"how long until every country decides that your "private" T1 connecting New York to Tokyo needs to pass through traffic sniffing tools so that both countries are sure nobody is using private corporations for terrorist activities?"
I'm trying to figure out how that's not buying it off the shelf? Yes you have to wait (and the wait is more like 4-6 months right now, I just waited 110 days to get a form4 back on a suppressor) but it's still purchased "off the shelf" ?
Not entirely incorrect but misleading. You do not have to have an FFL[SOT] to own a transferable machine gun. All you need to do is find a pre-may 1986 machine gun for sale, fill out your form 4 and associated forms, send the BATFE a check for $200, and wait 4-6 months for the paperwork to come back.
You can buy 7.62 ammunition all over the planet by the hundreds of thousands of rounds. Why would you manufacture it yourself? Ammunition supply has nothing to do with US gun laws.
If we somehow magically banned and destroyed every gun in the US tomorrow someone in South America would setup shop overnight and be cranking out firearms faster than you can imagine. Where there is a demand someone will create a supply.
An M-16 is a highly controller firearm. To own an M-16 (a fully automatic.223 caliber machine gun) you have to fill out a ton of paperwork and apply for a Tax Stamp from the BATFE (and pay $200). This includes fingerprint cards and a signed letter from your local Chief Law Enforcement Officer. Not to mention that on the low end a transferrable (meaning, can be owned by civillians) M-16 costs, BARE MINIMUM, $14,000, and up to over $20,000 for new unfired weapons.
Now, with all of that said - explain to me how we don't have "sensible gun control" ?
Sounds like a great reason to use something like SSL between memcached and the webservers. Forward the incoming SSL connections to memcached back to the memcached port over a loopback address. Now you can have authentication and encryption, if you want it, without adding any complexity to memcached. That way if someone wants to run it locally on a webserver you haven't really added any complexity.
Because in a lot of scenarios you might have a memcached server that uses public address space. You'd just use iptables to only allow ssh for remote management and then only allow your webservers to connect on your memcached port. No need for a big expensive hardware firewall. But I guess a warning wouldn't _hurt_ anything? Why not just "***WARNING - Only your webservers should be able to connect to memcache!"
Wow $500k to Citrix isn't cheap. How many users? What licensing level? Without divulging too much information we pay about $50k/year/1000-users to citrix if I remember correctly. You really need more than 100Mb fast-e for imaging software? What kind of bandwidth requirements does it have? I'm just really curious now (I work in healthcare as well).
You forgot: fiber infrastructure, administrators, money to audit/secure/acquire, power, hvac, square footage to host it. That's just off the top of my head I'm sure there's far more we're both forgetting.
You're only describing CapEx to acquire a storage system not the total cost of ownership. That's what I'm trying to illustrate.
Without getting into specifics it's impossible to say, but with your workload and the vast majority of the SAN being composed of SATA you might be able to get raw storage costs down to $3/GB across an entire large storage system. But at the petascale we're talking about dozens of controllers, shelves, UPS and a massive switching infrastructure to support it, not to mention software licensing costs, maintenance contracts and a staff to run something of that size. You mention the costs of things like 10GbE and the storage system itself, but I can't tell from context if you're including all of those things in that per-gigabyte figure.
I'd also point out that backing virtualization installations with fiber channel is not even remotely uncommon anymore because of the density that the current generation of servers can deliver. Single RMU dual socket six-core x86 machines supporting 10, 15, 20 or more VMs is standard operating procedure these days and without something like 10Gb DCE you're going to be hard pressed to deliver enough IOPS over ethernet via iSCSI. Shit 20 VMs on a single machine at 8Gb/s FC you're talking about splitting up 1GB/s of storage throughput between 20 machines would mean only 50MB/s total throughput distributed evenly between them (fictional case to prove the point). That's a fraction of a single consumer grade SATA drive.
I'd be curious to see you're actual CapEx for acquiring storage. When you consider all of these factors I wouldn't at all be surprised to see something in the $30-$40 per gigabyte range. You'd need to include things like:
- Controllers (cache, licensing, maintenance, etc)
- Assuming 10GbE iSCSI you have to allocate some portion of that cost/management to your storage system
- Redundant systems? Maybe not an issue for your installations?
- Backups (frequency, off-site rotation, etc)
- Cost per square foot (lease(?)/HVAC/power)
- Cost to design/acquire/maintain over lifespan (think storage architect, sales meetings, other soft costs - rough figures are fine obviously)
And as you pointed out, what you can buy at the petabyte scale for let's say $20/GB CapEx could easily be in the $40/GB range for someone acquiring a storage system in the 20TB range. So to make blanket statements like "$40/gig is way more than the capital costs of the storage" is definitely, if I give you absolute benefit of the doubt, misleading at best.
Wish I could, SSDs aren't supported by my SAN vendor (yet). What vendors are selling tier0 SSD cache at only $7/GB? That's a great price I would think?
Sounds like an architectural problem. Have you considered moving their operating environments closer to the storage using something like Virtual Desktops or Citrix Presentation Server? You'd only need about 30kb/s to each endpoint and you could provide massive throughput easily between app servers and storage in the datacenter.
Very rarely do I see FC SAN SLAs written in the form of raw throughput. Typically it's quoted in the form of capacity and IOPS. Do you guarantee a specific IOPS figure for that price?
When did power/hvac/floorspace become soft costs? Those are standard OpEx expenditures and as far as I know are considered "hard costs". It's like you're saying that "hard costs" are CapEx and "soft costs" are OpEx which isn't the case at all.
How many IOPS? What disk layout (disk sizes, types, any short-stroking) ? How much redundancy? Your numbers don't really seem unreasonable depending on the performance and SLA. We pay something like $30/GB for raw FC storage (including the FC infrastructure) and if I assumed I could get 4 years out of it I could pay for it at $7/GB/year.
Maybe $3/GB for raw SATA storage. There's no way you're building out a complete fiber channel storage system for $3/GB CapEx. Not even remotely possible. Raw (un-RAIDed) FC disks are about $3/GB. In RAID10 that jumps to $6/GB in disks alone. No HBAs, no FC ports, no controllers, no backups, no redundancy, no software, no caches - nothing other than bare disks.
I'm sorry where are you getting fiber channel storage for $0.10 per gigabyte? That's like the cost of bare RAID1 consumer 7200RPM sata disks ($0.05 * 2). We pay $1,000 for a 400GB 10K RPM fiber channel drive. That doesn't include: (redundant) controllers, (redundant) fiber channel HBAs, (redundant) fiber channel ports and a million other associated costs (everything from support to backups). I'm not necessarily agreeing with $30/GB/mo but $0.10 per gigabyte for the raw storage is not even remotely plausible for enterprise storage.
What type of storage is that? If I added all the CapEx (including HBAs, FC ports, software licensing, loss of _usable_ space via RAID, etc) we're probably well over $40/GB for FC SAN space.
read: I have an unlimited budget and can build perfect(tm) IT systems. I have many many friends who are government contractors and the money they waste makes me literally physically ill. If I ran IT shop like the government we'd be bankrupt before the end of the year.
Which would mean higher costs to end users which would allow another company to undercut their prices and steal their customers. Isn't capitalism great?
Slashdot Mirror
"how long until every country decides that your "private" T1 connecting New York to Tokyo needs to pass through traffic sniffing tools so that both countries are sure nobody is using private corporations for terrorist activities?"
Who cares?
I'm trying to figure out how that's not buying it off the shelf? Yes you have to wait (and the wait is more like 4-6 months right now, I just waited 110 days to get a form4 back on a suppressor) but it's still purchased "off the shelf" ?
Not entirely incorrect but misleading. You do not have to have an FFL[SOT] to own a transferable machine gun. All you need to do is find a pre-may 1986 machine gun for sale, fill out your form 4 and associated forms, send the BATFE a check for $200, and wait 4-6 months for the paperwork to come back.
Incorrect. Just requires a bit of paperwork and waiting.
You can buy 7.62 ammunition all over the planet by the hundreds of thousands of rounds. Why would you manufacture it yourself? Ammunition supply has nothing to do with US gun laws.
It doesn't really matter when you can buy the tooling to produce AK-47s pretty easily and inexpensively.
http://www.ak-builder.com/
If we somehow magically banned and destroyed every gun in the US tomorrow someone in South America would setup shop overnight and be cranking out firearms faster than you can imagine. Where there is a demand someone will create a supply.
An M-16 is a highly controller firearm. To own an M-16 (a fully automatic .223 caliber machine gun) you have to fill out a ton of paperwork and apply for a Tax Stamp from the BATFE (and pay $200). This includes fingerprint cards and a signed letter from your local Chief Law Enforcement Officer. Not to mention that on the low end a transferrable (meaning, can be owned by civillians) M-16 costs, BARE MINIMUM, $14,000, and up to over $20,000 for new unfired weapons.
Now, with all of that said - explain to me how we don't have "sensible gun control" ?
Oh and M-16 uses a magazine not a clip. Sorry, pet peeve.
HDMI to DVI or even the new laptops with WiDi
Sounds like a great reason to use something like SSL between memcached and the webservers. Forward the incoming SSL connections to memcached back to the memcached port over a loopback address. Now you can have authentication and encryption, if you want it, without adding any complexity to memcached. That way if someone wants to run it locally on a webserver you haven't really added any complexity.
Because in a lot of scenarios you might have a memcached server that uses public address space. You'd just use iptables to only allow ssh for remote management and then only allow your webservers to connect on your memcached port. No need for a big expensive hardware firewall. But I guess a warning wouldn't _hurt_ anything? Why not just "***WARNING - Only your webservers should be able to connect to memcache!"
literally the worst post in the history of slashdot.
Wow $500k to Citrix isn't cheap. How many users? What licensing level? Without divulging too much information we pay about $50k/year/1000-users to citrix if I remember correctly. You really need more than 100Mb fast-e for imaging software? What kind of bandwidth requirements does it have? I'm just really curious now (I work in healthcare as well).
You forgot: fiber infrastructure, administrators, money to audit/secure/acquire, power, hvac, square footage to host it. That's just off the top of my head I'm sure there's far more we're both forgetting.
You're only describing CapEx to acquire a storage system not the total cost of ownership. That's what I'm trying to illustrate.
Without getting into specifics it's impossible to say, but with your workload and the vast majority of the SAN being composed of SATA you might be able to get raw storage costs down to $3/GB across an entire large storage system. But at the petascale we're talking about dozens of controllers, shelves, UPS and a massive switching infrastructure to support it, not to mention software licensing costs, maintenance contracts and a staff to run something of that size. You mention the costs of things like 10GbE and the storage system itself, but I can't tell from context if you're including all of those things in that per-gigabyte figure.
I'd also point out that backing virtualization installations with fiber channel is not even remotely uncommon anymore because of the density that the current generation of servers can deliver. Single RMU dual socket six-core x86 machines supporting 10, 15, 20 or more VMs is standard operating procedure these days and without something like 10Gb DCE you're going to be hard pressed to deliver enough IOPS over ethernet via iSCSI. Shit 20 VMs on a single machine at 8Gb/s FC you're talking about splitting up 1GB/s of storage throughput between 20 machines would mean only 50MB/s total throughput distributed evenly between them (fictional case to prove the point). That's a fraction of a single consumer grade SATA drive.
I'd be curious to see you're actual CapEx for acquiring storage. When you consider all of these factors I wouldn't at all be surprised to see something in the $30-$40 per gigabyte range. You'd need to include things like:
- Controllers (cache, licensing, maintenance, etc)
- Assuming 10GbE iSCSI you have to allocate some portion of that cost/management to your storage system
- Redundant systems? Maybe not an issue for your installations?
- Backups (frequency, off-site rotation, etc)
- Cost per square foot (lease(?)/HVAC/power)
- Cost to design/acquire/maintain over lifespan (think storage architect, sales meetings, other soft costs - rough figures are fine obviously)
And as you pointed out, what you can buy at the petabyte scale for let's say $20/GB CapEx could easily be in the $40/GB range for someone acquiring a storage system in the 20TB range. So to make blanket statements like "$40/gig is way more than the capital costs of the storage" is definitely, if I give you absolute benefit of the doubt, misleading at best.
Wish I could, SSDs aren't supported by my SAN vendor (yet). What vendors are selling tier0 SSD cache at only $7/GB? That's a great price I would think?
Sounds like an architectural problem. Have you considered moving their operating environments closer to the storage using something like Virtual Desktops or Citrix Presentation Server? You'd only need about 30kb/s to each endpoint and you could provide massive throughput easily between app servers and storage in the datacenter.
Very rarely do I see FC SAN SLAs written in the form of raw throughput. Typically it's quoted in the form of capacity and IOPS. Do you guarantee a specific IOPS figure for that price?
When did power/hvac/floorspace become soft costs? Those are standard OpEx expenditures and as far as I know are considered "hard costs". It's like you're saying that "hard costs" are CapEx and "soft costs" are OpEx which isn't the case at all.
How many IOPS? What disk layout (disk sizes, types, any short-stroking) ? How much redundancy? Your numbers don't really seem unreasonable depending on the performance and SLA. We pay something like $30/GB for raw FC storage (including the FC infrastructure) and if I assumed I could get 4 years out of it I could pay for it at $7/GB/year.
No you didn't. Not unless your definition of "mid-range" is 1Gb ethernet iSCSI and 7200RPM sata disks. Even then I would find it highly unlikely.
Maybe $3/GB for raw SATA storage. There's no way you're building out a complete fiber channel storage system for $3/GB CapEx. Not even remotely possible. Raw (un-RAIDed) FC disks are about $3/GB. In RAID10 that jumps to $6/GB in disks alone . No HBAs, no FC ports, no controllers, no backups, no redundancy, no software, no caches - nothing other than bare disks.
I'm sorry where are you getting fiber channel storage for $0.10 per gigabyte? That's like the cost of bare RAID1 consumer 7200RPM sata disks ($0.05 * 2). We pay $1,000 for a 400GB 10K RPM fiber channel drive. That doesn't include: (redundant) controllers, (redundant) fiber channel HBAs, (redundant) fiber channel ports and a million other associated costs (everything from support to backups). I'm not necessarily agreeing with $30/GB/mo but $0.10 per gigabyte for the raw storage is not even remotely plausible for enterprise storage.
What type of storage is that? If I added all the CapEx (including HBAs, FC ports, software licensing, loss of _usable_ space via RAID, etc) we're probably well over $40/GB for FC SAN space.
read: I have an unlimited budget and can build perfect(tm) IT systems. I have many many friends who are government contractors and the money they waste makes me literally physically ill. If I ran IT shop like the government we'd be bankrupt before the end of the year.
Which would mean higher costs to end users which would allow another company to undercut their prices and steal their customers. Isn't capitalism great?