Read more.Essentially twin 7TB drives in a Helium filled chassis, this device is capable of 524MB/s.
Read more.Essentially twin 7TB drives in a Helium filled chassis, this device is capable of 524MB/s.
Pretty impressive for them to choose a colour scheme that already looks dated
It is not dated, but vintage!
What color you like your 67 SS in ? rainbow ??
Friesiansam (24-05-2021)
Sooo, a jbod array of 7tb disks inside a single physical unit?
Didn't I mention "hundreds of spindles"?
I'm used to working with enterprise drives sold by the petabyte, you get performance by distributing across racks full of drives. I'm just surprised that someone would pay for a doubling in performance, when if IOPS matters you get an all flash array or else you just buy twice as many cheap drives.
I guess if you have lots of smallish servers with just a few drives in each they might help.
So twice the failure rate in a single 3.5" form factor!
Had multiple failures in our dedicated server over at rackspace before moving to AWS, all the failure were seagate, personal usage I tried seagate and it failed along with the RMA they gave me!
Never again but this is my own anecdotal evidence of course.
Wonder why they have even bothered at this point as they cannot match the IOPS of SSD and they dont really really need two separate drives in a physical case as that will make it more prone to failure and more at risk.
It is a very lucrative market. The market for quick storage has gone SSD, but the market for things like bulk cloud storage is massive and still very much belongs to spinning rust.
At first that made sense to me; no-one has a driver optimised for a disk split across two regions, but if you just drive it as two disks then that's simple.
But on further thought it isn't. If one half dies, you are swapping both halves out of your RAID array so a classic RAID controller won't really know what to do with this and you will need to do some creative things with stripe sets. It works for Ceph, where the other copies of data will be on other machines so not on the other half of this spindle.
And it's a Seagate, so failure modes are important
The other possibility is that all drivers can do out of order tagged transactions, even SATA. So they could have just made this 14TB, That would make it harder to benchmark as twice as fast, but in the times when it mattered it would be.
There are currently 1 users browsing this thread. (0 members and 1 guests)