0

There is a file server (Windows Server 2008 R2) running on a hardware RAID 1 (HP SmartArray P212 with Read and Write Cache) with 2 standby disks. Additional, there is a Windows Backup writing to an attached SATA disk.

When the server hardware (or raid controller) crashes, afaik there is no possibility to directly access the data stored on the raid disks by simply attaching them to a foreign system (sata, usb) without the system having the same hardware raid controller? Is this correct?

To make the data still available on hardware crash, my intention is the following: Adding an iSCSI target (Synology DS720+) to the network, mount this target on the file server and create a (software) RAID1 containing the hardware raid + the mounted iSCSI target. Now in desaster case, i could simply mount the iSCSI target on any other machine and access its data.

Would this be possible?

JKB
  • 103
  • 3
  • Note that a simple NAS is another single point of failure - the result is *lower* availability than internal DAS since more parts can fail. If you're serious about resilience, you use a storage array with dual controllers, connected via redundant paths. – Zac67 Dec 14 '20 at 21:59

1 Answers1

2

When the server hardware (or raid controller) crashes, afaik there is no possibility to directly access the data stored on the raid disks by simply attaching them to a foreign system (sata, usb) without the system having the same hardware raid controller? Is this correct?

Yes, but if you are even borderline professional that does not matter because you have both, spare parts AND a backup that can both be used to get reasonably fast access to the data.

To make the data still available on hardware crash, my intention is the following: Adding an iSCSI target (Synology DS720+) to the network, mount this target on the file server and create a (software) RAID1 containing the hardware raid + the mounted iSCSI target. Now in desaster case, i could simply mount the iSCSI target on any other machine and access its data.

And that would be better compared to a spare raid controller (WAY cheaper) and a backup (WAY safer) how? Note that while it may handle a hardware failure reasonably fast (at the cost of performance) you are still totally open to data manipulation (i.e. through a failing memory chip) - which a backup ALSO allows you to roll back.

TomTom
  • 51,649
  • 7
  • 54
  • 136
  • I should try to get a spare controller, of course. But i would even need a spare server. To access the files from the windows backup, wouldn't i need another running windows server? Thats why i thought about the additional iSCSI: To be able to access the data immediately, and in parallel have the time to replace the failed hardware, restore a backup or setup a new system. It is only for SOHO. – JKB Nov 27 '20 at 11:18
  • " But i would even need a spare server. " - yes, which is why you should standardize the setup and then just have ONE more server (of x). That said, you need a spare server ANYWAY because the way you make it sound - well, if the server fails the iscsi target needs another server ;) – TomTom Nov 27 '20 at 11:50
  • But i could attach the iSCSI target to a client pc while setting up a new server! But you are right, i should better go for a complete spare server. Also, because i have move the 2008 r2 to a newer version somewhen... Last question: When replacing a hardware raid controller (with an identical spare part), and attaching the old disks, would the data immediatly be available? – JKB Nov 27 '20 at 12:27
  • Depends on the hardware. If you go i.e. for Adaptec / Microsemi / whatever they are called now, they write all raid configuration of a controller ON EVERY DISC - so if you attach another controller he can find the matching discs and see whether anything is missing. Some other controllers may require you to manually enter the configuration. Sirry, but this is a RTFM case - with the FM being specific per controller. Microsemi user here. – TomTom Nov 27 '20 at 14:15