Consumer Grade SSDs vs Enterprise Grade SSDs
-
@scottalanmiller said:
@DenisKelley said:
Nothing wrong with them. I have them in all my Workstations. What grinds my gears is that the HP drive is a re-branding of the Intel Enterprise drive at a substantial markup. See here:
That's what all warrantied drives have always been.
Sure (well, IBM did make their own for a while) - but the markup rate on SSD's is ridiculous compared to HDDs.
-
Is the markup that much more? Unlike HDs where there is no difference between the types essentially, I think that the markup tends to be a bit more dramatic.
I wonder if we do a comparison what the ratios look like.
-
@scottalanmiller said:
Is the markup that much more? Unlike HDs where there is no difference between the types essentially, I think that the markup tends to be a bit more dramatic.
I wonder if we do a comparison what the ratios look like.
Sure there are differences in the types of SSD - SLC, MSC, etc - but if consumer drives are now lasting 20+ GB a day, and that fits within your metric, the costs are outrageous.
The 1 TB drive is like $4500 from Dell/HP where you can get a consumer Samsung EVO 1 TB drive for $360, or the EVO Pro for under $500. -
I think that it is important to not think of the costs as outrageous but the reliability as being far too high. It's not that the cost of a tractor trailer is too much or that they are even expensive when you need to haul lots of stuff. But if you are just making a few trips a year to Home Depot then a used pickup truck is the better value for you. It's not that enterprise SSDs are expensive for what they are, it is that the are likely the wrong tool for the job.
-
Really? You think 9x the cost is even warranted for Enterprise mission critical stuff? Perhaps instead of rolling out RAID 5 on SSD with those less reliable drives you roll out RAID 10 and still save 4.5x the cost.
-
@Dashrender said:
Really? You think 9x the cost is even warranted for Enterprise mission critical stuff? Perhaps instead of rolling out RAID 5 on SSD with those less reliable drives you roll out RAID 10 and still save 4.5x the cost.
Are you looking at the differences in dollars per write? How much more expensive are they?
-
@scottalanmiller said:
@Dashrender said:
Really? You think 9x the cost is even warranted for Enterprise mission critical stuff? Perhaps instead of rolling out RAID 5 on SSD with those less reliable drives you roll out RAID 10 and still save 4.5x the cost.
Are you looking at the differences in dollars per write? How much more expensive are they?
Yeah OK - you definitely have a point there. Thanks. It would be interesting to see them side by side.
-
If you think about what people pay for enterprise hard drives versus consumer and how little you get. WD RE vs. WD Red Pro, for example. Sure you don't pay 10x more, but the only real different is in the URE rate. If you are not using parity RAID, that rate is worthless. So you pay 10% or 20% but get, quite often, nothing at all for it.
-
This doesn't help my understanding.
Going back to the pre write idea - It's likely that an enterprise could have huge amounts of writes versus a SMB (but some SMBs have huge writes too), the idea would be to see when the drives fail due to to many rights in a given situation and see if the extra cost of replacing the drives and the manpower to do said replacements, etc and see if the costs are justified.
The idea that HP/Dell/whomever will send you a free replacement drive doesn't sound all that worth while, since you can get the consumer drives with warranties as well, sure there's probably a bit more work (yes that work does have value for the equation).
-
@Dashrender said:
This doesn't help my understanding.
Traditionally, paying for enterprise drives might have gotten you literally nothing. So the extra cost was all waste, except for the warranty support. Today you at least, normally, get much higher reliability.
-
@Dashrender said:
The idea that HP/Dell/whomever will send you a free replacement drive doesn't sound all that worth while, since you can get the consumer drives with warranties as well, sure there's probably a bit more work (yes that work does have value for the equation).
Not comparable. HP and Dell will replace a drive even before it fully fails based on error rates, will do so with four hour response time and will run it to your shop and do the replacement for you. You are getting IT staff and extreme logistics as part of the warranty.
Standard warranties from the drive vendors mean that you have to wait for the drive to fail, do an RMA, ship the drive back, wait for a replacement and replace it yourself.
-
Take this to a datacenter level and it makes even more sense when you need people to send and receive the drives, do the replacement, etc.
-
@scottalanmiller said:
Take this to a datacenter level and it makes even more sense when you need people to send and receive the drives, do the replacement, etc.
Do you allow Dell/HP to enter the DC and replace drives when the reports say they are about to fail? At bare minimum I would expect the need to wait for a repair window.
-
@Dashrender said:
Do you allow Dell/HP to enter the DC and replace drives when the reports say they are about to fail? At bare minimum I would expect the need to wait for a repair window.
Assuming proper RAID redundency, there is no need for a maintenance window. WHy pay off hours rates?
-
@JaredBusch said:
Assuming proper RAID redundency, there is no need for a maintenance window. WHy pay off hours rates?
You mean anything other than a RAID 0, right?
I'm not sure about that if that was the case, then why is Scott so dead set against Hot Spares? Unless he's only against hot spares for spinning rust RAID 5.
-
@Dashrender said:
Do you allow Dell/HP to enter the DC and replace drives when the reports say they are about to fail? At bare minimum I would expect the need to wait for a repair window.
They do the work when you tell them to do the work. Generally, at least over 50% of the time, immediately drive replacement is best, but it all depends.
-
@JaredBusch said:
@Dashrender said:
Do you allow Dell/HP to enter the DC and replace drives when the reports say they are about to fail? At bare minimum I would expect the need to wait for a repair window.
Assuming proper RAID redundency, there is no need for a maintenance window. WHy pay off hours rates?
If you have a RAID 6 array, for example, you might have a performance impact during computational hours. Even working at the big bank, they generally made us wait until weekends to replace failed drives, which was a bit crazy.
-
@Dashrender said:
I'm not sure about that if that was the case, then why is Scott so dead set against Hot Spares? Unless he's only against hot spares for spinning rust RAID 5.
Hot Spares with R5 are insane as it could be a live part of the array making it a RAID 6. RAID 1, same thing, just make it part of the array. Both cases are insane to have hot spares.
With RAID 6 you can have them in situations where the hot spare does not push you unnecessarily close to RAID 10 and RAID 5.3 (aka RAID 7) is not available. But that's relatively uncommon.
In RAID 10 you can have them but they only make sense in very large arrays or cases where you just can't get to the array to swap the failed drives.
In some cases it is an architectural problem, in others it is that the cost is just not justified.
