Actually today if I was doing a SBS server I would probably use some sort of virtualization, and use snapshots for a level of backups and to allow the rollback of bad patches.
Yep that's what I just finished doing. We started testing last Friday but due to some changes going into production last weekend we weren't able to test all systems. We blocked all known paybload sites for the in-the-wild exploit at the firewall and set it to email us if those rules triggered. This morning we got a hit. We went into emergency response mode. The patch testing went to the top of the pile for all application groups. We located the affected workstation and pulled it from the network, confirmed it hadn't used the exploit on anything else on the network through analyzing server vlan sniffer traces (we mirror the entire vlan to a network general sniffer box with a couple TB array for about a week of storage). I just finished patching all of our servers except for the financials because they are running month end processing tonight but they will be done tomorrow night and all of the monitoring and blocking is still in place. Luckily all the systems like the proxy server, firewall and sniffer give us enough protection and visibility that we didn't have any big worries on this one.
When will people get it through their heads that subprime mortgages have little to do with the current problems? The bigger problem was the repealing of post-depression legislation limiting the leverage of banks and the insane amount of dollars injected into the economy to get us out of the LAST recession. That money had to find somewhere to go and it just happened to find realestate. That in and of itself wouldn't have been enough to lead to a 40+% decline in the markets, that took default credit swaps, mortgage derivatives and other accounting tricks to amplify the damage many fold. Even the total holdings of Freddie and Fannie are puny compared to the US GDP let alone the world economy. No, it took the shadow financial system and its crazy accounting tricks to multiply the problem to proportions that overshadow the real economy. The question now is can anyone safely unravel that crap without causing a decade long depression? No one knows but I can tell you it certainly ain't Paulson and his give me $700B and I will make it all better dictate that will do it.
Actually they already have, it's called EFI and like OpenFirmware it allows you to use a single keycode to get into a shell environment where you can program device settings in a standardized way. I'm not sure if the spec calls for the elimination of other keycode entry methods but I would think that eventually the others would die out since they are an unnecessaru expense to code.
As in the level of service we guarantee is none. If you are relying on this for your business you better have paid up your business insurance premiums and made damn sure that an outage is covered because you ain't even getting a refund for services not rendered by Amazon.
Try the Great Lakes, we've had shorts on Christmas and snow in LATE June within my lifetime and I'm only 30! Not only that but we are at the confluence of three weather patterns and get the fun of the lakes affecting all 3. The common quip around here is "If you don't like the weather wait around a few minutes, it will change".
Seriously, on a site called sciencedaily they printed an article which said "100 times colder than intergalactic space" and the quip about the magnet? I mean I can see doing that for the Tuesday NY Times Science section where it's businessmen and lay geeks reading it, but for a site like that I would expect units like K and T to be used. Besides AFAIK the most powerful electromagnets on earth are those used in the LHC.
Yeah I was a bit wigged out by something I saw the other day, while regular gas had fallen to $2.30/gal diesel was still almost $4! Not sure why the truckers are getting shafted but at some point they need to collectively decide that their increased costs are going to get passed on because their customers have little alternative for moving their goods. That would be bad in the short term for my dad because transportation is already one of his largest costs after raw materials but he'd adjust and pass along the cost to his customers.
If your margins are so thin you can't afford to absorb one lost shipment you are already out of business and just don't know it. My dad has customers file bankruptcy on a semi-regular basis and it never puts him out of business, it's just factored into his margin. Unless you have a single customer who accounts for more than 50% of your sales (like Walmart for many companies) the loss of payment from a single source shouldn't put you out of business.
It's like internet porn, some shops will have enough quality to attract direct paying customers who have more time than money, the rest will give their product away for free and pay for it with advertising. The barriers to entry are low enough that every joe schmo out of work can try their hand at the game so a bad economy will actually lead to MORE people throwing "free" content out there with the hope of hitting it big.
Sub 30C? During the Summer my ambient room temp is barely sub 30C (~28-28.5C) how does even rapid airflow get the drive that close to ambient when it's using a couple Watts in a fairly small volume? I try to keep my drives around 35-36C which is well within their specified range.
Actually, yes. Or at least that's one of the way my SAN vendor is dealing with increasing drive size. They are the first vendor to enable ANSI T10-DIF end to end checksumming which includes additional bits per block (AFAIR it's the same as mainframe drives have been using all along). They have also made the recoverable element the surface rather than the drive so recovery times are several times faster. Check them out.
That's why the quotes, but I have things like Oracle that will copy everything the do at initial load into swap, so I need at least 1x ram, and 1.5x ram being 80% used isn't out of the ordinary. Since the result of running out of swap space is the server falling over I go conservative and go with 2x ram.
RAID+snapshot's IS an effective first level of backup. For my dad's business I supplement that with local and remote copies to their workstations and then the local workstation also backs up to an offsite provider, but I've never needed anything more than the snapshots so far in 5 years of running this setup. Even in the S&P 500 company I admin there's very few non-legal requests to go back farther than my snapshots cover (just less than 3 months).
Yes, even 110xFC disks sucks for performance of swap, the problem is swap is orders of magnitude slower than ram yet it's made to stand in for ram at times. I wish there was an OS where the VM system was tuned to not need swap, but I've seen both Windows and Linux systems where enabling swap significantly increased performance even though the boxes shouldn't need it (IE a box with 64GB of ram and using ~17GB of it).
I have servers with swap files about 10% of that size and I can expect my next DB server will be 20% of that. If your server has 128GB of ram then a 256GB swap file is 'normal'. Of course I would never use a single SATA drive for that, it goes on the faster part of the SAN =)
Come on, any SATA drive can play an HD movie, even a BluRay rip comes out at what 45Mbit/s max, that's a punny 4.5MB/s, something IDE drives could do almost two decades ago.
Another big market for big drives is archival systems and disk to disk to tape systems. I don't always need 15K FC performance, but if it was significantly slower than comparable 1TB drives I might just go with more 1TB drives to make up the capacity difference.
Oh I agree completely, a converged network is desirable and some would say inevitable. What I was at Storage Networking World this spring there were multiple presenters who basically said the same thing, FC as a physical transport is dying. As an end user I think this is a GOOD thing, give me a single duel fabric to manage for data, voice, storage, video, etc and I'll be very happy because it means fewer things to break, fewer technologies to learn and fewer products to master. I guess if I was a SAN specialist I might not like it so much, but I'm a generalist so convergent technologies just mean more hats I can wear =)
No they want it because a LOT of money gets sunk into the development of new standards and PHY chips for other protocols that could be more easily implemented as a personality on top of a reliable ethernet transport. A good example is Fiberchannel, most of the physical equipment and low level signal encoding is the same between say 8Gbit FC and 10Gbit ethernet so if you had a reliable ethernet layer with reliable switches you would only need one type of switch, one type of wire, and one type of admin. This represents a real cost savings to both the industry and the industries largest consumers, the enterprise.
Huh, in Ethernet which is CSMA/CD you listen to the wire before starting to transmit, this doesn't avoid all possible garbled packets but it does avoid most if things are working to spec. Also because VTT goes higher than normal transmit levels during a collision there IS detection. The reason that ethernet won over tokenring is that IBM charged a hefty fee per port for tokenring. There was also real world reliability problems with tokenring as early designs used a physical ring or string layout instead of the star topology of ethernet (later tokenring switches did allow for physical star topology but this was well past the point where ethernet had won the mass market).
Since 10Gbit Ethernet has the collision domain defined as the two endpoints there IS no longer a collision domain on the wire, just a virtual one in an oversubscribed switch. This isn't about guaranteeing transmission over the internet, it's about guaranteeing reliability in a LAN/MAN/WAN Ethernet network. The idea is you will have one set of wires, one physical protocol with several personalities sitting on top. The biggies are TCP/IP and FCOE but there are other things like remote DMA that can greatly benefit from a reliable network layer transport.
Slashdot Mirror
Actually today if I was doing a SBS server I would probably use some sort of virtualization, and use snapshots for a level of backups and to allow the rollback of bad patches.
Yep that's what I just finished doing. We started testing last Friday but due to some changes going into production last weekend we weren't able to test all systems. We blocked all known paybload sites for the in-the-wild exploit at the firewall and set it to email us if those rules triggered. This morning we got a hit. We went into emergency response mode. The patch testing went to the top of the pile for all application groups. We located the affected workstation and pulled it from the network, confirmed it hadn't used the exploit on anything else on the network through analyzing server vlan sniffer traces (we mirror the entire vlan to a network general sniffer box with a couple TB array for about a week of storage). I just finished patching all of our servers except for the financials because they are running month end processing tonight but they will be done tomorrow night and all of the monitoring and blocking is still in place. Luckily all the systems like the proxy server, firewall and sniffer give us enough protection and visibility that we didn't have any big worries on this one.
When will people get it through their heads that subprime mortgages have little to do with the current problems? The bigger problem was the repealing of post-depression legislation limiting the leverage of banks and the insane amount of dollars injected into the economy to get us out of the LAST recession. That money had to find somewhere to go and it just happened to find realestate. That in and of itself wouldn't have been enough to lead to a 40+% decline in the markets, that took default credit swaps, mortgage derivatives and other accounting tricks to amplify the damage many fold. Even the total holdings of Freddie and Fannie are puny compared to the US GDP let alone the world economy. No, it took the shadow financial system and its crazy accounting tricks to multiply the problem to proportions that overshadow the real economy. The question now is can anyone safely unravel that crap without causing a decade long depression? No one knows but I can tell you it certainly ain't Paulson and his give me $700B and I will make it all better dictate that will do it.
Actually they already have, it's called EFI and like OpenFirmware it allows you to use a single keycode to get into a shell environment where you can program device settings in a standardized way. I'm not sure if the spec calls for the elimination of other keycode entry methods but I would think that eventually the others would die out since they are an unnecessaru expense to code.
As in the level of service we guarantee is none. If you are relying on this for your business you better have paid up your business insurance premiums and made damn sure that an outage is covered because you ain't even getting a refund for services not rendered by Amazon.
Try the Great Lakes, we've had shorts on Christmas and snow in LATE June within my lifetime and I'm only 30! Not only that but we are at the confluence of three weather patterns and get the fun of the lakes affecting all 3. The common quip around here is "If you don't like the weather wait around a few minutes, it will change".
Seriously, on a site called sciencedaily they printed an article which said "100 times colder than intergalactic space" and the quip about the magnet? I mean I can see doing that for the Tuesday NY Times Science section where it's businessmen and lay geeks reading it, but for a site like that I would expect units like K and T to be used. Besides AFAIK the most powerful electromagnets on earth are those used in the LHC.
I would try something more along the lines of an industrial surplus store like HGR.
Yeah I was a bit wigged out by something I saw the other day, while regular gas had fallen to $2.30/gal diesel was still almost $4! Not sure why the truckers are getting shafted but at some point they need to collectively decide that their increased costs are going to get passed on because their customers have little alternative for moving their goods. That would be bad in the short term for my dad because transportation is already one of his largest costs after raw materials but he'd adjust and pass along the cost to his customers.
If your margins are so thin you can't afford to absorb one lost shipment you are already out of business and just don't know it. My dad has customers file bankruptcy on a semi-regular basis and it never puts him out of business, it's just factored into his margin. Unless you have a single customer who accounts for more than 50% of your sales (like Walmart for many companies) the loss of payment from a single source shouldn't put you out of business.
It's like internet porn, some shops will have enough quality to attract direct paying customers who have more time than money, the rest will give their product away for free and pay for it with advertising. The barriers to entry are low enough that every joe schmo out of work can try their hand at the game so a bad economy will actually lead to MORE people throwing "free" content out there with the hope of hitting it big.
Sub 30C? During the Summer my ambient room temp is barely sub 30C (~28-28.5C) how does even rapid airflow get the drive that close to ambient when it's using a couple Watts in a fairly small volume? I try to keep my drives around 35-36C which is well within their specified range.
Actually, yes. Or at least that's one of the way my SAN vendor is dealing with increasing drive size. They are the first vendor to enable ANSI T10-DIF end to end checksumming which includes additional bits per block (AFAIR it's the same as mainframe drives have been using all along). They have also made the recoverable element the surface rather than the drive so recovery times are several times faster. Check them out.
That's why the quotes, but I have things like Oracle that will copy everything the do at initial load into swap, so I need at least 1x ram, and 1.5x ram being 80% used isn't out of the ordinary. Since the result of running out of swap space is the server falling over I go conservative and go with 2x ram.
For a home user a first level may be all they need, in fact it's probably more than they have today....
It's called eSATA, much better than either of those for external storage and it's here today.
RAID+snapshot's IS an effective first level of backup. For my dad's business I supplement that with local and remote copies to their workstations and then the local workstation also backs up to an offsite provider, but I've never needed anything more than the snapshots so far in 5 years of running this setup. Even in the S&P 500 company I admin there's very few non-legal requests to go back farther than my snapshots cover (just less than 3 months).
Yes, even 110xFC disks sucks for performance of swap, the problem is swap is orders of magnitude slower than ram yet it's made to stand in for ram at times. I wish there was an OS where the VM system was tuned to not need swap, but I've seen both Windows and Linux systems where enabling swap significantly increased performance even though the boxes shouldn't need it (IE a box with 64GB of ram and using ~17GB of it).
I have servers with swap files about 10% of that size and I can expect my next DB server will be 20% of that. If your server has 128GB of ram then a 256GB swap file is 'normal'. Of course I would never use a single SATA drive for that, it goes on the faster part of the SAN =)
Come on, any SATA drive can play an HD movie, even a BluRay rip comes out at what 45Mbit/s max, that's a punny 4.5MB/s, something IDE drives could do almost two decades ago.
Another big market for big drives is archival systems and disk to disk to tape systems. I don't always need 15K FC performance, but if it was significantly slower than comparable 1TB drives I might just go with more 1TB drives to make up the capacity difference.
Oh I agree completely, a converged network is desirable and some would say inevitable. What I was at Storage Networking World this spring there were multiple presenters who basically said the same thing, FC as a physical transport is dying. As an end user I think this is a GOOD thing, give me a single duel fabric to manage for data, voice, storage, video, etc and I'll be very happy because it means fewer things to break, fewer technologies to learn and fewer products to master. I guess if I was a SAN specialist I might not like it so much, but I'm a generalist so convergent technologies just mean more hats I can wear =)
No they want it because a LOT of money gets sunk into the development of new standards and PHY chips for other protocols that could be more easily implemented as a personality on top of a reliable ethernet transport. A good example is Fiberchannel, most of the physical equipment and low level signal encoding is the same between say 8Gbit FC and 10Gbit ethernet so if you had a reliable ethernet layer with reliable switches you would only need one type of switch, one type of wire, and one type of admin. This represents a real cost savings to both the industry and the industries largest consumers, the enterprise.
Huh, in Ethernet which is CSMA/CD you listen to the wire before starting to transmit, this doesn't avoid all possible garbled packets but it does avoid most if things are working to spec. Also because VTT goes higher than normal transmit levels during a collision there IS detection. The reason that ethernet won over tokenring is that IBM charged a hefty fee per port for tokenring. There was also real world reliability problems with tokenring as early designs used a physical ring or string layout instead of the star topology of ethernet (later tokenring switches did allow for physical star topology but this was well past the point where ethernet had won the mass market).
Since 10Gbit Ethernet has the collision domain defined as the two endpoints there IS no longer a collision domain on the wire, just a virtual one in an oversubscribed switch. This isn't about guaranteeing transmission over the internet, it's about guaranteeing reliability in a LAN/MAN/WAN Ethernet network. The idea is you will have one set of wires, one physical protocol with several personalities sitting on top. The biggies are TCP/IP and FCOE but there are other things like remote DMA that can greatly benefit from a reliable network layer transport.