NAS advice

[stevech]

http://storagemojo.com/2007/02/26/netapp-weighs-in-on-disks/

Interesting read - not the blog post - but the netapp discussion within it - (NetAPP builds big enterprise grade SAN's, and they have a lot of data on drives, RAID, and failure modes)

Focus is on disk farms. I think that in SOHO/SMB NASes, theft, human error, file system corruption, are the key risks - moreso than drive failure. This, after the initial new-drive failure time is past.

 

[sfx2000]

Focus is on disk farms. I think that in SOHO/SMB NASes, theft, human error, file system corruption, are the key risks - moreso than drive failure. This, after the initial new-drive failure time is past.

Nice to see stats - RAID5 is esp at risk - and while we have a bathtub failure curve, the 2 year mark for drives in service is a key point with a distributed file system - which even a target market like this - SOHO/SMB - is a concern, as most folks don't understand that while RAID might provide some level of redundancy - more disks, more risks, as MBTF is per disk, and goes down when in a RAID...

I've seen this first hand in a data center environment...

 

[kamina]

http://storagemojo.com/2007/02/26/netapp-weighs-in-on-disks/

Interesting read - not the blog post - but the netapp discussion within it - (NetAPP builds big enterprise grade SAN's, and they have a lot of data on drives, RAID, and failure modes)

I used to manage a netapp cluster with several shelves of FC and Sata disk. It was a really cool device, but having worked with it I can feel some PR in the netapp comments... On the other hand they were pretty good at the things they highlight so I would assume the ideas are correct (Raid 6 or in their case Raid DP instead of Raid 5). They also have a pretty good understanding of firmware I would imagine since even though the drives in their systems are manufacturers by various vendors they make their own firmwares for them.

 

[sfx2000]

I used to manage a netapp cluster with several shelves of FC and Sata disk. It was a really cool device, but having worked with it I can feel some PR in the netapp comments... On the other hand they were pretty good at the things they highlight so I would assume the ideas are correct (Raid 6 or in their case Raid DP instead of Raid 5). They also have a pretty good understanding of firmware I would imagine since even though the drives in their systems are manufacturers by various vendors they make their own firmwares for them.

One of the older platforms I had (48 Sun T1000's) had a big NetApp as a filer... easy enough to manage and cheap relative to the Sun solution for the same function.

 

[stevech]

My related story: On a professional project I lead, I chose a local data center provider. Substantial. Paid for a VM service for a long time. One day, it was gone. After 3 days, the service provider says:
your VM was on a system with RAID5 and that array died. Try as we might, we could not rebuild. All is lost. Sorry about the SLA contract... the salesman who sold that mistakenly put on your order that there was an SLA but there isn't an SLA for the plan you have and we don't sell that plan anymore. Unless you had bought plan xxx that would have had a frequent backup... [where that was 20 times more costly]. Scam.

I had done my own backup across the internet a few weeks earlier, and some small incrementals. The service provider created a new blank VM (with MS Windows again) and I was, after a lot of time, able to get back on the air. Due solely to my own backup. Despite the faux paus, they would not let me enter their data center with my USB drive to do a fast restore, nor would they do so. They said "our staff isn't trained on how to do that".

The server was the application monitor for syslog events for 100 unattended sites in 3 countries. It also logged application message metrics for this fielded pre-production system.

This highly marketed data center is on my black list, of course.

 

[L&LD]

My related story: On a professional project I lead, I chose a local data center provider. Substantial. Paid for a VM service for a long time. One day, it was gone. After 3 days, the service provider says:
your VM was on a system with RAID5 and that array died. Try as we might, we could not rebuild. All is lost. Sorry about the SLA contract... the salesman who sold that mistakenly put on your order that there was an SLA but there isn't an SLA for the plan you have and we don't sell that plan anymore. Unless you had bought plan xxx that would have had a frequent backup... [where that was 20 times more costly]. Scam.

I had done my own backup across the internet a few weeks earlier, and some small incrementals. The service provider created a new blank VM (with MS Windows again) and I was, after a lot of time, able to get back on the air. Due solely to my own backup. Despite the faux paus, they would not let me enter their data center with my USB drive to do a fast restore, nor would they do so. They said "our staff isn't trained on how to do that".

The server was the application monitor for syslog events for 100 unattended sites in 3 countries. It also logged application message metrics for this fielded pre-production system.

This highly marketed data center is on my black list, of course.

You should post a link to this story in the thread where people think everything 'cloud' is all rosy and cheerful.

As I've said; if you don't own it, you shouldn't trust it. And there is nothing you own 'in the cloud'. I am guessing that even with an SLA in place (at 20x the cost) they would have still left you high and dry. They already got your money, your problems? They can now do without.

 

[stevech]

I have to admit... the VM service was "too cheap" compared to competitors'. I should have not trusted that sales guy. He's long gone.

To be fair... using a cloud service for backups is a far lower risk, than using a VM service that is live 24/7.

 

[kamina]

I guess the important thing to understand is that the cloud is just somebody elses 24/7 VM...


Sent from my iPhone using Tapatalk