This message is from the T13 list server.
I'm leaning in Hale and Bob's direction as well (speaking as a RAID implementer). However, we'll have to see how this proposal evolves. I have general security concerns about WRITE "WRONG", but it's no worse that the position today. (And it's really more of a complaint about pass-through than the commands themselves) ::>-----Original Message----- ::>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On ::>Behalf Of Hale Landis ::>Sent: Thursday, June 09, 2005 3:31 PM ::>To: [email protected] ::>Subject: Re: [t13] WRITE LONG, SCT Write Long, and WRITE WRONG EXT ::> ::>This message is from the T13 list server. ::> ::> ::>Sheffield, Robert L wrote: ::>> I think when RAID vendors begin to use the WRITE WRONG commands in ::>> practice, we'll find the capability it offers will be weak ::>with respect ::>> to what's needed for robust RAID algorithms. ::> ::>I would agree. I don't see how Write Wrong has much value ::>for any kind ::>of RAID implementation. And I'm not sure how it would work when the ::>physical sector contains more than one logical sector or how ::>it works ::>with the more complex ECC algorithms that will be used in the future. ::> ::>And related to this: I don't see how attempting to read and ::>use the data ::>in a sector that has an uncorrectable ECC error is much ::>value either. ::>Given today and future ECC algorithms it is likely that none ::>of the data ::>in a sector is of any value if there is an uncorrectable ECC error ::>anywhere in the sector's data. ::> ::>Metadata? Cool... Define a way the host can store metadata ::>associated ::>with each sector on the disk - expand the basic sector size ::>by 32 or 64 ::>bytes for the metadata and define new commands to access the ::>metadata. ::>(This would probebly be of more value to more people than ::>that stupid NV ::>cache thing.) ::> ::>Hale ::> ::>-- ::> ::>++ Hale Landis ++ www.ata-atapi.com ++ ::> ::>
