groberts101
Golden Member
- Mar 17, 2011
- 1,390
- 0
- 0
you don't have to understand something to make it fact. Google is your friend.
Best analogy I can give you is a simple question.. why run "old gas" when the whole idea is to start fresh?
SE Q&A is here.
http://cmrr.ucsd.edu/people/Hughes/SecureErase.shtml
more details here.
http://www.destructdata.com/secure-erase.html
as far as the USB requirements? different standards altogether and I can only assume they don't intend them to be used as OS based drives in the first place.
For any other question as to the validity or need to SE your SSD's? First off.. this is Sandforce we're talking about here. To understand that up front eliminates many of these debates. They are NOT the same as other SSD's in many regards, most of which in this discussion is related to the drives mapping process.
If you go back to the first gen SF(or even this one for that matter), the drive keeps what's called a Durawrite map after all nand has been touched with writes at least once. This happens faster than you realize as even surfing the web will scratch data to your SSD(even moreso with streaming media) and the controller will ALWAYS attempt to write to fresh "clean reserve pool" blocks.
Reformatting/reinstalling or using "TRIM and erasing tools"(other than the proper SE command) will ABSOLUTELY NOT wipe a SF internal map. The biggest kicker is this. Whatever the controller was dealing with from a recovery standpoint(amount of clean blocks in reserve) prior to that reformat/reinstall will only get compounded that much more when you write more data to the drive(such as a fresh install or reimage will do). SE wipes ALL maps and will eliminate all but the worst LTT algorithm(such as hitting the drive with 40TB's worth of data in less than a week to see how long the flash will last).
Also consider that the Sandforce controller will have the ability to write contiguously after all maps have been wiped and all throttles(often called a "settled state"). This is because Durawrite requires the map to be in place to regain full control of wear leveling and "settled state" performance once again. This effect is much more exaggerated with the first gen SF.. but even this newer one has similar tendencies and internal Durawrite coding. It just skewes most peoples perspectives because it recovers in a much more aggressive manner due to better TRIM functionality(it's not so lazy as it used to be) and on-the-fly garbage collection which is all a direct result of the larger recycling engine.
So, just because you don't see the drive slow down from a throttle or the various hardware's internal ACPI tabling processes(maps)?.. does not mean that they don't exist. Because they do.. and many of your drives(not just Sandforce) will still have remnant data(some of which can "freak a Sandforce controller's tables all to hell") if not properly erased at the entire physical level. Simply forcing the controller and its internal mapping to "just deal with it".. is not the correct way to baseline an SSD.
And for that matter.. if it's mission critical enterprise type implementation we're talikng about here?(and I realize we're not but who are you to judge whether or not my data is worthy of such lofty standards?).. even those with simple HDD usually flip all bits to a known standard for the sake of proper baselines to build from. If it's good enough for enterprise?.. it's certainly good enough for me.
SE is cheap insurance and it cost's you barely a penny of electricity to do.
Best analogy I can give you is a simple question.. why run "old gas" when the whole idea is to start fresh?
SE Q&A is here.
http://cmrr.ucsd.edu/people/Hughes/SecureErase.shtml
more details here.
http://www.destructdata.com/secure-erase.html
as far as the USB requirements? different standards altogether and I can only assume they don't intend them to be used as OS based drives in the first place.
For any other question as to the validity or need to SE your SSD's? First off.. this is Sandforce we're talking about here. To understand that up front eliminates many of these debates. They are NOT the same as other SSD's in many regards, most of which in this discussion is related to the drives mapping process.
If you go back to the first gen SF(or even this one for that matter), the drive keeps what's called a Durawrite map after all nand has been touched with writes at least once. This happens faster than you realize as even surfing the web will scratch data to your SSD(even moreso with streaming media) and the controller will ALWAYS attempt to write to fresh "clean reserve pool" blocks.
Reformatting/reinstalling or using "TRIM and erasing tools"(other than the proper SE command) will ABSOLUTELY NOT wipe a SF internal map. The biggest kicker is this. Whatever the controller was dealing with from a recovery standpoint(amount of clean blocks in reserve) prior to that reformat/reinstall will only get compounded that much more when you write more data to the drive(such as a fresh install or reimage will do). SE wipes ALL maps and will eliminate all but the worst LTT algorithm(such as hitting the drive with 40TB's worth of data in less than a week to see how long the flash will last).
Also consider that the Sandforce controller will have the ability to write contiguously after all maps have been wiped and all throttles(often called a "settled state"). This is because Durawrite requires the map to be in place to regain full control of wear leveling and "settled state" performance once again. This effect is much more exaggerated with the first gen SF.. but even this newer one has similar tendencies and internal Durawrite coding. It just skewes most peoples perspectives because it recovers in a much more aggressive manner due to better TRIM functionality(it's not so lazy as it used to be) and on-the-fly garbage collection which is all a direct result of the larger recycling engine.
So, just because you don't see the drive slow down from a throttle or the various hardware's internal ACPI tabling processes(maps)?.. does not mean that they don't exist. Because they do.. and many of your drives(not just Sandforce) will still have remnant data(some of which can "freak a Sandforce controller's tables all to hell") if not properly erased at the entire physical level. Simply forcing the controller and its internal mapping to "just deal with it".. is not the correct way to baseline an SSD.
And for that matter.. if it's mission critical enterprise type implementation we're talikng about here?(and I realize we're not but who are you to judge whether or not my data is worthy of such lofty standards?).. even those with simple HDD usually flip all bits to a known standard for the sake of proper baselines to build from. If it's good enough for enterprise?.. it's certainly good enough for me.
SE is cheap insurance and it cost's you barely a penny of electricity to do.
Last edited: