Sign in with
Sign up | Sign in
Your question

Silicon Image Brings Virtualization to eSATA

Last response: in Storage
Share
May 29, 2007 12:14:49 PM

Finally, SATA comes of age: Silicon Image's new SteelVine storage processors combine USB 2.0 flexibility with eSATA performance. The technology allows users to cascade external hard drives into virtual storage volumes. RAID options or mirroring also help to increase data safety.
May 29, 2007 2:26:59 PM

(As usual) ... inaccuracies in the article:
[*:D ab638ca0e]The article implies that port multipliers are an eSATA-specific feature. This is not true, port multipliers can be used on any SATA connection, internal/external, SATA/eSATA if the controller supports them.
[*:D ab638ca0e]The description of the SAFE modes should not call the individualized sections of the RAID "partitions". A partition is a term used to describe a section of a device that contains a file system. The individualized RAID areas are not partitions, since they have nothing to do with the file system, but instead operate at the underlying virtualization layer. They are properly called "logical drives" or "containers".
[*:D ab638ca0e]The auto drive locking mode/feature appears to be somewhat of a misnomer. The security is reliant on the SIL57xx device to prevent access to the drive. Removing the drive from the external enclosure and connecting it internally to a different computer will probably bypass the security. I see nowhere that the SIL57xx devices do full drive encryption, which means the data would then be fully accessible.
[*:D ab638ca0e]Average read access time graph is meaningless. You're comparing external devices that use 3.5" drives against your SIL units which use 2.5" drives. The comparison is apples and oranges.
[*:D ab638ca0e]Read and write transfer rate graph is almost as meaningless. Multiple things vary across the devices (different interfaces, different drives, different controllers). The comparison means virtually nothing unless all variables are held constant except the one you're testing. Since the article is focusing on the logic embedded in the external controller, all tests should have been performed with the same drive and the same interface, varying only the external enclosure/controller. This is basic scientific method, which appears to have been thrown out the window here.
[*:D ab638ca0e]Nowhere in the article is it mentioned that virtualization of storage across multiple drives increases risk of data loss. If you operate these drives in BIG or FAST modes, risk increases proportional to the number of drives used, just like RAID 0. In BIG mode, some data can usually be recovered from the non-failed drives with a drive recovery program like GetDataBack, but nevertheless users will likely not understand the risk they're taking with their data. Especially now that the risk has been extended to external drives, which are subject to more shock, riskier electrical environments, and external cases which are notorious for overheating.
As storage needs increase, better solutions need to be found to ensure data integrity while offering better storage management. SIL's solution here is not ideal in my mind, but perhaps at least RAID 1 is better than nothing.
May 29, 2007 5:33:12 PM

Quote:
Finally, SATA comes of age: Silicon Image's new SteelVine storage processors combine USB 2.0 flexibility with eSATA performance. The technology allows users to cascade external hard drives into virtual storage volumes. RAID options or mirroring also help to increase data safety.


hmmm, I read "most" of this, it was long.., so do I understand this is a chip/hardware solution imbedded in an external device, right? I mean the host computer will not need an accompanying card to host this?

Also, would the slowest drive in your chain limit the performance of the whole chain it being the lowest common denominator? Just curious.

I would have like to have seen this tested on a variety of boards not just one uncommon server system unless its only designed for this type environment.
Related resources
May 29, 2007 11:13:28 PM

Quote:
Multiple drives make it difficult for users to manage their files

Now you know why *nix has never followed Microsoft's "treat every drive as a separate letter" attitude, cos it just "makes it difficult for users to manage their files" haha

For all of you who have no idea what i'm talking about, on Unix Linux or BSD there is no such thing as C: ,D: or A: etc.. everything begins with / and you can put any drive anywhere you want.

eg.
/home/mikky/movies can be sata 1
and /home/mikky/movies/porno can be sata 2
May 30, 2007 12:35:47 AM

Quote:
For all of you who have no idea what i'm talking about, on Unix Linux or BSD there is no such thing as C: ,D: or A: etc.. everything begins with / and you can put any drive anywhere you want.


Actually, Windows NT/2K/XP/Vista does support this, it's just very rarely used.

Go to Computer Management, go to Disk Management, right-click any non-boot partition, select "Change Drive Letter/Path", click "Change", and select "Mount in the following empty NTFS folder". This gives you the same Unix/Linux-like ability to mount different drives/partitions into the C:\ file system (or D:\, etc. if you want).

The limitation is that only local hard drive devices can be treated as such. Optical drives must still get their own drive letter, as must network drives or they must be referred to by UNC path (\\Server\Share).
!