTK.....
For others in search of speed, you may want to try it.
Sun has a program called BoB. It stands for the best of the best. While you're in college, and just after you graduate, Sun invests in you by setting up a regimine of 70-90% training and the rest working with Sun people, shadowing engineers, etc. I'm sure there's info on the website, or you can contact the local team or the HR department. Also check out www.sunsandiego.com-- there are some engineering groups there if you're interested in hard core stuff.
And by the way, Sun is not "evil". Yes, perhaps I'm biased, but I don't think anyone that really knows the organization would think Sun is"evil".
I run linux on my Sun issued laptop-- and I know I'm not the only one. Unix excels at flexibility, so it's easy to integrate the systems. Use the right tool for the right job....
Does anyone else find it interesting that the time on the update is 11:30PM? It's currently about...
:)
A couple things:
I just installed these things for the first time on friday, and was thoroughly impressed. They're working well in our industrial complex, despite poor placement on our part (haven't run any benchmarks yet, it could be dropping to a lower speed).
A couple observations: The gateway is really just a little bridge. Don't plan on getting 11Mbps out of it though-- it has a 10BaseT interface on the other side! Perhaps someone could put together a nice little embedded linux solution. :)
Also, you're back to shared media days. It's like putting everyone on a 10Mb hub. I don't know if there's any way to get it to scale (probably not if they all use one frequency) by breaking it down into more networks.
- Matt Ingenthron
Which version are you running? In M11 they made the GUI quite a bit faster, though the mail/news components were still fairly slow. In M12, it's as fast as 4.x under Linux and Windows for me.
Some good suggestions out there, but one thing hasn't been mentioned. It's good to test the site with some kind of load generation tool (I think there's one at apache.org) when you're trying out different configurations *before* you go live. Every site is likely to be optimally tuned a bit different.
:).
Also overlooked is possibly tuning the filesystem for caching and the like (file descriptors) and networking (maximum connections).
Possibly most of all, when I've seen performance problems, it's been due to how the code was written
I work with Solaris at work and Linux at home. The reason I got a job working with Solaris (on the big boxes mind you) is because of my experience with Linux.
I work for a Sun reseller, Sun supports their resellers selling and supporting machines with Linux on them. Sun is committed to delivering solutions on Solaris and for support and advanced computing reasons should continue with Solaris development.
To continue to give Unix the scalability and reliabilty with advanced reliability features like Alternate Pathing and Dynamic Reconfiguration that allow Unix systems to be used for the largest OLTP environments, we need companies like Sun. With regard to Unix computing, SGI has fuzzy plans these days, IBM has fuzzy plans these days, HP is very fuzzy these days. Sun is pretty much, last time I checked, the only company without some kind of deal with Microsoft to push NT boxes.
I honestly don't see how this is a bad thing. Some software packages might like to port to Unix, but don't want to go open source (for any one of a number of reasons). This gives them the vehicle. It's better than looking at the Unix world and saying they can't afford to port/support the myriad of OS variants.
I also don't see the anti-establishment banter helping. Support companies that support open computing: Sun's supported open computing for a long time.
I thought I saw once that Linux can mount UFS (probably not logging though)-- since UFS is Solaris's default fs, why not use UFS?