I fail to see your objectivity.. what is your basis for calling Solaris clunky ? You haven't explained anything towards your comment..
I've been using both Linux and Solaris for over 12 years now.. I can share this with you, I haven't had a kernel panic due to a driver in Solaris since 1998.
Not the case with Linux. I don't need support for all the funky hardware in Solaris. What I need is sure-shot reliability and predicatibility. And clear information when stuff does break. I can tell you Linux today is no where near being reliable/predictable or debuggable.
I run a ton of Linux boxes and for those purposes which I don't care for reliability or predictability or where I don't need to troubleshoot. Would I trust Linux with my mission critical applications.. NO. Would I trust Linux with my webserver.. Yes.
Choose it, Use it.
Sun software on the other hand (Solaris) I tend to have very little respect for. It works and works well, but is clunky and very outdated IMHO.
You've gotta be kidding me or you are the kind that thinks my embraer EJ-145 gets me from SFO to Dallas (because thats what your realm activities are limited to) and you can do your job so all the Boeing 747s or Airbus 300 are overatted in your IMHO...
Linux is no-where near the heavy lifting capabilites of Solaris 10. Just because bash isn't installed as the default root shell does not make Solaris outdated.
It doesn't wear the lipstick or the new hot mini-skirt that linux flashes.. but it certainly can do the job in day and night over and over, sideways and whichever ways and still be willing to service more..
Solaris 10 ships with/usr/sfw/bin utils (i.e. gnu utils).. this can be resolved with a simple 'exec bash' and the backspace 'stty erase ^H'.. simple unix stuff (agreed quirky, but thats no way the determining the factor to qualify an OS or otherwise)
And you can set user shells to which ever you want. Only the root shell is defaulted to statically linked 'sh' so if for some reason your/lib gets trashed you still have access to a working shell.
Just because you can't scroll history with you up/down arrows or you can't get backspace to work the way you want doesn't mean Solaris is immature.
If you get a chance, explore its DR capabilities or Zones. Also, if you ever come across a kernel panic due to a Solaris shipped driver.. I will take back everything I said about Solaris being mature. Agreed it doesn't support all the hardware you want it to, but it does a few and does it extremely well. And well enough to trust my multi-million dollar OLTP stuff and be able to sleep soundly at night.
I have worked extensively with both Linux and Solaris (and AIX, HPUX and OSF/1 before).. Linux is lightyears behind any of the unixes, but yes it does the job for a webserver or any of the simple tcp related services.. When it comes to trusting my money with it, it just ain't there.
Why not use Solaris?
It works out of the box, its free, its supported by a brilliant engineering, opensource, very modular, very advanced, very easy to use and very stable..
Why would you want to waste time with something immature when you have a mature alternative ?
Let me try and jerk out into reality.
Here is the press release from Mellanox announcing $69 per IB HCA chip in volume. http://www.mellanox.com/news/press_releases/pr_030 105.php Agreed its not widespread yet. But there are systems manufacturers actively looking at deploying this silicon on the mother board. IWILL is one company thats already started. Asus, Tyan are next.
The reality is that there is only one *correct* way to do a fast interconnect, and that is to build it into the CPU itself. Oh wait, AMD intends to do just that!
You do realize on board interconnects can only scale so much.. before you run into serious data coherency issues. Intel Xeon's can scale up to 4 processors per board. IBM has tweaked so they can scale up to 8 procs on their proprietory motherboards. AMD Hypertransport allows you to scale upto 8 sockets per motherboard. To scale beyond that you have to go use NUMA like interconnect (pricey) or push it on a fast pipe to other boxes (not so pricey). AMD recommends this. AMD specifically recommends to use InfiniBand to push high bandwidth. Go talk to AMD.
Hypertransport is available *today* in the market. Go talk to supermicro. You don't have to wait for it.
All this other junk has a shelf life of maybe a few years at best (as does pretty much all networking gear, but the difference is that this stuff costs 10 times as much).
Pretty much everything in computers has a shelf life of few years hardware and software.. what is your point ?
It's a huge waste of time for all but the most extreme clustered applications for which there is no algorithmic solution to the latency issue (read: people like to throw hardware at badly written programs more often then they should).
People are not complete idiots when they are spending millions of dollars setting up 100's or even 1000's of nodes of beowulf clusters over IB.
There are specific applications that are latency sensitive and bandwidth hungry. And that is the segment that IB addresses very well and where ethernet is considered an outsider.
Tell me you want to use 1G ethernet to process a satellite feed coming in at 3.2 Gbps.. again, I think you are either clueless or ignorant.. just because IB doesn't fit your bill doesn't mean it doesn't have its play.
I think you have no clue about what your saying.
1) InfiniBand is an open standard hosted by IBTA which is a consortium of companies. The spec is available for anyone who wants to understand/build InfiniBand hardware. Not IEEE does not make it proprietary.
2) The major roadblock with 10Gbps is physics. You can only reach so far with copper without retiming the signal. And optics are expensive. 10 GbE has the same problem and it won't be cheap any time soon.
3) InfiniBand has already reached a volume where on-board IB chips are available in $70-80 range.. 10 GbE is no where close. And IB DDR will be shipping next month (20 Gbps wire / 16 Gbps data).
4) Beowulfs are popular for a reason.. Cache Coherency is a bitch.
5) A round trip node-to-node latency in IB is 2.7 usecs (best case of course). With all the optimization in the world, you won't be able to get ethernet anywhere near that number.
6) InfiniBand is being WIDELY deployed. Sandia Thunderbird is a 9216 processor IB fabric in production. NCSA has Tungsten2 which is 1024 processor IB fabric. NCSA also has a Microsoft Windows Cluster running CCE over IB with 880 processors. There are several large firms Oil&Gas, BioTech, Banks, Market Data houses which run several large multi-hundred/multi-thousand processor IB clusters.
7) Just as with any technology it will take time for new technologies to be accessible to the masses.. so don't write off anything yet.
8) Do you research before you open your mouth.
"Beowulf can mean cheap hardware, Sun doesn't." -- Wrong! Sun X64 servers are the least expensive and the most ideal (branded) boxes for Beowulf type cluster.
"Trusted/Secure Solaris adds huge amount of overheads to installing and configuring and using a system that they just might not need." -- Wrong again. Trusted Solaris is or for that matter securing any Unix like OS is about similar effort. Now the difference is with Solaris, you have a vendor (ie Sun) willing to put their money where their mouth is.. can you say the same for Linux ?
That said, If as an administrator (presumably) you think securing your boxes is too much work, you ought not to be an administrator to begin with.
"In my opinion you have forced your customer to make a move on questionable grounds.. and in my opinion, you're looking at it from a loosing vendor's point of view, talking about what's bad in this solution." -- It is clear the topic poster hadn't done his homework before suggesting the migration. This itself is enough evidence to point out the suggestion is questionable.
That said again, have you looked into Solaris 10 RBAC ? (Role Based Access Control).. And as the parent poster mentioned, if you are running MPI jobs, you do not need to be a root user.
If cost of concern and hence the need for Beowulf type cluster, the user might as well go about with Solaris 10 beowulf cluster with Sun MPI-2 stack and Sun Compiler (available for free), which I would think would be a lot less painful than moving to Linux and mucking around for a few months to get the cluster up and running.. And yes, I build supercomputers for a living.
Do you really want the future of web processing to be entirely web based and saved on somebody else's machine? G-mail bothers me like that -- even though I pretty much use it exclusively for e-mail now.
Go back a 100 years and read it this way: Do you really want the future of money transactions to be entirely banking based and saved in somebody else's safes? Banking bothers me like that -- even though I pretty much use it exclusively to safe-keep my money now.
Or would you rather be happy with your money buried in a hole in your backyard ?
Slashdot Mirror
I fail to see your objectivity .. what is your basis for calling Solaris clunky ? You haven't explained anything towards your comment ..
I've been using both Linux and Solaris for over 12 years now .. I can share this with you, I haven't had a kernel panic due to a driver in Solaris since 1998.
Not the case with Linux. I don't need support for all the funky hardware in Solaris. What I need is sure-shot reliability and predicatibility. And clear information when stuff does break. I can tell you Linux today is no where near being reliable/predictable or debuggable.
I run a ton of Linux boxes and for those purposes which I don't care for reliability or predictability or where I don't need to troubleshoot. Would I trust Linux with my mission critical applications .. NO. Would I trust Linux with my webserver .. Yes.
Choose it, Use it.
Sun software on the other hand (Solaris) I tend to have very little respect for. It works and works well, but is clunky and very outdated IMHO.
...
Linux is no-where near the heavy lifting capabilites of Solaris 10. Just because bash isn't installed as the default root shell does not make Solaris outdated.
It doesn't wear the lipstick or the new hot mini-skirt that linux flashes .. but it certainly can do the job in day and night over and over, sideways and whichever ways and still be willing to service more ..
You've gotta be kidding me or you are the kind that thinks my embraer EJ-145 gets me from SFO to Dallas (because thats what your realm activities are limited to) and you can do your job so all the Boeing 747s or Airbus 300 are overatted in your IMHO
Solaris 10 ships with /usr/sfw/bin utils (i.e. gnu utils) .. this can be resolved with a simple 'exec bash' and the backspace 'stty erase ^H' .. simple unix stuff (agreed quirky, but thats no way the determining the factor to qualify an OS or otherwise)
And you can set user shells to which ever you want. Only the root shell is defaulted to statically linked 'sh' so if for some reason your /lib gets trashed you still have access to a working shell.
Just because you can't scroll history with you up/down arrows or you can't get backspace to work the way you want doesn't mean Solaris is immature.
If you get a chance, explore its DR capabilities or Zones. Also, if you ever come across a kernel panic due to a Solaris shipped driver .. I will take back everything I said about Solaris being mature. Agreed it doesn't support all the hardware you want it to, but it does a few and does it extremely well. And well enough to trust my multi-million dollar OLTP stuff and be able to sleep soundly at night.
I have worked extensively with both Linux and Solaris (and AIX, HPUX and OSF/1 before) .. Linux is lightyears behind any of the unixes, but yes it does the job for a webserver or any of the simple tcp related services .. When it comes to trusting my money with it, it just ain't there.
Why not use Solaris? It works out of the box, its free, its supported by a brilliant engineering, opensource, very modular, very advanced, very easy to use and very stable ..
Why would you want to waste time with something immature when you have a mature alternative ?
Let me try and jerk out into reality. Here is the press release from Mellanox announcing $69 per IB HCA chip in volume. http://www.mellanox.com/news/press_releases/pr_030 105.php
.. before you run into serious data coherency issues. Intel Xeon's can scale up to 4 processors per board. IBM has tweaked so they can scale up to 8 procs on their proprietory motherboards. AMD Hypertransport allows you to scale upto 8 sockets per motherboard. To scale beyond that you have to go use NUMA like interconnect (pricey) or push it on a fast pipe to other boxes (not so pricey). AMD recommends this. AMD specifically recommends to use InfiniBand to push high bandwidth. Go talk to AMD.
Hypertransport is available *today* in the market. Go talk to supermicro. You don't have to wait for it.
All this other junk has a shelf life of maybe a few years at best (as does pretty much all networking gear, but the difference is that this stuff costs 10 times as much).
Pretty much everything in computers has a shelf life of few years hardware and software .. what is your point ?
It's a huge waste of time for all but the most extreme clustered applications for which there is no algorithmic solution to the latency issue (read: people like to throw hardware at badly written programs more often then they should).
People are not complete idiots when they are spending millions of dollars setting up 100's or even 1000's of nodes of beowulf clusters over IB.
There are specific applications that are latency sensitive and bandwidth hungry. And that is the segment that IB addresses very well and where ethernet is considered an outsider.
Tell me you want to use 1G ethernet to process a satellite feed coming in at 3.2 Gbps .. again, I think you are either clueless or ignorant .. just because IB doesn't fit your bill doesn't mean it doesn't have its play.
Agreed its not widespread yet. But there are systems manufacturers actively looking at deploying this silicon on the mother board. IWILL is one company thats already started. Asus, Tyan are next.
The reality is that there is only one *correct* way to do a fast interconnect, and that is to build it into the CPU itself. Oh wait, AMD intends to do just that!
You do realize on board interconnects can only scale so much
I think you have no clue about what your saying. 1) InfiniBand is an open standard hosted by IBTA which is a consortium of companies. The spec is available for anyone who wants to understand/build InfiniBand hardware. Not IEEE does not make it proprietary. 2) The major roadblock with 10Gbps is physics. You can only reach so far with copper without retiming the signal. And optics are expensive. 10 GbE has the same problem and it won't be cheap any time soon. 3) InfiniBand has already reached a volume where on-board IB chips are available in $70-80 range .. 10 GbE is no where close. And IB DDR will be shipping next month (20 Gbps wire / 16 Gbps data).
4) Beowulfs are popular for a reason .. Cache Coherency is a bitch.
5) A round trip node-to-node latency in IB is 2.7 usecs (best case of course). With all the optimization in the world, you won't be able to get ethernet anywhere near that number.
6) InfiniBand is being WIDELY deployed. Sandia Thunderbird is a 9216 processor IB fabric in production. NCSA has Tungsten2 which is 1024 processor IB fabric. NCSA also has a Microsoft Windows Cluster running CCE over IB with 880 processors. There are several large firms Oil&Gas, BioTech, Banks, Market Data houses which run several large multi-hundred/multi-thousand processor IB clusters.
7) Just as with any technology it will take time for new technologies to be accessible to the masses .. so don't write off anything yet.
8) Do you research before you open your mouth.
"Beowulf can mean cheap hardware, Sun doesn't." -- Wrong! Sun X64 servers are the least expensive and the most ideal (branded) boxes for Beowulf type cluster. .. can you say the same for Linux ? .. and in my opinion, you're looking at it from a loosing vendor's point of view, talking about what's bad in this solution." -- It is clear the topic poster hadn't done his homework before suggesting the migration. This itself is enough evidence to point out the suggestion is questionable. .. And as the parent poster mentioned, if you are running MPI jobs, you do not need to be a root user. .. And yes, I build supercomputers for a living.
"Trusted/Secure Solaris adds huge amount of overheads to installing and configuring and using a system that they just might not need." -- Wrong again. Trusted Solaris is or for that matter securing any Unix like OS is about similar effort. Now the difference is with Solaris, you have a vendor (ie Sun) willing to put their money where their mouth is
That said, If as an administrator (presumably) you think securing your boxes is too much work, you ought not to be an administrator to begin with.
"In my opinion you have forced your customer to make a move on questionable grounds
That said again, have you looked into Solaris 10 RBAC ? (Role Based Access Control)
If cost of concern and hence the need for Beowulf type cluster, the user might as well go about with Solaris 10 beowulf cluster with Sun MPI-2 stack and Sun Compiler (available for free), which I would think would be a lot less painful than moving to Linux and mucking around for a few months to get the cluster up and running
Do you really want the future of web processing to be entirely web based and saved on somebody else's machine? G-mail bothers me like that -- even though I pretty much use it exclusively for e-mail now.
Go back a 100 years and read it this way: Do you really want the future of money transactions to be entirely banking based and saved in somebody else's safes? Banking bothers me like that -- even though I pretty much use it exclusively to safe-keep my money now. Or would you rather be happy with your money buried in a hole in your backyard ?