I will have to disgree with you on this one.
We have had 80GB's SSD and lower size for quite a while now. People want bigger SSD's aswell as lower cost one but not lower cost smaller ssd which is already on the market.
People also want TRIM working in RAID which would greatly increase the sales on smaller SSD's (maybe this is the reason they haven't bothered to fix it)
And why would you need 200 mb/s read and write on a boot drive?
I'm still scared to see what that 600GB SSD will cost.
Lower cost will come with the smaller process flash memory, and with increasing production. Higher capacity will also come, at the same and higher price points -as the technology becomes more established.
Right now, SSDs are firmly targetted at the enterprise market, where the very high speeds are most desirable, and where the customers are most willing to pay. In the server market, it's been random read/write that has been the bottleneck, not capacity. Customers who currently buy 12 15k SAS drives in RAID may instead choose to upgrade to a single SSD. Even the value type drives tend to be targetted at business desktops where the apps justify the cost, and where large storage isn't required. The thing is that the business market is the market most likely to accept the cost, so as long as there is a limit in terms of production capacity - they may as well only target that segment.
TRIM in RAID isn't really in issue from the manufacturer's perspective. That's an OS issue - and as it is, a very difficult one. TRIM violates the fundamental assumption of RAID, which is that the storage contents are determinate. The contents of a TRIMmed sector are undefined (they may be 0, they may be random, etc). - this puts huge restrictions of using TRIM in a redundant array (e.g. whole stripe only in RAID 5/6) and still causes a problem - because it makes it impossible to check the array for consistency (TRIMmed sectors on different drives may not necessarily contain consistent data). Not only that but TRIM is horribly slow and will 'stall' a drive while it runs - this isn't ideal for a high performance storage system. Only RAID 0 is simple to implement TRIM on - but RAID 0 has no place in the business segment where SSDs are currently targetted - so there's little to be gained by investing R&D time in TRIM on RAID support (especially as implementing TRIM on RAID support would approach the complexity of building the entire RAID system itself).
What would be better than TRIM support - would be good quality background garbage collection which would ensure a decent pool of immediately writeable flash pages - without the need to rely on specific OS versions and specific drivers for optimal performance. We already have the sandforce controllers that do this, and the next gen Intel controllers should also do this - hopefully, these designs should render TRIM largely obsolete.
There are many reasons why you might want speed on a 'boot drive'. Even small SSDs have plenty of space for a OS and core apps. Some apps (like databases, development work, etc.) are highly dependent on rapid access to their data, yet may only need a few GB of capacity. For that a 40 or 80 GB drive is plenty.
As for the maximum capacity drives - if sir has to ask the price, sir would not be able to afford it.
Facebook
Twitter