Seagate Is Bringing RAID 0 Performance to Single HDDs via Multi-Actuator Tech

Megalith

24-bit/48kHz
Staff member
Joined
Aug 20, 2006
Messages
13,000
Seagate has unveiled its new Multi Actuator technology, a breakthrough that can double the data performance of its future-generation hard drives in hyperscale data centers. With two actuators, a hard drive can double its performance while maintaining the same capacity as that of a single actuator drive.

An actuator is the component that moves a hard drive’s heads over the media surface, to read and write data. In its first generation, Seagate’s Multi Actuator technology will equip hard drives with dual actuators (two actuators). With two actuators operating on a single pivot point, each actuator will control half of the drive’s arms. Half the drive’s recording heads will operate together as a unit, while the other half will operate independently as a separate unit.
 
If each actuator were cable of reading the whole drive if needed then doubling them would add not only performance gains but would also build some redundancy in.

At least that would be possible if it weren't a Seagate.
 
2wice the speed 2wice the parts 2wice the failure.

Seagate: cause your data? fuck it.
 
This could actually greatly increase the chance of recovery. With that said I remember hearing this idea over a decade ago and it has not happened.

Edit: The picture is not the same as the old idea. I thought the actuators were independent and had 2 heads on each side of a platter.
 
This could actually greatly increase the chance of recovery. With that said I remember hearing this idea over a decade ago and it has not happened.
How? All you need is one crashed head from either of the two arms and your data is hosed.
 
i dont know why, but i had thought that the data was striped across all the heads for some time now.
 
Guess ssd tech is putting pressure on the HD market and encouraging development of new tech.

You know the other companies in the HD field will be the first in line to get one and follow closely how well the tech works.
 
twice the performance for more than twice the failure rate. considering abysmal HDD performance, twice of very little is still very little, and less reliable.
 
I'm kinda surprised no one released this sort of tech already given how old HDDs are.

It's one of those things that make so much sense given how "obvious" it sort of is. Even using something like stripped down 2.5" drives one could get two, maybe three, inside a 3.5" drive and have them running a native RAID mode.

I mean, their idea here is better I'm sure but, yeah, why has this not been done before? Double or triple performance HDD's would be nice.
 
They didn't because it's a terrible idea. Doubling the chance of failure in a mechanical device is just stupid.

You're right that it adds failure points, but if all of the heads can read the whole drive then it adds redundancy.
 
this is the dumbest thing I think I have ever seen. one of the #1 reasons for having a mirror in the first place, is that you can replace the drive that failed. can you replace the drive that failed in this scenario? not without a clean room you can't.
 
You're right that it adds failure points, but if all of the heads can read the whole drive then it adds redundancy.
From the pic it seems half and half.
I guess technically if you can do Raid 0 you could also do Raid1. Not that I think its a good Idea.
 
This has been tried before, several times actually... The first was by Conner back in the early 90s, but failure rates were too high. Seagate talked about the idea several times but no drives were ever released that I can find, probably because of failure rates and the design being financially impractical (or because the drive would have to be longer, adding compatibility to the list of challenges).

Here is a conversation about it from 2009 where Seagate said it was impractical http://www.tomshardware.com/news/seagate-hdd-harddrive,8279.html - and that was before the age of practical SSDs that hard drives could never keep up with! At this point, the whole idea of a "premium" hard drive makes no sense, the price of it will be as high as an SSD without even coming close to the speed.
 
Seagate has unveiled its new Multi Actuator technology, a breakthrough that can double the data performance of its future-generation hard drives in hyperscale data centers. With two actuators, a hard drive can double its performance while maintaining the same capacity as that of a single actuator drive.

An actuator is the component that moves a hard drive’s heads over the media surface, to read and write data. In its first generation, Seagate’s Multi Actuator technology will equip hard drives with dual actuators (two actuators). With two actuators operating on a single pivot point, each actuator will control half of the drive’s arms. Half the drive’s recording heads will operate together as a unit, while the other half will operate independently as a separate unit.

so basically two harddrives in the same shell. But I don't see the benefit at all. Can't you stripe the data across 8 platters and put 1 bit on each platter? Or 2 bits over 4 platters. (Plus parity on a 5th?) What's the point?
 
This looks like a bad ROI/idea to me. You have two actuators, but they are stacked vertically so each can only access half the platters at a time...so you only get double performance if the data you want is spread equally on the two platters...or are they planning on mirroring all the data on the two sets of platters? Two actuators would make sense if they were on opposite sides of the drive, but then you would have size/space issues. Looks like another solution in search of a problem. They would be better off spending their R&D money on reliability improvement or SSD's. Or, dump some extra cheap cache in there and move on.
 
You're right that it adds failure points, but if all of the heads can read the whole drive then it adds redundancy.

No way they can read the whole drive. The two sets of heads are on opposite ends of the platter stack. I would hope they can at least operate with half the drive's total capacity if one actuator fails, but that's the best you could hope for.
 
I'm curious--when hard drives fail, what is usually the cause? Is it a head crash? Firmware corruption in the controller board? Overheating leading to demagnetization?

here's another idea: could you put a second set of heads on the opposite side of the platter stack, i.e. diametrically opposed?
 
No way they can read the whole drive. The two sets of heads are on opposite ends of the platter stack. I would hope they can at least operate with half the drive's total capacity if one actuator fails, but that's the best you could hope for.

There's an animation in the article that I misinterpreted. If they do something like RAID 1 across both sets then in theory you're fine.
 
No way they can read the whole drive. The two sets of heads are on opposite ends of the platter stack. I would hope they can at least operate with half the drive's total capacity if one actuator fails, but that's the best you could hope for.

I would expect that would not be an operating mode for the same reason when a head fails on a 3 platter drive you don't 5 / 6 ths of a drive. With that said the firmware could be written to support this. They just don't do it in practice.
 
so sequential read/writes would not go up since it can't read/write the same sectors at the same time.. but I guess random speeds would be double, so good for servers etc?
 
I would expect that would not be an operating mode for the same reason when a head fails on a 3 platter drive you don't 5 / 6 ths of a drive. With that said the firmware could be written to support this. They just don't do it in practice.

It's not practical because head crashes and many other "partial" failures result in the inside of the drive being contaminated with debris. Even microscopic particles can wreck havoc on a drive.
 
God I feel old, since I remember these (and still have that's working!)
TrueX Multi-read tech

72x_box.jpg
 
from the animation it doesn't seem like its actually doing RAID0, but some parallel accses.
 
from the animation it doesn't seem like its actually doing RAID0, but some parallel accses.

The heads are split into two sets across two sets of platters, so it could be RAID0 or RAID1, however they want to configure it.

It's kinda like taking two sets of HDD internals and stacking them.
 
Why cant they use each heads output in parallel instead of reading from only one at a time?
I've looked for reasons why this isnt done and havent found any yet.
Then DTR for all use would increase, it would perform like 2 or more drives in RAID but without having to sync platters so comparative seek times would reduce slightly as well.

All this drive does is act like 2 discrete drives.
Data transfer is only doubled for 2 separate file read/writes, no better than using 2 drives.
If you use this in RAID, you need double redundancy for a single head failure because 2 effective drives will need replacing.
I'm not seeing the advantage, especially for data centre use with Seagates legendary reliability.
 
Back
Top