Stagnation? You need to look again. Sun has a LOT of new products that have shown up recently. Dual core AMD processor 1U servers starting at $795 isn't something new?
If things were happening in a New York State Court and NOT Utah, I think you'd have already seen IBM doing a Victory dance. However, This case is in Federal Court so the chances of "buying justice" are less (judges have life appointments). That said, Utah is SCO's home turf, and I think in the early days they used it to their advantage. Another reason for the time stretching out is that SCO vs IBM is not the only case Judge Kimball is handling. I'm still not sure what Hatch has to do with all this other than his Son is part of the litigation team for SCO. It would be highly unusual to see a US Senator trying to intercede in the case (at least not publically). But in Utah people of the Mormon faith do tend to stick together, so it could happen.
Item 0: Requirements reviews by peers and independants. If you don't have good requirements you obviously don't know things well enough to be building them. Sure you can catch some requirements issues in 1 and 2 but the longer you wait the costlier it is to fix.
A MSCS is NOT a Software Engineering Degree, so why WOULD you take courses in SE?I'd say that CS and SE are two different professions. There are places to get a MS SE (Texas Tech comes to mind) if you are interested.
IBM has asked for Summary Judgement, the courts have not allowed it (yet..maybe when Discovery is over in early 2006). SCO is really being allowed massive amounts of rope by the Courts. That is the reason things are taking so long. Which might be good if there is an appeal. Based on what I have read on Groklaw from legal experts, the length of time this case is taking is really not that long compared to others of a similar nature.
Yes, IBM could have bought them for a fraction of the cost of litigation, but I think IBM is 1)standing on prinicple they didn't do anything 2) won't be "blackmailed", if they give in to SCO God only know how many others will be coming after IBM 3) winning in Court would clear Linux completely from any shadow of being derived from UNIX.
Most Cash donations to non-profit hospitals go to cover medical/admin costs so the children can get the care for litte or no cost. Of course that is the goal of these Hospitals and they have to take care of that obligation first. They do a very good job making the money stretch. Places like you mentioned as well as others like St. Judes do an awesome job treating sick kids on limited funds.
After covering treatment/admin/Overhead/Facility costs very little is leftover for "extras" like books, toys, or games to keep the kids minds occupied during treatment. I know a positive mental attitude is very important to the healing process so if this helps with that I'm just as much for it as just donating cash.
I wonder if these guys could go to Nintendo, Amazon, etc. and get them to give the games up for FREE or at cost to make the donations go further. The shipping fees are not that big a deal for Amazon to kick in if they are making the usual retail markup on the game. In fact, it may be UPS that picks up part of that tab and not just Amazon. Every little bit helps but those who can do more should do more.
I've seen it both ways, but the BIG difference is that with the knowledge you get in college you can learn new things quicker, are broader in your understanding and are productive in different contexts and likely in different programming langauges and domains . Just writing code is NOT all there is to being a "good programmer",it's understanding why you do things and what impact they will have. Just learning Java from a book isn't going to give you that. However, once you have 5-7 yrs practical experience the degree isn't going to help you out unless you need it to advance due to company policy or you eventually want an advanced degree.
Everyone knows LE is an assh()le, but CFOs leaving right and left probably indicates a deeper issue. My bet is some funny accounting on all these mergers/buyouts Oracle has been doing. I wonder if they are trying to pull some fast moves with how they handle the acquisition expenses. As I recall from my MBA the excess of cost over book value has to be handled as Goodwill and amortized over 10? yrs. Oracle may be trying to avoid that, because it will show they paid WAY too much for Peoplesoft.
P.S. Enjoyed your blog. Tried to post a comment but I have to login it seems. I'm also a DFW techno person. The Halloween post was funny but the poor kids!!
Bad comparison, DEC has been gone from the business for quite a while. DEC went under and Compaq bought the leftovers. SUN is holding on pretty good, they lost a penny a share but are growing slightly. If you used HP instead of DEC you would have a better illustration.
In either case, you could always buy Insurance for the damage, but you better have a backup data center for a while. Pipes rarely burst unless something was done wrong in the plumbing but can you take the chance?
Ethylene Gylcol (basic anti-freeze) does conduct electricity (didnt know that) but does Propylene Gylcol? That is actually a better anti-freeze but it is very toxic to animals and is rarely used.
There must be cheaper options than flourinert that are better than water, say the new environmentally safe "Puron" used in A/C units. I may take some time tomorrow and look into that.
I wouldn't use water but something that if a leak occurs nothing bad happens. Anti-freeze is pretty much inert and transfers heat well. IIRC, some of the Cray supercomputers were water cooled. So I guess that technology belongs to SGI (for now) since they bought Cray.
Well, for one thing the temp & humidity of outside air varies a lot, which can make more work for your A/C systems. The air within the datacenter only raises a pretty constant amount (heat from servers) and the humidity is constantly low. Even with turns your A/C systems work less. Convection cooling while very energy efficient is really slow unless you are somewhere where you can use the wind to pull a vacuum and suck it out the vents.
Wow, flamebait mod for telling the truth!!! Yes, these were around on mainframes in the 1970/80s and as well as being in DEC VMS as well. As usual, M$ re-invents (or copies) other's work and claims it is new.
Idiot. The $70M came mostly from funds he had leftover from running for Governor twice in Texas. The old laws said what you don't spend you get to keep and can use in just about any way you want. GWB was tagged as a possible Presidental candidate about halfway into his second term as Governor. Far as I know you don't actually have to be a "official" candidate to raise money, you can just say you are forming an "exploratory committee" and start raising money for things like people, offices, name identification surveys, etc. It's done all the time by BOTH parties
I wish I had the time to go craft up some scenarios and really sit down and look at the results...
I agree, I did some benchmarking similar to this about 5-7 years ago for some avionics architecture but was not funded to get into this level of detail. I was looking for best CPU, best network, best OS combinations. Somedays I miss the embedded world,as now I work in Enterprise IT where such stuff is WAY below the level of concern.
If you are doing a lot of FFTs on an x86 processor you probably have either the wrong processor (DSPs are much better at FFTs) or the wrong algorithm. Estimating things (or pre-calculating the common values and interpolating the rest) and using table lookups is much faster even with using the extra registers. Memory alignement on structures USED to be an issue, but not anymore. I did some research about 4-5 yrs ago that showed on a PowerPC processor alignment of structures made no difference, the tiny bit of lost time was greatly offset in other areas. Sure, this tweak can squeeze out the last 1/2 of 1% performance and that is sometimes needed but for 99.9% of the applications it doesnt matter.
Umm, how is tweaking the innermost loops going to result in more cache hits? With the size of caches today you can bet that loop is already in cache after 1 iteration. Instrumenting code is old stuff, a good High level language debugger can tell you so much more, and if you need to know more than that learn to read circuit schematics and how to operate an oscilliscope. Data size might be an issue, but some CPUs are smart enough to see you are "walking" thru memory and do pre-fetches to help you out. Most CPUs when they DO have to reload the cache are not grabbing a few instructions or bytes of data, they go grab 4K or 8K blocks.
No, you CAN pare down the variables into discrete sets so that you cover 95% of the threats for a reasonable cost and time. Occam's Razor and Paretos Rule apply in security just as in other parts of IT. You cannot make a system 100% secure for all situations unless you have infinte time or money so you have to compromise wisely. People and weak security processes are more likely to cause loss than purely a technological attack. After all, 90% of the hacks come from INSIDE the company.
Yep, good analogy. The Law of Diminishing Marginal Returns kicks in Immediately. There are some interesting "solutions" being tried (such as the NUMA Architecture from SGI) and other shared memory approaches to reduce the memory hit.
So switch to Sun Blades or the new Xseries boxes, and X4200 would work for you. Those run Solaris just fine and boot from SAN. Of course since you just bought IBM Blade Centers I guess that's not an option.
If something I get for Free is a Pile of Crap, I'm going to say it's a Pile of Crap, same as if I paid for it. My integrity is NOT for Sale.
Buying back a used car with high-mileage and doing a tear down to actually evaluate the wear and tear sounds like something that should be done more often!!
Yea, and get no support. You can also get OpenSolaris for Free. Support starts @ $120/yr and up. IBM tells you up front they won't support Solaris, you gotta go to Sun for that.
Slashdot Mirror
Nahh..play the squeals and grunts of a group of Pigs...that should really make them mad and deaf at the same time ;)
Stagnation? You need to look again. Sun has a LOT of new products that have shown up recently. Dual core AMD processor 1U servers starting at $795 isn't something new?
If things were happening in a New York State Court and NOT Utah, I think you'd have already seen IBM doing a Victory dance. However, This case is in Federal Court so the chances of "buying justice" are less (judges have life appointments). That said, Utah is SCO's home turf, and I think in the early days they used it to their advantage. Another reason for the time stretching out is that SCO vs IBM is not the only case Judge Kimball is handling. I'm still not sure what Hatch has to do with all this other than his Son is part of the litigation team for SCO. It would be highly unusual to see a US Senator trying to intercede in the case (at least not publically). But in Utah people of the Mormon faith do tend to stick together, so it could happen.
You mised one..
Item 0: Requirements reviews by peers and independants. If you don't have good requirements you obviously don't know things well enough to be building them. Sure you can catch some requirements issues in 1 and 2 but the longer you wait the costlier it is to fix.
A MSCS is NOT a Software Engineering Degree, so why WOULD you take courses in SE?I'd say that CS and SE are two different professions. There are places to get a MS SE (Texas Tech comes to mind) if you are interested.
IBM has asked for Summary Judgement, the courts have not allowed it (yet..maybe when Discovery is over in early 2006). SCO is really being allowed massive amounts of rope by the Courts. That is the reason things are taking so long. Which might be good if there is an appeal. Based on what I have read on Groklaw from legal experts, the length of time this case is taking is really not that long compared to others of a similar nature.
Yes, IBM could have bought them for a fraction of the cost of litigation, but I think IBM is 1)standing on prinicple they didn't do anything 2) won't be "blackmailed", if they give in to SCO God only know how many others will be coming after IBM 3) winning in Court would clear Linux completely from any shadow of being derived from UNIX.
Valid points, not a troll.
Most Cash donations to non-profit hospitals go to cover medical/admin costs so the children can get the care for litte or no cost. Of course that is the goal of these Hospitals and they have to take care of that obligation first. They do a very good job making the money stretch. Places like you mentioned as well as others like St. Judes do an awesome job treating sick kids on limited funds.
After covering treatment/admin/Overhead/Facility costs very little is leftover for "extras" like books, toys, or games to keep the kids minds occupied during treatment. I know a positive mental attitude is very important to the healing process so if this helps with that I'm just as much for it as just donating cash.
I wonder if these guys could go to Nintendo, Amazon, etc. and get them to give the games up for FREE or at cost to make the donations go further. The shipping fees are not that big a deal for Amazon to kick in if they are making the usual retail markup on the game. In fact, it may be UPS that picks up part of that tab and not just Amazon. Every little bit helps but those who can do more should do more.
I've seen it both ways, but the BIG difference is that with the knowledge you get in college you can learn new things quicker, are broader in your understanding and are productive in different contexts and likely in different programming langauges and domains . Just writing code is NOT all there is to being a "good programmer",it's understanding why you do things and what impact they will have. Just learning Java from a book isn't going to give you that. However, once you have 5-7 yrs practical experience the degree isn't going to help you out unless you need it to advance due to company policy or you eventually want an advanced degree.
Everyone knows LE is an assh()le, but CFOs leaving right and left probably indicates a deeper issue. My bet is some funny accounting on all these mergers/buyouts Oracle has been doing. I wonder if they are trying to pull some fast moves with how they handle the acquisition expenses. As I recall from my MBA the excess of cost over book value has to be handled as Goodwill and amortized over 10? yrs. Oracle may be trying to avoid that, because it will show they paid WAY too much for Peoplesoft.
P.S. Enjoyed your blog. Tried to post a comment but I have to login it seems. I'm also a DFW techno person. The Halloween post was funny but the poor kids!!
Bad comparison, DEC has been gone from the business for quite a while. DEC went under and Compaq bought the leftovers. SUN is holding on pretty good, they lost a penny a share but are growing slightly. If you used HP instead of DEC you would have a better illustration.
In either case, you could always buy Insurance for the damage, but you better have a backup data center for a while. Pipes rarely burst unless something was done wrong in the plumbing but can you take the chance?
Ethylene Gylcol (basic anti-freeze) does conduct electricity (didnt know that) but does Propylene Gylcol? That is actually a better anti-freeze but it is very toxic to animals and is rarely used.
There must be cheaper options than flourinert that are better than water, say the new environmentally safe "Puron" used in A/C units. I may take some time tomorrow and look into that.
I wouldn't use water but something that if a leak occurs nothing bad happens. Anti-freeze is pretty much inert and transfers heat well. IIRC, some of the Cray supercomputers were water cooled. So I guess that technology belongs to SGI (for now) since they bought Cray.
Electronics and water don't mix. Leaks can be deadly to servers and people.
Well, for one thing the temp & humidity of outside air varies a lot, which can make more work for your A/C systems. The air within the datacenter only raises a pretty constant amount (heat from servers) and the humidity is constantly low. Even with turns your A/C systems work less. Convection cooling while very energy efficient is really slow unless you are somewhere where you can use the wind to pull a vacuum and suck it out the vents.
Wow, flamebait mod for telling the truth!!! Yes, these were around on mainframes in the 1970/80s and as well as being in DEC VMS as well. As usual, M$ re-invents (or copies) other's work and claims it is new.
Idiot. The $70M came mostly from funds he had leftover from running for Governor twice in Texas. The old laws said what you don't spend you get to keep and can use in just about any way you want. GWB was tagged as a possible Presidental candidate about halfway into his second term as Governor. Far as I know you don't actually have to be a "official" candidate to raise money, you can just say you are forming an "exploratory committee" and start raising money for things like people, offices, name identification surveys, etc. It's done all the time by BOTH parties
ANyone got a login? BugMeNot isn't getting me in nor is Test/Test, Nature/Nature or Demo/Demo which are common "backdoors".
I wish I had the time to go craft up some scenarios and really sit down and look at the results...
I agree, I did some benchmarking similar to this about 5-7 years ago for some avionics architecture but was not funded to get into this level of detail. I was looking for best CPU, best network, best OS combinations. Somedays I miss the embedded world,as now I work in Enterprise IT where such stuff is WAY below the level of concern.
If you are doing a lot of FFTs on an x86 processor you probably have either the wrong processor (DSPs are much better at FFTs) or the wrong algorithm. Estimating things (or pre-calculating the common values and interpolating the rest) and using table lookups is much faster even with using the extra registers. Memory alignement on structures USED to be an issue, but not anymore. I did some research about 4-5 yrs ago that showed on a PowerPC processor alignment of structures made no difference, the tiny bit of lost time was greatly offset in other areas. Sure, this tweak can squeeze out the last 1/2 of 1% performance and that is sometimes needed but for 99.9% of the applications it doesnt matter.
Umm, how is tweaking the innermost loops going to result in more cache hits? With the size of caches today you can bet that loop is already in cache after 1 iteration. Instrumenting code is old stuff, a good High level language debugger can tell you so much more, and if you need to know more than that learn to read circuit schematics and how to operate an oscilliscope. Data size might be an issue, but some CPUs are smart enough to see you are "walking" thru memory and do pre-fetches to help you out. Most CPUs when they DO have to reload the cache are not grabbing a few instructions or bytes of data, they go grab 4K or 8K blocks.
Programmers working in India who are using Servers hosted in the USA to do development for a US firm will owe US Income Tax ;)
No, you CAN pare down the variables into discrete sets so that you cover 95% of the threats for a reasonable cost and time. Occam's Razor and Paretos Rule apply in security just as in other parts of IT. You cannot make a system 100% secure for all situations unless you have infinte time or money so you have to compromise wisely. People and weak security processes are more likely to cause loss than purely a technological attack. After all, 90% of the hacks come from INSIDE the company.
Nope, Sun servers work fine with EMC, IBM and others SANs. Just put the right interface cards in there to connect to the switch/server.
Yep, good analogy. The Law of Diminishing Marginal Returns kicks in Immediately. There are some interesting "solutions" being tried (such as the NUMA Architecture from SGI) and other shared memory approaches to reduce the memory hit.
So switch to Sun Blades or the new Xseries boxes, and X4200 would work for you. Those run Solaris just fine and boot from SAN. Of course since you just bought IBM Blade Centers I guess that's not an option.
If something I get for Free is a Pile of Crap, I'm going to say it's a Pile of Crap, same as if I paid for it. My integrity is NOT for Sale.
Buying back a used car with high-mileage and doing a tear down to actually evaluate the wear and tear sounds like something that should be done more often!!
Yea, and get no support. You can also get OpenSolaris for Free. Support starts @ $120/yr and up. IBM tells you up front they won't support Solaris, you gotta go to Sun for that.