Intel 520 240GB SSD The rumour mill was working overtime throughout 2011 with suggestions that Intel was considering forsaking the Marvell controllers it had used for its previous generation of consumer SSDs, the 510 series. The story went that the company was jumping into bed with another third party controller manufacturer, namely Sandforce, for …
"Hence, the 240GB capacity, which, when the drive is formatted, drops even further to 224GB"
Actually the capacity doesn't drop when formatted, rather you change from using the hard disk drive manufacturer's standard decimal-based measurement of 1 GB = 10 to the power 9 bytes, to using the more widely used (at least, by filesystems) binary-based measurement of 1GB as 1024 to the power 3.
"""Although 16 x 16GB memory chips populate the 240GB Intel 520, one is used for firmware and other tasks"""
Actually, I believe you'll find that Sandforce do a Raid5-style striped XOR across their flash, so they can tolerate an entire NAND die failure with no data loss. They still have ~7% spare capacity for firmware and things, from the GiB to GB conversion.
@Giles: "The question is why is Intel (aka Chipzilla) using other people's controllers? where's their own design?"
They've still got their own controller in the 320, which is still untouchable as far as price / features / performance for large segments of enterprise users. The 5x0 series is for gamers, who need every last benchmark win - when Intel released the 510 they said they could make the most money in that segment with a third party controller.
@Anon: "Super capacitor?"
1) Super Caps are so last year - they tend to wear out quickly at the elevated temperatures found within computers. They've been replaced by arrays of SMD electrolytic caps, which last quite a lot longer. 2) Clearly from the circuit board pictures, this drive doesn't have either sort of power fail protection. 3) Sandforce only offers power fail on their enterprise controllers (2500 for SATA and 2600 for SAS,) though not all drives with those controllers actually have the capacitors.
But yeah, power fail protection is essential, even for desktop use.
I imagine most people are most interested in SSDs for their speed. The next most interesting number to me is power consumption. It's a shame that it doesn't feature much in these reviews.
My laptop runs continuously, mostly used as a desktop replacement rather than being lugged around. I was keen to replace the HDD with a SSD to increase reliability and lower power consumption, until I discovered that most SSD seem to have about the same power consumption as the equivalent HDD. That was disappointing. Would be nice if power consumption were included in the summary box.
I still need convincing that SSDs are reliable.
We know they've limited rewrite life, and at densities such as 480GB, reliability is a tall ask.
If one takes the time to understand how storage is achieved within SSDs then you'll wonder how they work at all. Essentially, a charge has to be stored in a sort of squishy glass dielectric and either remain there indefinitely or be rewritten to/refreshed indefinitely (and in practice, we're told this is limited). Moreover, cells become more strained when opposite charges--data states 0,1--are nearby/adjacent to one another (this is a full study in and of itself).
Dielectrics fail with use and/or over-stressing, it's a well know material science problem and high speed SSDs are very new and the coupling between cells is high. Albeit very different but it'll do for an analogy, electrolytic capacitors are often changed because of dielectric failure, and for capacitors, this has been an intrinsic problem since the early days of radio nearly a century ago (and the problem is still with us).
Essentially, SSD storage is intrinsically unstable over the long term (and we still don't have practical experience/data to know how reliable they'll be over time). So it remains to be seen how they work out in practice, certainly for the moment I only use them for OSes or where backups are guaranteed and regular.
This issue is further complicated by the fact that manufactures don't publish proper failure rates or the modus operandi of failures versus time and usage.
What FUD, it's not like spinning drives are that reliable... I've had more of them fail on me in the last year or two than SSDs, none of which have had any issues.
"certainly for the moment I only use them for OSes or where backups are guaranteed and regular."
You'd be a complete dumbass to trust a single hard drive with something important and not regularly backed up. This is why humanity invented RAID, and why backup is still a word that gets used rather than consigned to history.
... with dropping scales, the write cycles available drop too. We're down to what, a couple hundred cycles for this size two-level cells, and it gets worse for the even smaller and even more levels that're coming up. Wear level and all that mitigate the direct trouble quite a bit, but reliability hasn't really been pinned down yet, plus wonky firmwares and all that. Besides, while spinning platters aren't that great, they've been around for a long time and we have some inkling when to move on to the next. Not so much with SSDs yet. Things are bound to change, of course. But right now, where using disk without backups isn't wise, using SSDs without backup is much less so, about as much less as they're faster than disk.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
