We use Linux to teach the Operating Systems class now at Johns Hopkins. This is new for last semester and it went rather well. Of course VMware wasn't nice enough to GIVE us licenses. I think we made a good impact on students and our projects were very good examples of how to learn kernel internals and operating system fundamentals.
I designed the assignments (the second was the hardest and needs revision or replacement), but I think it went very well and we had a positive response from students. I hope we do it the same way next year!
It is a module that incurs almost NO overhead. You can mark directories or locations with Load Balancing policies and BOOM. That is it. It communicates with other Apache servers via multicast and handles the rest. You can even plug in your own decision making algorithms. It is super simple to load balance cgi-scripts to some machines, mod_perl database script to another set and images based on a completely different policy. Or just use our default;)
It curently runs under Linux and Solaris, but the next release will support BSDI as well.
It is a software solution that can be combined with any hardware solution you choose (if you need that too). You can't loose with this. The install process and set up time combined is very minimal.
Many distributed systesm can benefit greatly from a good group communications system. It provides a network architecture that allows for reliable, effecient network communications between many hosts as well as other useful features. You get the notion of groups and without much effort, you system can gracefully handle network paritions and systems coming up and down.
SPREAD is a great systems. It provides a consistent API on all platforms and is open source. REALLY FAST TOO! It is a client/server architecture and client can by Java or C or C++ (our port it to your favorite language!) Server is in C.
Last time I checked the big old NSA HQ. With the BIG rocks and fences and armed security guards is not in D.C. Rather it is 20 miles north of D.C.'s outter beltway loop; between Columbia and Odenton. (Sharing land with the Fort George Meade (Fort Meade) military base. I pass by it every day.
If you had visited this you would know it is a hike from D.C. (no metro or anything like that)
Just thought I would set that part of the record straight.
Unfortunately, top tier doesn't really apply to CS grad schools as it does to undergraduate education. As an undergraduate, you study computer science. As a PhD tracked graduate student your areas of interest will be much more particular. So now, this "top tier" applies only to individual fields inside computer science. For example, if you think that you would like to do research in MITs media labs, then that is one of the best places in the world to do so. However, if you want to do research in networks and group communications, look at JHU or Cornell where they have established research centers and EXCELLENT faculty that work in those areas.
Sounds to me like you have a lot of research to do. Also, don't hesitate to fly out to a school to check it out (small investment in the big picture). If the faculty member you are interested in working with is not receptive, you can make a big part of you decision based on that and the environment you see there.
I actually am pursuing my PhD at the school I attended for my BS and my MSE. I knew the environemtn here and knew that it was right for me and what I want to research.
So, if someone goes to work for 8 hours a day, they are addicted to it? I think that the phsycologist is perhaps on crack?! I am on the Internet for about 8-14 hours a day. That is how I make a living. My research, my job and many social acquaitances are on the Net. It is a different medium for communications. I find it much more disturbing when my wife talks on the phone for 4 hours at a time.
Working with mail servers and news servers that have massive amounts of small files (all less that 1MB, most around 4k), we see a lot of problems with ext2. When we have a systems charash (though it be rare) we must get up and running immediately.
My experience with ext2 exhibits extrememly painful fscks and information loss due to controller errors (the person bringing ir back online and walking it through the fsck).
On the other hand, XFS (no experience with JFS) is EXTEMELY quick about these things. I can bring online my 50 GB file system in minutes rather than hours.
To address the XFS is slower issue, I do a lot of I/O intensive processing and I find that ext2 is faster for somethings and slower for others, but neither is substantially better.
We need solutions to handling incoming/outgoing messages at about ~500k hour on a single machine, and we con't have the resources for solid state drives.
As we run linux on those machines, we have purchase many small (2-4GB) SCSI LVD drives and split the mail spool over all of them (14). We run ext2 and this FLY!
I personally believe that once XFS is available for Linux, I wil not have to do this kludgery and I can move back to having my spool on 1-3 disks.
Using XFS on my SGI's running IRIX has been nothing but a pleasant expereince. Using IRIX itself has been less than pleasant (not painful, though).
The whole idea is to be able to have console access. You plain and simple cannot do this over telnet. If you are running 18 boxes like I am and are hacking networking drivers or other OS things, it is a pain to debug a crashed kernel or faulty hardware over the network when there is no networking stack (crashed kernel). You need true console access (SGIs and Sparcs do provide this). PCs don't (as of yet).
Actually, multicast is not the greatest solution if you are talking about teleconferencing. If you are talking about sending it to thousands of people, sure, but there are better tools out there for 10-20. Check out the SPREAD group communication toolkit from The Center for Networks and Distributed Systems at JHU. It is pretty powerful and simple to develope around.
-- my slashdot for your jesus
e^(-jx) or sqrt(x) functions probably not patented
on
Hyperbolic Trees
·
· Score: 1
Well, this may not be all bad. I would imagine that they don't have a patent on everything that looks remotely similar. So instead of a hyberbolic function, We can modify the domain and make it a parabolic function or something like that. It won't be the same, but it would probably be just as effective (and comfortable) for the user.
All these comments about how this is useful.. I agree... I think it would really be nice for utilities like CHEOPS.
And if it supports recursive trees (read cyclic) then it would be pretty nice for a man page reader and a web site browser (like LINBOT).
That would mean of course that we would need a GTK port. Doesn't sound impossible, or even that hard. And I would definitely use it.
Anybody game for coding this one up? Sounds like a GOOD, SHORT TERM, free software project, same license as GTK (of course).
Just for a moment, ponder something quite nice...
on
MS Office for Linux
·
· Score: 1
As a relaxing excerise for the reader who is usually frustrated with the death grip MS has on the industry... Think some happy (though unrealistic) thoughts. It calms tensions...
Wouldn't it be nice if:
after MS is broken into pieces, the "apps" side of the corporation decided to repent. I think I would really enjoy using Microsoft Office if it was made open source (even if not GPLed). Besides, I think that they don't have too much to loose by this. People will still use their product (those that already do) and it would help mend deep wounds inflicted on the free software community.
It should be something for little MSs to think about if and whne Ma-MS is broken.
Slashdot Mirror
We use Linux to teach the Operating Systems class now at Johns Hopkins. This is new for last semester and it went rather well. Of course VMware wasn't nice enough to GIVE us licenses. I think we made a good impact on students and our projects were very good examples of how to learn kernel internals and operating system fundamentals.
Check it out at CS318/CS418 class projects.
I designed the assignments (the second was the hardest and needs revision or replacement), but I think it went very well and we had a positive response from students. I hope we do it the same way next year!
If you like Apache, check out mod_backhand. It is a module load-balancer that is under development (but works well now) over at The Center for Networks and Distributed Systems at Johns Hopkins.
It is a module that incurs almost NO overhead. You can mark directories or locations with Load Balancing policies and BOOM. That is it. It communicates with other Apache servers via multicast and handles the rest. You can even plug in your own decision making algorithms. It is super simple to load balance cgi-scripts to some machines, mod_perl database script to another set and images based on a completely different policy. Or just use our default
It curently runs under Linux and Solaris, but the next release will support BSDI as well.
It is a software solution that can be combined with any hardware solution you choose (if you need that too). You can't loose with this. The install process and set up time combined is very minimal.
Of course, I am a little biased
-- Theo Schlossnagle
Many distributed systesm can benefit greatly from a good group communications system. It provides a network architecture that allows for reliable, effecient network communications between many hosts as well as other useful features. You get the notion of groups and without much effort, you system can gracefully handle network paritions and systems coming up and down.
SPREAD is a great systems. It provides a consistent API on all platforms and is open source. REALLY FAST TOO! It is a client/server architecture and client can by Java or C or C++ (our port it to your favorite language!) Server is in C.
Check it out at http://www.spread.org/
Last time I checked the big old NSA HQ. With the BIG rocks and fences and armed security guards is not in D.C. Rather it is 20 miles north of D.C.'s outter beltway loop; between Columbia and Odenton. (Sharing land with the Fort George Meade (Fort Meade) military base. I pass by it every day.
If you had visited this you would know it is a hike from D.C. (no metro or anything like that)
Just thought I would set that part of the record straight.
Unfortunately, top tier doesn't really apply to CS grad schools as it does to undergraduate education. As an undergraduate, you study computer science. As a PhD tracked graduate student your areas of interest will be much more particular. So now, this "top tier" applies only to individual fields inside computer science. For example, if you think that you would like to do research in MITs media labs, then that is one of the best places in the world to do so. However, if you want to do research in networks and group communications, look at JHU or Cornell where they have established research centers and EXCELLENT faculty that work in those areas.
Sounds to me like you have a lot of research to do. Also, don't hesitate to fly out to a school to check it out (small investment in the big picture). If the faculty member you are interested in working with is not receptive, you can make a big part of you decision based on that and the environment you see there.
I actually am pursuing my PhD at the school I attended for my BS and my MSE. I knew the environemtn here and knew that it was right for me and what I want to research.
Good luck in your decisions.
So, if someone goes to work for 8 hours a day, they are addicted to it? I think that the phsycologist is perhaps on crack?! I am on the Internet for about 8-14 hours a day. That is how I make a living. My research, my job and many social acquaitances are on the Net. It is a different medium for communications. I find it much more disturbing when my wife talks on the phone for 4 hours at a time.
This is ridiculous!
Working with mail servers and news servers that have massive amounts of small files (all less that 1MB, most around 4k), we see a lot of problems with ext2. When we have a systems charash (though it be rare) we must get up and running immediately.
My experience with ext2 exhibits extrememly painful fscks and information loss due to controller errors (the person bringing ir back online and walking it through the fsck).
On the other hand, XFS (no experience with JFS) is EXTEMELY quick about these things. I can bring online my 50 GB file system in minutes rather than hours.
To address the XFS is slower issue, I do a lot of I/O intensive processing and I find that ext2 is faster for somethings and slower for others, but neither is substantially better.
We need solutions to handling incoming/outgoing messages at about ~500k hour on a single machine, and we con't have the resources for solid state drives.
As we run linux on those machines, we have purchase many small (2-4GB) SCSI LVD drives and split the mail spool over all of them (14). We run ext2 and this FLY!
I personally believe that once XFS is available for Linux, I wil not have to do this kludgery and I can move back to having my spool on 1-3 disks.
Using XFS on my SGI's running IRIX has been nothing but a pleasant expereince. Using IRIX itself has been less than pleasant (not painful, though).
The whole idea is to be able to have console access. You plain and simple cannot do this over telnet. If you are running 18 boxes like I am and are hacking networking drivers or other OS things, it is a pain to debug a crashed kernel or faulty hardware over the network when there is no networking stack (crashed kernel). You need true console access (SGIs and Sparcs do provide this). PCs don't (as of yet).
KVMs are the only solution I have used reliably.
Actually, multicast is not the greatest solution if you are talking about teleconferencing. If you are talking about sending it to thousands of people, sure, but there are better tools out there for 10-20.
Check out the SPREAD group communication toolkit from The Center for Networks and Distributed Systems at JHU. It is pretty powerful and simple to develope around.
-- my slashdot for your jesus
Well, this may not be all bad. I would imagine that they don't have a patent on everything that looks remotely similar. So instead of a hyberbolic function, We can modify the domain and make it a parabolic function or something like that. It won't be the same, but it would probably be just as effective (and comfortable) for the user.
All these comments about how this is useful.. I agree... I think it would really be nice for utilities like CHEOPS.
And if it supports recursive trees (read cyclic) then it would be pretty nice for a man page reader and a web site browser (like LINBOT).
That would mean of course that we would need a GTK port. Doesn't sound impossible, or even that hard. And I would definitely use it.
Anybody game for coding this one up? Sounds like a GOOD, SHORT TERM, free software project, same license as GTK (of course).
As a relaxing excerise for the reader who is usually frustrated with the death grip MS has on the industry... Think some happy (though unrealistic) thoughts. It calms tensions...
Wouldn't it be nice if:
after MS is broken into pieces, the "apps" side of the corporation decided to repent. I think I would really enjoy using Microsoft Office if it was made open source (even if not GPLed). Besides, I think that they don't have too much to loose by this. People will still use their product (those that already do) and it would help mend deep wounds inflicted on the free software community.
It should be something for little MSs to think about if and whne Ma-MS is broken.