SSDs - State of the Product?

snowhiker said:
Anand reviews the OCZ Z-Drive R4 CM88 (1.6TB PCIe SSD) Review

http://www.anandtech.com/show/4879/ocz-zdrive-r4-cm88-16tb-pcie-ssd-review
Click to expand...

It seems to me that there are only two issues with that: Cost and reliability. If you truly need that level of performance then you can probably pay. If you need increased reliability as well as the performance then mirror-raid it, and run backups of course. Further, if you can afford one then you can still probably afford two.

One does wonder how software-level mirroring would affect performance though...
 
It would make an excellent cache drive to a 100TB+ array. I can only imagine how well that would perform, and reliability would be much less of an issue.
 
BingBangBop said:
It seems to me that there are only two issues with that: Cost and reliability.
Click to expand...

What about service? Does OCZ treat the enterprise customers better than they do the peasants buying a few Z-Drives on the internet?
 
ddrueding said:
It would make an excellent cache drive to a 100TB+ array. I can only imagine how well that would perform, and reliability would be much less of an issue.
Click to expand...

I imagine write performance would be good, but read performance would be slow most of the time and fast only if the data happens to be in the SSSD.
 
The only way to increase cache hits for reads is to make it smarter or bigger. Making the read cache >1% of the total drive size is quite generous compared to most, and would probably work quite well with some decent software making the predictions.

Of course, my largest corporate (many simultaneous users) file share would practically fit on the largest of these drives, and then the problem would truly be solved ;)
 
SR too. I am underwhelmed and still waiting for the Intel 720 (PCIe version).
 
Indeed. That they have finally built a drive that nearly matches the X25-E is not what I wanted to hear. Let's see what Intel can do when they get to roll their own all the way to the PCIe interface.
 
Remember, I had the OCZ Vertex Turbo fail. It was replaced with a drive that would not hold a SATA cable. That was replaced with a drive that is in the HTPC, and, as little as it's used, is working fine.
 
I'm sure that large companies would rather buy Intel or Micorn/Crucial and not split hairs over performance of individual drives.
 
Is RAID1 even an issue here? In theory RAID1 means doing the same thing to both drives regardless. There's no splitting data or parity calculations, so I'm thinking there isn't a complicating factor.
 
Other than RAID-0, I don't see a compelling need in the SSD space at this time. The cost/GB is still high enough that your redundancy should be in HDDs wherever possible.
 
So where are we now, 3 months later? Which is the best performing, most reliable single SSD these days?
 
ddrueding said:
Have one coming now. I'll stick it in a DB VM and see how long it lasts. The last drive in that machine (480GB Corsair Force GT) kept disconnecting itself under really high IO.
Click to expand...

I'm more curious to know if that's a characteristic of the qualification methodology they use. The conventional wisdom still says to keep SSDs out of critical business stuff unless there are a bunch of other layers of redundancy available. Did anybody at Corsair even test that kind of workload?
 
ddrueding said:
The last drive in that machine (480GB Corsair Force GT) kept disconnecting itself under really high IO.
Click to expand...
I don't associate disconnection as a performance issue or a longevity problem rather isn't it more likely to be a design flaw that can be solved by picking a totally different design. Of course the 830 is totally different so good luck.

If the usage is a database VM that has a significant amount of writing going on then the 710 might have been a better choice as an enterprise SSD designed for lots of writes rather than a consumer SSD that isn't. In reality, a server running a database that is written to a lot can wear out a consumer drive relatively quickly. That was the whole point of the article that was linked to for the Intel 710 -- How long different SSD's would last in a server database environment.
 
Mercutio said:
I'm more curious to know if that's a characteristic of the qualification methodology they use. The conventional wisdom still says to keep SSDs out of critical business stuff unless there are a bunch of other layers of redundancy available. Did anybody at Corsair even test that kind of workload?
Click to expand...

No idea. That disk will now go into a high-end workstation and will likely never have an issue again. From Intel drives I expect more ;)
 
BingBangBop said:
If the usage is a database VM that has a significant amount of writing going on then the 710 might have been a better choice as an enterprise SSD designed for lots of writes rather than a consumer SSD that isn't. In reality, a server running a database that is written to a lot can wear out a consumer drive relatively quickly. That was the whole point of the article that was linked to for the Intel 710 -- How long different SSD's would last in a server database environment.
Click to expand...

Yeah, I'm not into throwing that much money at this particular problem yet. I do have a good enough backup/failover strategy that it isn't really an issue.
 
You must log in or register to reply here.
Back
Top