Greetings,
Have been looking at the specs of Boston Venom T4000.
Boston Venom T4000 http://www.bostonindia.in/products/bos-t4000.aspx
I could not figure out the lowest base price points, HA features and KVM support
Yet to work out the storage part of say 8TB storage (with HA features, of course) for this beast.
and accessability issues from devices.
I am in the middle of trying to find out about What AMD has been doing in its end.
and Lastly, wireless fencing devices
Yet to look into easy monitoring and management tools.
Any opinions please?
Regards,
Rajagopal
On 07/16/10 11:01 AM, Rajagopal Swaminathan wrote:
Greetings,
Have been looking at the specs of Boston Venom T4000.
Boston Venom T4000 http://www.bostonindia.in/products/bos-t4000.aspx
I could not figure out the lowest base price points, HA features and KVM support
is your goal a "server" or "supercomputing"? all that tesla stuff sorta says supercomputing, while HA etc says 'server'.
supercomputer clusters eschew HA in favor of having many independent compute units in a loose cluster that can tolerate any node dying by simply reassigning its last work unit to another node. only the persistent storage (usually a SAN or a clustered file system), and the cluster controller needs conventional HA.
if your goal is a 'server', then something from here would likely be more suitable. http://h10010.www1.hp.com/wwpc/us/en/sm/WF04a/15351-15351-3328412-241644-241... http://www-03.ibm.com/systems/x/hardware/rack/index.html http://www.dell.com/us/en/business/servers/rack_optimized/ct.aspx?refid=rack... http://www.dell.com/us/en/business/servers/rack_optimized/ct.aspx?refid=rack_optimized&s=bsd&cs=04 http://www.supermicro.com/products/system/2U/
along with the appropriate storage etc, depending on your requirements.
Greetings,
On 7/17/10, John R Pierce pierce@hogranch.com wrote:
On 07/16/10 11:01 AM, Rajagopal Swaminathan wrote:
is your goal a "server" or "supercomputing"? all that tesla stuff sorta says supercomputing, while HA etc says 'server'.
supercomputer clusters eschew HA in favor of having many independent compute units in a loose cluster that can tolerate any node dying by simply reassigning its last work unit to another node. only the persistent storage (usually a SAN or a clustered file system), and the cluster controller needs conventional HA.
I was thinking more about a "personal supercomputer in a cloud"
HA is a requirement for cloud. So I guess I have to think further about it.
ideas welcome.
Regards,
Rajagopal
On 07/16/10 7:12 PM, Rajagopal Swaminathan wrote:
HA is a requirement for cloud.
it is?
Most clouds have their own integral resilience that negates the need for conventional HA techniques.
but, really, "Cloud" is about the most meaningless buzzword in current usage. it can mean anything from renting Amazon E2 server instances, to deploying distributed applications on google's BigTable infrastructure to moving your older physical servers to vmware esx farms.
Greetings,
On 7/17/10, John R Pierce pierce@hogranch.com wrote:
On 07/16/10 7:12 PM, Rajagopal Swaminathan wrote:
Perhaps I have a distorted understanding of the cloud and HA technologies.
But i have been looking at the GPU based SCs.
I am looking at creating/building a Centos based infrastructure which would have which can have secure, private clouds which can host secure, private supercomputers.
I do realise that Sun had almost realised its "Net is the computer" dream. I have not followed up on their self healing system.
I have build Monitoring systems, HA clusters, VM Infrastructure etc. But they were all for entities who paid my salary. period.
Now I want every stakeholder and the less fortunate to benefit.
An affordable computing model for the common man.
Please realise that I am trying to do this in India, where anybody who wants to do have to face the hurdle of huge red tapes (exactly opposite to the model of "perl is the duct tape of Internet") that every government loves to do.
I posted the link as I felt that if a computer can be made "available" then a supercomputer too. I am technology agnostic as far as the solution is concerned.
I am in fact investigating the possibility of getting desiged a monitoring and control device for the CTOS Hardware. I have technical and business contacts which may help me to pull this off.
We are looKing at sub usd1000 infrastructure which would help young gamers, students, grandma and grandpa to have their own secure private computers as and when, where they want it and which should cost more than a typical utility bill from any device they choose to use.
My aim is to have it all in open source model. Including the revenue distribution.
Can we look at the model as perpetual revenue rolling rather than sharing?
Just to give an idea of the environment, an experienced server engineer gets less than usd150 per month after years of training and he may not know about HA, Virtualization techs, power availability is an act of god in most part of the country.
Last I heard, an APC fence device costs somewhere around 35K INR (~ usd 700). I mean INR 7000/socket is ridiculous price for a over-decorated power strip with less about basic cellphone type electronics with a little bit of power stuff thrown in. I know software projects and all that looting. but what the heck people should have access affordable computing.
On the instrumentation side, I was looking at arduino wireless, off band, so to say.
List, wish ourselves luck as if we fail, we have to answer our next genration for bad engineering.
I had seen one such project at VLSI Lab at IIT, Chennao during Shastra 2008 using some Open source hardware and eminently affordable
Regards,
Rajagopal
Raj,
Here is the problem.
All of us who've been in this industry for any period of time have heard an endless stream of "grand ideas". 99% never go anywhere at all. After a while, we can't help but get cynical and doubtful of yet another grand idea. Add to that your liberal use of "buzz words" and you plan is hard to believe.
If you really want to do this, you will need to show that you are serious *and* capable. You need to build the foundation yourself. You need to have concrete evidence; code, working prototypes, etc. Only then will people know that you are serious and can actually pull off your great idea.
I'm not telling you to give up. I am explaining why you are not being taken very seriously. If you really think you can do this, then start work and start asking specific questions. Then, when you can show the prototype, put together a business plan. Then pitch your idea again.
Rajagopal Swaminathan wrote:
My aim is to have it all in open source model. Including the revenue distribution.
Have you looked at Ubuntu's setup? I don't think it deals with GPU's but it might be an easier starting point than building from scratch, and if anything outgrows your resources it can move to Amazon's ec2.
http://www.ubuntu.com/cloud/private
But, I think you need to look at storage and compute facilities differently in a cloud model. Storage needs to be HA and redundant. Computing needs to be able to fail and be replaced.
Greetings,
Thanks Les for your reply. (I dont top-post normally, but this was an emergency)
On 7/17/10, Les Mikesell lesmikesell@gmail.com wrote:
Have you looked at Ubuntu's setup? I don't think it deals with GPU's but it might be an easier starting point than building from scratch, and if anything outgrows your resources it can move to Amazon's ec2.
The point is, why should we not have and use our own resources?
http://www.ubuntu.com/cloud/private
But, I think you need to look at storage and compute facilities differently in a cloud model. Storage needs to be HA and redundant. Computing needs to be able to fail and be replaced.
1. I am not able to understand your above statement clearly. (see my earlier replies on this thread)
2. I believe (and have experienced including and mother and father), that nothing and/or everything is (ir)replacable.
4. I am not confused (See point 1)
Regards,
rajagopal
Rajagopal Swaminathan wrote:
Greetings,
Thanks Les for your reply. (I dont top-post normally, but this was an emergency)
Emergency? Sorry, but your posts are leading me to think that you have lost it.
Have you looked at Ubuntu's setup? I don't think it deals with GPU's but it might be an easier starting point than building from scratch, and if anything outgrows your resources it can move to Amazon's ec2.
The point is, why should we not have and use our own resources?
/me blinks.
http://www.ubuntu.com/cloud/private
But, I think you need to look at storage and compute facilities differently in a cloud model. Storage needs to be HA and redundant. Computing needs to be able to fail and be replaced.
- I am not able to understand your above statement clearly. (see my
earlier replies on this thread)
- I believe (and have experienced including and mother and father),
that nothing and/or everything is (ir)replacable.
/me stares.
- I am not confused (See point 1)
Does not seem to be a matter of confusion.
You start on this list with a sharing model that sounds akin to time-sharing investments of say a private yacht. Then you post a string of stuff that are rather general, controversial, ill-informed but most of all, nothing to do with Centos. Keep this up and you'll be getting the boot. You've been 'warned' already by Karanbir.
Am 18.07.2010 um 00:14 schrieb Rajagopal Swaminathan:
Greetings,
Thanks Les for your reply. (I dont top-post normally, but this was an emergency)
On 7/17/10, Les Mikesell lesmikesell@gmail.com wrote:
Have you looked at Ubuntu's setup? I don't think it deals with GPU's but it might be an easier starting point than building from scratch, and if anything outgrows your resources it can move to Amazon's ec2.
The point is, why should we not have and use our own resources?
The (other) point is, I cannot really see the (technical) point of your posting(s). It seems non-technical. As such, the discussion has no real place on this list, I'm afraid.
Look, it's perfectly understandable that you in India might want to do your own thing. But maybe you want to found your own mailing-list for discussing that? (Please refrain from subscribing us on it, OK?)
There cannot come anything good out of this thread from here on, if there ever could...
Rajagopal Swaminathan wrote:
Greetings,
Thanks Les for your reply. (I dont top-post normally, but this was an emergency)
On 7/17/10, Les Mikesell lesmikesell@gmail.com wrote:
Have you looked at Ubuntu's setup? I don't think it deals with GPU's but it might be an easier starting point than building from scratch, and if anything outgrows your resources it can move to Amazon's ec2.
The point is, why should we not have and use our own resources?
If you had read the link you would have seen that it does use your own resources - that's why it is called a private cloud. Or did you mean that CentOS should have an equivalent to match the Ubuntu configuration? You probably could, but it would be a lot more work.
But, I think you need to look at storage and compute facilities differently in a cloud model. Storage needs to be HA and redundant. Computing needs to be able to fail and be replaced.
- I am not able to understand your above statement clearly. (see my
earlier replies on this thread)
Distributed (super) computing normally involves techniques that break the work up into independent units. If a node doesn't complete a unit it can be given to a different node.
- I believe (and have experienced including and mother and father),
that nothing and/or everything is (ir)replacable.
- I am not confused (See point 1)
Storage, on the other hand, needs to be reliable. The hadoop project is an approach at distribution, but the interface is hard to deal with.
Greetings,
On 7/18/10, Les Mikesell lesmikesell@gmail.com wrote:
Rajagopal Swaminathan wrote:
On 7/17/10, Les Mikesell lesmikesell@gmail.com wrote:
Dear Les,
Or did you mean that CentOS should have an equivalent to match the Ubuntu configuration?
Exactly. and I poised to do that as Centos x86_64 was attached thanks to this month's Linux for You magazine's usual attachment CD/DVD.
You probably could, but it would be a lot more work.
Of course, and I am willing to share my work with this list along with whatever income that incurr with it.
(Moderator: Kindly hange the subject line restrict my posting rights if I have erred)
Distributed (super) computing normally involves techniques that break the work up into independent units. If a node doesn't complete a unit it can be given to a different node.
I know that.
Storage, on the other hand, needs to be reliable. The hadoop project is an approach at distribution, but the interface is hard to deal with.
Indeed and of course.
Regards,
Rajagopal
On Saturday 17 July 2010, Rajagopal Swaminathan wrote:
Greetings,
On 7/17/10, John R Pierce pierce@hogranch.com wrote:
On 07/16/10 11:01 AM, Rajagopal Swaminathan wrote:
is your goal a "server" or "supercomputing"? all that tesla stuff sorta says supercomputing, while HA etc says 'server'.
supercomputer clusters eschew HA in favor of having many independent compute units in a loose cluster that can tolerate any node dying by simply reassigning its last work unit to another node. only the persistent storage (usually a SAN or a clustered file system), and the cluster controller needs conventional HA.
I was thinking more about a "personal supercomputer in a cloud"
HA is a requirement for cloud. So I guess I have to think further about it.
Lol, saturday morning entertainment :-)
Combining HA + "personal supercomputer" + "in a cloud" + using GPUs sure maxed out my troll-o-meter. Very creative.
/Peter
Greetings,
On 7/17/10, Peter Kjellstrom cap@nsc.liu.se wrote:
On Saturday 17 July 2010, Rajagopal Swaminathan wrote: Lol, saturday morning entertainment :-)
Combining HA + "personal supercomputer" + "in a cloud" + using GPUs sure maxed out my troll-o-meter. Very creative.
duh..
Q. If I can compute in cloud, in which cloud can I supercompute at an affordable or sponsored cost?
I may be a gamer, AE student, Graphic designer, renderer and so forth.
Now which part of above you did not understand?
sigh... how I hate my (in)ability to communicate tech to people inspite of having been kindly classified as a "lightweight fossil" in this list? Why elders are made to shout when they don't want to?
Regards,
Rajagopal
On 07/17/10 2:11 AM, Rajagopal Swaminathan wrote:
Q. If I can compute in cloud, in which cloud can I supercompute at an affordable or sponsored cost?
what is it you want to compute? what hardware resources do you want to bring to the table? what is this cloud of which you speak?
From all what I gather, its not that easy to use the Tesla GPU-as-a-processor stuff, its really only suitable for a specific class of problems, it won't virtualize (I think you implied in one of your vague
I may be a gamer, AE student, Graphic designer, renderer and so forth.
and that helps define your requirements... how?
Now which part of above you did not understand?
all the parts you left out.
Greetings,
Thanks for all your replies.
On 7/17/10, John R Pierce pierce@hogranch.com wrote:
On 07/17/10 2:11 AM, Rajagopal Swaminathan wrote:
Q. If I can compute in cloud, in which cloud can I supercompute at an affordable or sponsored cost?
what is it you want to compute?
anything that is computable.
what hardware resources do you want to bring to the table?
Whatever minimal cost hardware resources that are locally available.
what is this cloud of which you speak?
The one that I want to attempt to experiment with.
From all what I gather, its not that easy to use the Tesla GPU-as-a-processor stuff, its really only suitable for a specific class of problems, it won't virtualize (I think you implied in one of your vague
now *that* comes to slightly closer to my answer. But then, I am slightly confused after seeing some of the things on the web in the nvidia site. I will have to search again and will post the link which confused me.
and that helps define your requirements...
partially, yes.
how?
Try one of cellphone transmitter simulations in terms of Radiative Transfer (Check out University of Chicago's Prof. S. Chandrashekhar's textbook)
all the parts you left out.
Like?
Regards,
Rajagopal
On 7/17/2010 2:11 AM, Rajagopal Swaminathan wrote:
Q. If I can compute in cloud, in which cloud can I supercompute at an affordable or sponsored cost?
I may be a gamer, AE student, Graphic designer, renderer and so forth.
Now which part of above you did not understand?
sigh... how I hate my (in)ability to communicate tech to people inspite of having been kindly classified as a "lightweight fossil" in this list? Why elders are made to shout when they don't want to?
Everything you listed is interactive realtime or near-realtime graphics intensive. A cloud is not really suited to that kind of task to begin with. And you appear to be additionally attempting to find out if you could use an *existing* cloud (for example Amazon EC2) to do it - meaning not only are you talking about an architecture that isn't really suited to the problem, you are talking about putting it behind *SLOW* network connections to boot.
Never-mind how *fast* a cloud is (or is not), you can't move the rendered bits back and forth to a desktop over a remote network connection at any kind of sane speed.
Greetings,
Thanks for the interest shown by all the responders.
On 7/18/10, Jerry Franz jfranz@freerun.com wrote:
On 7/17/2010 2:11 AM, Rajagopal Swaminathan wrote: Everything you listed is interactive realtime or near-realtime graphics intensive. A cloud is not really suited to that kind of task to begin with.
I don't inderstand why it should be so.
How does the online gaming work? and who renders it?
And you appear to be additionally attempting to find out if you could use an *existing* cloud (for example Amazon EC2) to do it - meaning not only are you talking about an architecture that isn't really suited to the problem, you are talking about putting it behind *SLOW* network connections to boot.
Agreed, as EC2 performance as limited by last mile speed. But Then why should I be limited by something which is hosted not in India?
But again, I am thinkinking about the future. not present.
Never-mind how *fast* a cloud is (or is not), you can't move the rendered bits back and forth to a desktop over a remote network connection at any kind of sane speed.
Again, the questions that are raised by the "sane speed" are:
1. What if it is within a campus? 2. Why should Indian universities and its students be denied such a computing facility despite having fibre speed connectivity within campus/area? 3. Why we (the Centos community) should hold ourselves hostage to vendor grip? 4. Why it should not be private?
I am clear that it is a technological possibility (rather, more of a probability).
Regards,
Rajagopal
On Sun, Jul 18, 2010 at 03:01:32AM +0530, Rajagopal Swaminathan wrote:
On 7/18/10, Jerry Franz jfranz@freerun.com wrote:
Everything you listed is interactive realtime or near-realtime graphics intensive. A cloud is not really suited to that kind of task to begin with.
I don't inderstand why it should be so.
How does the online gaming work? and who renders it?
This is where what you don't know is hurting you. Online gaming works because the gamers have computers - whether PCs or game machines - with considerable resources. It will not work to a dumb terminal or thin client.
And you appear to be additionally attempting to find out if you could use an *existing* cloud (for example Amazon EC2) to do it - meaning not only are you talking about an architecture that isn't really suited to the problem, you are talking about putting it behind *SLOW* network connections to boot.
Agreed, as EC2 performance as limited by last mile speed. But Then why should I be limited by something which is hosted not in India?
It's the "last mile" to the client that matters. Where the hosts are isn't the problem or the solution.
Never-mind how *fast* a cloud is (or is not), you can't move the rendered bits back and forth to a desktop over a remote network connection at any kind of sane speed.
- Why should Indian universities and its students be denied such a
computing facility despite having fibre speed connectivity within campus/area?
What is this, a social justice argument? If you want good computers for students and house holders, go to One Laptop Per Child. If you want good computers for university students, give them a budget to get ahold of the parts and build their own. You can build an incredibly powerful system, for local use, for very little money, even in rupees.
I am clear that it is a technological possibility (rather, more of a probability).
So is doing this all with quantum-computing nanomachines hovering invisibly in the air around us. However let me say with considerable confidence that those won't run CentOS. And if they form a cloud, it will be totally unlike any cloud we know today.
Whit
Greetings,
On 7/16/10, Rajagopal Swaminathan raju.rajsand@gmail.com wrote:
<fact of an incomeless (ex-) professional who has a debt burden of about $2000> Despite all the discussions, I have not been able spend even 1 INR during the duration as I just don't have it. </fact of an incomeless (ex-) professionala debt burden of about $2000>
Regards,
Rajagopal
-
Chan Chung Hang Christopher -
Digimer -
Jerry Franz -
John R Pierce -
Karanbir Singh -
Les Mikesell -
m.roth＠5-cent.us -
Peter Kjellstrom -
Rainer Duffner -
Rajagopal Swaminathan -
Whit Blauvelt