I'd love this, as I could keep a week backspool of CNN, CNN/HN, Fox News, MSNBC, BBC, CSPAN, and a tuner left over for Fox/Cartoon Network/whatever else I feel like watching. With the amount of stories that get quashed after one broadcast, it would be handy to be able to keep things around for a while.
Find a good way to virtualize ramdisks in commodity servers, then bind it together into addressable luns, and then you can market a replacement. Better yet, find an off the shelf server that can hold more than 3TB of ram, and just mount it as a ramdisk, and use the remaining 500G as memory for the system.
The TMS solution is a drop in replacement for SAN based disk.
I was seriously looking at this product as a solution for a really bad database performance issue for a major financial institution. Instead of redesigning the whole database (small, around 2-3TB), we were thinking about putting a couple of these in to replace the slow EMC 8830/DMX infrastructure. Unfortunately, the PCI busses of the DB server weren't fast enough to keep up with it (Unisys ES7000 420). When you look at the TCO, the man years required to redesign the DB and applications were much more expensive than dropping in another few million in hardware. The only downside that I saw to the terraramsan solution was that it eats power and generates way too much heat. This thing would be great for horizontally scaled databses, as long as your physical plant could support it. Brute force always wins over recoding. That is, until we got offshored by Indians billing at $30/hour.
If it runs on Pocketpc, why can't they just make an app that will run on all softphones? Its trivial to intercept the mic and speaker calls and route them through an encryption/decryption routine. Hell, you could use bluetooth for it and just make a headset profile that handles the encryption/decryption. Then you could use your PDA as a handset for your bluetooth enabled phone, with encryption over the public network segment. The PAN would be encrypted as well.
What's to keep people from finding out the data format of the outgoing packets and seeding their database with a couple of terabytes of bogus statistics? Say every couple seconds a few thousand users send out packets stating that everyone is listening to the best of the 700 club or something? This seems relatively east to make useless, as there can't be too much authentication going on here...
So can you get an account from out of the country? Say I got a one2free.com.hk account, but live in the US. Then can I pay the monthly fee, never use the voice services, and have unlimited GPRS service here, for the price of the monthly rate?
Mmmmm. loopholes.
(Don't chastise me for not reading one2free's site yet. I just thought of this...)
I always wondered what that error was. I always assumed it was some star trek style geek message.. Close Enough. DomainOS was still badass for its time, even if the hardware was a little goofy.
Interesting thing I noticed from one of the subpages on this.. The baby vio comes with a MD drive. Now, as I've been too cheap to go out and replace my MD walkman with an mp3 system, this thing could have the potential to be seriously useful. Question is, can you also write data to the MDs? Last I had heard, there was only one MD data drive from a long time back that only handled about 100M. If they upped the capacity, this would be incredibly useful to me, since I have tons of MDs around, and I like the idea of security through obscurity (who else has MD drives available?). That, and maybe (obligatory matrix reference), I can start getting people to buy the 0day off me before I go out and follow the rabbit...
Anyone else notice that board had what looked to be a PC/104 bus? That just adds even more to the potential. Put it in a new case, add a PC/104 board with another PCMCIA controller, and you've got 3x ethernet + 802.11b . Hells yah. Firewall potential gets nice. External, Internal, DMZ, and 802.11b DMZ.
I have no problems with 2.4.17 on an UltraAXi based machine running heavily modified Redhat 6.2. Its under moderate load as an NFS server, and runs some other menial tasks. My only complaint is that pcmcia-cs doesn't work under sparc, which forces me to use my w2k box as my 802.11 gateway. Damn lack of floppy drive slots for the cardreader. 2.4.17 hasn't given me any problems under Redhat 7.2 in my jukebox machine either.
Linux sparcstor 2.4.17 #3 SMP Thu Jan 10 06:12:38 EST 2002 sparc64 unknown
Anyone recall that right after that F117 was shot down, the US "accidentally" bombed the Chinese embassy there? Was the majority of the wreckage of the F117 ever reclaimed by the US? Reason would dictate that perhaps there were interests within Kosovo that were interesting in developing stealth technology on someone else's nickel... Did the US possibly "accidentally and regretably" keep US technology out of unfriendly hands who would have been able to reproduce it?
Anyone notice this thread is being over moderated? There are/were a ton of posts about alleged child molestation, etc. Having seen this man in action, I hate to say it, but these people do have a valid point. Is slashdot afraid of a Libel Lawsuit against them? There's probably hundreds of people that could testify with first knowledge on this. This man is not eccentric. He's got a problem which people overlook because at one point, he did something that people thought was cool.
Having met John on several occasions, and had lengthy security discussions with him, it seems that this technical savvy expired after writing EasyWriter. This guy did do a lot for the "scene" back in the day, but he's simply outclassed in today's security business. The company is riding on a name. Would you hire Eric Corely to secure your network? Of course not. He's got a name known to the media, but that's about it. Unless the company has some technical people, its merely riding on a name that's 30 years out of date...
$.02
There's also other simple reasons besides "Linux is the in thing".
OpenBSD would probably have been a good place for them to start, except for some serious factors against it. OpenBSD isn't scalable. At all. It's great for small corporate networks, or home firewalls and such, but wiht no plans for SMP in the future, it can't compete as a server environment. OpenBSD is great for IDS sensors, and specific appliance type hardened boxes, but it's not well rounded enough to put into a big multi user production environment. Have you ever heard of any major e-commerce site using OBSD as their primary server software? Then there's the whole problem with Theo not playing well with others. And being Canadian. Because Theo wholly manages the project himself, it would cause issues. I believe the NSA folks are looking to put together something that can go into general release, possibly as an option on any distribution. OBSD can't do this, because a) the NSA would have to pay Theo to audit their code [see how OBSD architecture ports end up getting made.. its interesting], or he wouldn't let them integrate it into his source tree, and B) there's all kinds of weird issues with the project maintainer not being a US national. I'm not knocking OpenBSD. I'm a big supporter. I run it on a lot of appliance type boxes, rnuning security centric tasks. However, don't believe for a second its secure. It requires the same amount of tweaking as any other operating system to get it into shape. I've had OBSD machines get owned before, where there were serious user errors in judgement. Just because there aren't any *remote* exploits, doesn't mean your users aren't going to get drunk and give away their account. Trusted OSes are a little more forgiving when this kind of thing happens.
My $.02 . Take it for what its worth. Or ask for change back.
The only way this is going to work is to have a large common database on the backend. How long is it going to take the banner ad/porn people to start seeding it with keywords?
Sure Solaris does load balancing across disks. Too many people have been playing around with "Free Solaris" to get a good feel for what real Solaris is all about. Solaris comes with DiskSuite, which lets you stripe data, and if you get into any of the SparcStorage Arrays, you get a RTU on Veritas. Veritas lets you do pretty much anything you want with the drives RAIDwise. My SSA112's and A5000's work mighty fine with Oracle. You cant go wrong with a fistfull of RAID 0+1 FCAL drives at 10k rpm. As far as scalability, again, don't look at "Free Solaris", its aimed towards the workstation market. Look serverside and you'll see that 16CPU machines are quite common, and I doubt highly that many people are using one SCSI bus to handle any type of workload. My machines tend to have 6 4-6 SCSI busses, as well as older Fiber Channel and FCAL. I don't think anything but static data and the root drive is sitting on SCSI at all. Everything else on SCSI is longterm storage devices, DLT's, library robotics, etc.
Slashdot Mirror
I'd love this, as I could keep a week backspool of CNN, CNN/HN, Fox News, MSNBC, BBC, CSPAN, and a tuner left over for Fox/Cartoon Network/whatever else I feel like watching. With the amount of stories that get quashed after one broadcast, it would be handy to be able to keep things around for a while.
Find a good way to virtualize ramdisks in commodity servers, then bind it together into addressable luns, and then you can market a replacement. Better yet, find an off the shelf server that can hold more than 3TB of ram, and just mount it as a ramdisk, and use the remaining 500G as memory for the system.
The TMS solution is a drop in replacement for SAN based disk.
I was seriously looking at this product as a solution for a really bad database performance issue for a major financial institution. Instead of redesigning the whole database (small, around 2-3TB), we were thinking about putting a couple of these in to replace the slow EMC 8830/DMX infrastructure. Unfortunately, the PCI busses of the DB server weren't fast enough to keep up with it (Unisys ES7000 420). When you look at the TCO, the man years required to redesign the DB and applications were much more expensive than dropping in another few million in hardware. The only downside that I saw to the terraramsan solution was that it eats power and generates way too much heat. This thing would be great for horizontally scaled databses, as long as your physical plant could support it. Brute force always wins over recoding. That is, until we got offshored by Indians billing at $30/hour.
If it runs on Pocketpc, why can't they just make an app that will run on all softphones? Its trivial to intercept the mic and speaker calls and route them through an encryption/decryption routine. Hell, you could use bluetooth for it and just make a headset profile that handles the encryption/decryption. Then you could use your PDA as a handset for your bluetooth enabled phone, with encryption over the public network segment. The PAN would be encrypted as well.
What's to keep people from finding out the data format of the outgoing packets and seeding their database with a couple of terabytes of bogus statistics? Say every couple seconds a few thousand users send out packets stating that everyone is listening to the best of the 700 club or something? This seems relatively east to make useless, as there can't be too much authentication going on here...
So can you get an account from out of the country? Say I got a one2free.com.hk account, but live in the US. Then can I pay the monthly fee, never use the voice services, and have unlimited GPRS service here, for the price of the monthly rate?
Mmmmm. loopholes.
(Don't chastise me for not reading one2free's site yet. I just thought of this...)
I always wondered what that error was. I always assumed it was some star trek style geek message.. Close Enough. DomainOS was still badass for its time, even if the hardware was a little goofy.
Interesting thing I noticed from one of the subpages on this.. The baby vio comes with a MD drive. Now, as I've been too cheap to go out and replace my MD walkman with an mp3 system, this thing could have the potential to be seriously useful. Question is, can you also write data to the MDs? Last I had heard, there was only one MD data drive from a long time back that only handled about 100M. If they upped the capacity, this would be incredibly useful to me, since I have tons of MDs around, and I like the idea of security through obscurity (who else has MD drives available?). That, and maybe (obligatory matrix reference), I can start getting people to buy the 0day off me before I go out and follow the rabbit...
Anyone else notice that board had what looked to be a PC/104 bus? That just adds even more to the potential. Put it in a new case, add a PC/104 board with another PCMCIA controller, and you've got 3x ethernet + 802.11b . Hells yah. Firewall potential gets nice. External, Internal, DMZ, and 802.11b DMZ.
Bablefish works fine translating Japanese for me. Comes out reading like late 70s calculator instructions, but it *is* some semblance of english.
I have no problems with 2.4.17 on an UltraAXi based machine running heavily modified Redhat 6.2. Its under moderate load as an NFS server, and runs some other menial tasks. My only complaint is that pcmcia-cs doesn't work under sparc, which forces me to use my w2k box as my 802.11 gateway. Damn lack of floppy drive slots for the cardreader. 2.4.17 hasn't given me any problems under Redhat 7.2 in my jukebox machine either.
Linux sparcstor 2.4.17 #3 SMP Thu Jan 10 06:12:38 EST 2002 sparc64 unknown
Too bad chinese fighter pilots have a navigation tolerance of +-3 meters or so...
Anyone recall that right after that F117 was shot down, the US "accidentally" bombed the Chinese embassy there? Was the majority of the wreckage of the F117 ever reclaimed by the US? Reason would dictate that perhaps there were interests within Kosovo that were interesting in developing stealth technology on someone else's nickel... Did the US possibly "accidentally and regretably" keep US technology out of unfriendly hands who would have been able to reproduce it?
Anyone notice this thread is being over moderated? There are/were a ton of posts about alleged child molestation, etc. Having seen this man in action, I hate to say it, but these people do have a valid point. Is slashdot afraid of a Libel Lawsuit against them? There's probably hundreds of people that could testify with first knowledge on this. This man is not eccentric. He's got a problem which people overlook because at one point, he did something that people thought was cool.
Having met John on several occasions, and had lengthy security discussions with him, it seems that this technical savvy expired after writing EasyWriter. This guy did do a lot for the "scene" back in the day, but he's simply outclassed in today's security business. The company is riding on a name. Would you hire Eric Corely to secure your network? Of course not. He's got a name known to the media, but that's about it. Unless the company has some technical people, its merely riding on a name that's 30 years out of date... $.02
There's also other simple reasons besides "Linux is the in thing". OpenBSD would probably have been a good place for them to start, except for some serious factors against it. OpenBSD isn't scalable. At all. It's great for small corporate networks, or home firewalls and such, but wiht no plans for SMP in the future, it can't compete as a server environment. OpenBSD is great for IDS sensors, and specific appliance type hardened boxes, but it's not well rounded enough to put into a big multi user production environment. Have you ever heard of any major e-commerce site using OBSD as their primary server software? Then there's the whole problem with Theo not playing well with others. And being Canadian. Because Theo wholly manages the project himself, it would cause issues. I believe the NSA folks are looking to put together something that can go into general release, possibly as an option on any distribution. OBSD can't do this, because a) the NSA would have to pay Theo to audit their code [see how OBSD architecture ports end up getting made.. its interesting], or he wouldn't let them integrate it into his source tree, and B) there's all kinds of weird issues with the project maintainer not being a US national. I'm not knocking OpenBSD. I'm a big supporter. I run it on a lot of appliance type boxes, rnuning security centric tasks. However, don't believe for a second its secure. It requires the same amount of tweaking as any other operating system to get it into shape. I've had OBSD machines get owned before, where there were serious user errors in judgement. Just because there aren't any *remote* exploits, doesn't mean your users aren't going to get drunk and give away their account. Trusted OSes are a little more forgiving when this kind of thing happens. My $.02 . Take it for what its worth. Or ask for change back.
The only way this is going to work is to have a large common database on the backend. How long is it going to take the banner ad/porn people to start seeding it with keywords?
Actually, you can get an Ultra5 for under 3 grand... check out www.recurrent.com .
Sure Solaris does load balancing across disks. Too many people have been playing around with "Free Solaris" to get a good feel for what real Solaris is all about. Solaris comes with DiskSuite, which lets you stripe data, and if you get into any of the SparcStorage Arrays, you get a RTU on Veritas. Veritas lets you do pretty much anything you want with the drives RAIDwise. My SSA112's and A5000's work mighty fine with Oracle. You cant go wrong with a fistfull of RAID 0+1 FCAL drives at 10k rpm. As far as scalability, again, don't look at "Free Solaris", its aimed towards the workstation market. Look serverside and you'll see that 16CPU machines are quite common, and I doubt highly that many people are using one SCSI bus to handle any type of workload. My machines tend to have 6 4-6 SCSI busses, as well as older Fiber Channel and FCAL. I don't think anything but static data and the root drive is sitting on SCSI at all. Everything else on SCSI is longterm storage devices, DLT's, library robotics, etc.