The problem with RAID6 is that it's not supported by most controllers, especially if it's a year or two old. My new EMC AX4-5 doesn't even support it. I'm doing RAID10 on it, though 6 would have given me some more usable space. I just set it up with two hotspares so I feel alright about the possibility of an error.
This weakness is really only an issue once you start getting towards an immense amount of data. As the blog entry states, one failure is statistically likely during 14TB of data recover (not storage volume, data volume). So 7 1TB drives in a full RAID5 array has a 50% chance of failure during recovery if one of the drives were to fail.
Home users won't be affected, at least at this point. I imagine that by the time it does become an issue, manufacturing tech will have caught up and it'll be 14PB.
Except that we receive client data through email. So if we don't get client data for 2 days, we're going to get fired, and that means a heck of a loss of revenue.
I work in a 24x7 computing environment for a financial services company. I literally don't have a service window for disabling client access. This means I've got to have backup systems that can take over at a moment's notice.
Just because there are 20 people in my company doesn't mean that there isn't a large, large amount of money at stake to our clients. Individual machine uptime isn't as big of a deal as service availability being as close to 100% as possible.
We are rapidly approaching the point where "Microsoft didn't adopt it" doesn't matter any more. In case you've not noticed, Microsoft has not done a lot recently to ingratiate themselves with anyone in the IT world.
Since I've been shopping for email hosting, I've encountered a few misguided sales people who try to convince me that they actually provide 100% uptime.
The most amusing case was the company who offered two hosting levels. One, which guaranteed 99.9% uptime for their shared hosting, kept their shared customers' accounts on a clustered set of Exchange hosts, with data stored on an enterprise SAN. The 100% uptime was reserved for those people who paid an extra $500 a month. Those lucky customers had an exchange server running on an off-the-shelf Dell box that wasn't connected to the SAN for backup at all.
I really did laugh at the engineer and salesweenie as I thanked them for their time and hung up.
If Google's datacenter burns to the ground, how long do you think you'll be without your GMail account?
But that's one of the strengths of using gmail.
The loss of a complete datacenter for them would be a minor issue which would cause slow access. Accidental removal of one part of their infrastructure doesn't cause much of a problem. It's the administrative issues that get replicated across the network that cause major issues.
It took a while to decide, but I submitted a proposal to host it with Apptix. I spoke with an engineer, and it sounds like they've got a very reliable clustered backend. Add their Blackberry services and compliant archiving, and it sounds like a good solid solution.
Of course, two years and I might change my mind, but that's the risk you take.
It's an issue for personal customers, but for corporate entities who have outsourced their email to Google, it's a bit more than an irritation. For the admin who decided to outsource the mail to there, it could be a Resume Generating Event
My wife and I struggle with that all the damned time. I keep wanting to see him as a pretty boy whiner kid, but like you said, he keeps doing things like that. I eventually forgot that it was him in Blood Diamond. He's a good actor, as much as I hate to admit it.
Not to be the trek fanboy I used to be, but Kirk was the youngest captain in starfleet history. I'm assuming this is before he was legendary, and I'm sure they're going to be making the movies about how he/became/ legendary. Anyway, carry on.
PS - The Klingons didn't have a word for surrender...until they met Kirk
Worked tech support in a crappy company. Used that experience to do tech support in a small ISP. Used my linux/networking skills to become the assistant network admin, got the main network admin job when the old guy left, and used that experience to get my current fun job.
Slashdot Mirror
The problem with RAID6 is that it's not supported by most controllers, especially if it's a year or two old. My new EMC AX4-5 doesn't even support it. I'm doing RAID10 on it, though 6 would have given me some more usable space. I just set it up with two hotspares so I feel alright about the possibility of an error.
wow, I don't envy that period of time in your life, to be sure.
Nice shock value, but I'm not sure I'd trust to fate to rebuild the array every time. The risk/reward ratio isn't in balance to me.
This weakness is really only an issue once you start getting towards an immense amount of data. As the blog entry states, one failure is statistically likely during 14TB of data recover (not storage volume, data volume). So 7 1TB drives in a full RAID5 array has a 50% chance of failure during recovery if one of the drives were to fail.
Home users won't be affected, at least at this point. I imagine that by the time it does become an issue, manufacturing tech will have caught up and it'll be 14PB.
I did cover this earlier today, too.
I came into this thread to post the same message for the most part. Though as size increases, so does risk of failure, as I'm finding out.
*sigh* I know ;-)
Except that we receive client data through email. So if we don't get client data for 2 days, we're going to get fired, and that means a heck of a loss of revenue.
I work in a 24x7 computing environment for a financial services company. I literally don't have a service window for disabling client access. This means I've got to have backup systems that can take over at a moment's notice.
Just because there are 20 people in my company doesn't mean that there isn't a large, large amount of money at stake to our clients. Individual machine uptime isn't as big of a deal as service availability being as close to 100% as possible.
Trust me, I curse Apple developers pretty frequently...it's not just you
I think they're using this as a lever to push the audio-editors to the Pro models.
We are rapidly approaching the point where "Microsoft didn't adopt it" doesn't matter any more. In case you've not noticed, Microsoft has not done a lot recently to ingratiate themselves with anyone in the IT world.
Since I've been shopping for email hosting, I've encountered a few misguided sales people who try to convince me that they actually provide 100% uptime.
The most amusing case was the company who offered two hosting levels. One, which guaranteed 99.9% uptime for their shared hosting, kept their shared customers' accounts on a clustered set of Exchange hosts, with data stored on an enterprise SAN. The 100% uptime was reserved for those people who paid an extra $500 a month. Those lucky customers had an exchange server running on an off-the-shelf Dell box that wasn't connected to the SAN for backup at all.
I really did laugh at the engineer and salesweenie as I thanked them for their time and hung up.
Out of curiosity, what did he say that was untrue?
As long as it's a planned outage. Otherwise, you'll be letting them know as you're trying to fix it.
If Google's datacenter burns to the ground, how long do you think you'll be without your GMail account?
But that's one of the strengths of using gmail.
The loss of a complete datacenter for them would be a minor issue which would cause slow access. Accidental removal of one part of their infrastructure doesn't cause much of a problem. It's the administrative issues that get replicated across the network that cause major issues.
"so what" isn't a valid option for corporate clients, which they sell to.
I'm getting ready to move my company to a hosted exchange provider.
It took a while to decide, but I submitted a proposal to host it with Apptix. I spoke with an engineer, and it sounds like they've got a very reliable clustered backend. Add their Blackberry services and compliant archiving, and it sounds like a good solid solution.
Of course, two years and I might change my mind, but that's the risk you take.
You've got to love IT, where one-in-a-million events occur every other day...
It's an issue for personal customers, but for corporate entities who have outsourced their email to Google, it's a bit more than an irritation. For the admin who decided to outsource the mail to there, it could be a Resume Generating Event
Eh, I just forcequit the updater process and reboot on my own terms
My wife and I struggle with that all the damned time. I keep wanting to see him as a pretty boy whiner kid, but like you said, he keeps doing things like that. I eventually forgot that it was him in Blood Diamond. He's a good actor, as much as I hate to admit it.
Not to be the trek fanboy I used to be, but Kirk was the youngest captain in starfleet history. I'm assuming this is before he was legendary, and I'm sure they're going to be making the movies about how he /became/ legendary. Anyway, carry on.
PS - The Klingons didn't have a word for surrender...until they met Kirk
Much love to the Intellivision.
I miss Sewer Sam, Stampede, and Donkey Kong on there.
I've used ppracer since they closed the source, but the maps seem freaky and inconsistent
You've really come across companies who will look at someone over 6 months out of school and ask what their GPA was, in a technical field?
And from what I've seen of the many engineers I know, co-ops are really the key to getting employed after school
That's exactly how I became a sysadmin.
Worked tech support in a crappy company. Used that experience to do tech support in a small ISP. Used my linux/networking skills to become the assistant network admin, got the main network admin job when the old guy left, and used that experience to get my current fun job.
It can happen.