Quick answer

Exos idle at 2.8 bels of sound power; Ultrastar and Toshiba MG idle at 2.0. Under seek all three are 3.2. So the gap is idle noise, and the thing you actually hear is the chassis panel resonating. Isolate the enclosure before you shop for a quieter drive.

On this page

Exos are loud. Not louder-than-a-laptop loud — loud enough that where the array lives is a decision you make before you buy the drives, not after they arrive.

I bought mine on price and capacity. The noise is the part I regret.

But the drive is not what you hear. The chassis panel resonating with the drive is what you hear, and that changes what you should do about it.

Some retailer links below are affiliate links; if you buy through them TechFuel HQ may earn a small commission at no extra cost to you. As an Amazon Associate I earn from qualifying purchases. Commissions never influence what gets recommended — see our disclosure.

The three brands are identical under seek

All three manufacturers publish sound power in their own product manuals, each measured to ISO 7779:

24TB driveIdle (typ)Seek (typ)
Seagate Exos X242.8 bels3.2 bels
WD Ultrastar DC HC5802.0 bels3.2 bels
Toshiba MG112.0 bels3.2 bels

Figures from the Seagate Exos X24 SATA product manual (Rev. C, Table 9), the Ultrastar DC HC580 SATA specification (Rev. 2.1, Table 21) and the Toshiba MG11 product overview, which prints sound power as 20 dB idle and 32 dB seek — the same 2.0 and 3.2 bels in different units.

Read that table again, because it is not the shape people expect. Under seek there is no difference at all. Every one of these is a 7200 RPM enterprise drive, and every one of them is 3.2 bels when the heads are moving. Seagate and Western Digital both publish the same 3.4 bel maximum on top of it.

The whole gap is idle. Eight decibels of sound power, on a machine that spends the overwhelming majority of its life idle. That is why HGST’s Ultrastar line — Western Digital’s since the acquisition — and Toshiba’s MG carry the quiet reputation, and it is a narrower, more specific advantage than the reputation suggests.

You pay for it. In my August 17 pricing the Toshiba MG at 24TB ran $43.12 per terabyte against $39.50 for the Exos — about 9% more, which is $87 a drive and $522 across six bays.

I can’t tell you idle draw per drive from memory and I’m not going to guess — it’s the number everyone asks for on deal threads and nobody answers.

The panel is the noise

Vibration is the part people miss when they shop this by spec sheet.

Seagate rates the Exos X24 to hold greater than 90% throughput on sequential and random writes while being shaken at 12.5 rad/s² of rotary vibration between 20 and 1500 Hz, and to run without unrecoverable errors under 0.70 Grms of linear vibration from 5 to 500 Hz. The drive is engineered to keep working while it shakes.

Nothing in that specification says your enclosure stays quiet while it happens. The energy goes somewhere. In a thin steel side panel it goes into a large flat surface with a resonant frequency, and a 7200 RPM spindle is very good at finding it. That is the sound you are actually annoyed by — not the drive, the sheet metal the drive is bolted to.

Which means isolation fixes more than shopping for a quieter model does. Grommets or rubber-isolated sleds in the bays, a heavier or braced enclosure, and drives that are not rigidly coupled to one big panel. The 0.8 bel difference between an Exos and an Ultrastar is real, and it is smaller than the difference between a bare metal cage and an isolated one.

The manual is specific about the mounting itself, and it is worth following: 6-32 UNC screws only, inserted no more than 3.56 mm into the side or bottom holes, at a maximum torque of 6 in-lb. Cranking a drive down hard into a panel does not stop the noise. It couples the two more tightly.

The number on the box is the outer edge of the platter

Sequential throughput comes in under spec. Not by a lot, and not in a way that matters for a NAS, but the number printed on the listing is not the number you get.

That is by design rather than by disappointment. Seagate’s 285 MB/s is the maximum sustained rate at the outer diameter — the manual prints it as 272 MiB/s OD. Outer tracks are physically longer, so more sectors pass under the head per rotation. Fill the pool and you are working further in, at a lower rate, with nothing wrong anywhere.

If you are sizing a rebuild window, that is the direction the error runs: slower than the floor, never faster.

How to buy recertified

Recertified is a normal way to buy drives now, not a corner cut. Seagate runs the program itself and describes the process as sanitize, test against factory performance standards, then certify.

Test every recert on arrival, before it joins anything. A full SMART extended self-test:

smartctl -t long /dev/sdX      # start it
smartctl -c /dev/sdX           # the drive reports its own expected polling time
smartctl -a /dev/sdX           # read the result when it finishes

The smartctl documentation puts the extended test at “tens of minutes to several hours.” At 24TB you are at the long end of that. Start it before bed. If a drive is going to fail early, it fails there — outside your array, while you still have a return window.

The warranty is the real tradeoff, and it is not one number

This is where recertified actually costs you, and the spread is wider than most people expect. Same drive family, four different answers, all live today:

Where you buy itStated warranty
New Exos X24, from Seagate5-year limited warranty
Seagate’s own recertified program6 months
Recertified Exos X24 at Newegg (ST24000NM000C)90 days
Recertified Exos X24 at ServerPartDeals (ST24000NM000H)3 years

Sources: Seagate’s Exos X24 support page and recertified drives page, the Newegg recertified listing, and the ServerPartDeals listing.

The mechanism behind the spread is the thing to understand. ServerPartDeals defines a manufacturer-recertified drive as one “tested, recertified, and approved by the original manufacturer… without a manufacturer warranty.” The original coverage does not travel with the drive. What you get instead is whatever the seller chose to promise, and on the same 24TB Exos that runs from 90 days to three years.

So the rule is not “recert has a shorter warranty.” It is read the listing you are actually buying from, because the number is set there and nowhere else. Three years of seller coverage on a recert is a different purchase from ninety days at a similar price.

“It already broke once” misses how parity works

The standard objection to recertified drives is that they are drives that already failed. It misunderstands what the array is for.

A redundant array does not assume its members are healthy. It assumes they will die and holds enough parity to survive it while a replacement copies. Recert plus redundancy plus backups is the whole argument — and if you are missing the third one, the first two do not save you. Parity survives a drive dying. It does not survive a deleted directory, or the enclosure’s power supply taking the shelf with it.

That is the actual condition on buying recertified. Not “is this drive suspicious,” but “do I have all three.” If you only have two, buying new drives leaves you in the same place, just more expensively.

Check for SMR anyway

SMR is a consumer-drive problem. Western Digital’s own SMR disclosure is titled “WD Internal Client HDD Recording Technology” and lists WD Red 3.5-inch at 2/3/4/6TB plus some WD Blue and WD Black models — client drives, none above 6TB — with the note that “all other capacity points use CMR recording technology.”

The three enterprise families here are all CMR, confirmed in their own documentation: Seagate describes the 24TB Exos X24 as “conventional magnetic recording (CMR)-based,” Western Digital’s HC580 manual labels its 22/24TB power table the “CMR Model,” and Toshiba’s MG11 overview opens by calling it a “conventional magnetic recording (CMR) HDD.”

Check the model number anyway. It costs thirty seconds, and SMR discovered during a resilver is the worst afternoon you will have.

Buy the Exos, and decide where it goes before you order

For a homelab array, buy the Seagate Exos. It costs less per terabyte than the Toshiba, it is identical to both of the others under seek, and the idle penalty is a room problem you solve with rubber and a closed door rather than with $522 of extra drives. Buy recertified if the listing’s warranty is one you would accept in writing, and run the long test the day it lands.

Search NeweggLink checked 2026-09-07Search AmazonLink checked 2026-09-07

Prices move. Check before you buy six.

Buy the Ultrastar HC580 or the Toshiba MG11 instead if the array has to sit in a room you occupy quietly — a home office you take calls in, a studio apartment, a living room. On the Toshiba you are buying 0.8 bels at idle for about 9% more per terabyte, and in that room it is worth it.

Do not put any of these in a room you sleep in. Not the Exos, not the quiet ones. Move the array or accept that you will hear it, because 2.0 bels at idle is still an enterprise drive four feet from your head, and it goes to 3.2 the moment a scrub starts at 2am.

Who should not buy these

  • Anyone putting the array in a bedroom. The quieter models narrow the gap at idle and close nothing under load.
  • Anyone trusting a single recert with unbacked data. Recert works because parity and a backup sit behind it. Alone it is a used drive.
  • Anyone who wants one quiet external drive. These are 7200 RPM enterprise disks built for a chassis full of their siblings. Buy a consumer drive for that job.

The Seagate drive comparison covers where Exos sits against IronWolf and IronWolf Pro, and the capacity calculator runs the parity maths for whatever bay count you land on.

Frequently asked questions

Are Seagate Exos drives too loud for a home NAS?
They are loud enough that the room matters. Seagate’s Exos X24 manual publishes 2.8 bels of sound power at idle and 3.2 bels under seek, measured to ISO 7779. Western Digital’s Ultrastar DC HC580 and Toshiba’s MG11 both publish 2.0 bels idle and 3.2 seek. So all three are the same under seek, and the Exos is 0.8 bels — eight decibels of sound power — noisier when nothing is happening, which is most of the time. In an office or a closet that is fine. In a bedroom it is not, and no amount of drive shopping fixes it.
What is the quietest 24TB enterprise hard drive?
On published figures, the Western Digital Ultrastar DC HC580 and the Toshiba MG11 tie at 2.0 bels of sound power at idle, against 2.8 for the Seagate Exos X24. Under seek all three publish 3.2 bels typical, so the advantage disappears the moment the array is working. You pay for the quieter idle: in my August 17 Newegg pull the Toshiba MG 24TB ran $43.12 per terabyte against $39.50 for the Exos, about 9% more.
Are recertified Exos drives safe to buy?
Yes, if you run redundancy and keep a real backup. Recertified is a normal way to buy drives now rather than a corner cut. The actual tradeoff is warranty, and it is not one number — the same recertified Exos X24 carries 90 days at Newegg and three years at ServerPartDeals, against a five-year limited warranty on a new one from Seagate. Read the specific listing, because the manufacturer’s warranty does not transfer and what replaces it is whatever the seller decided.
How do I test a recertified hard drive when it arrives?
Run a full SMART extended self-test before the drive joins anything: smartctl -t long /dev/sdX, then smartctl -a to read the result. The smartctl documentation describes the extended test as taking tens of minutes to several hours; at 24TB you are at the long end, and the drive itself reports its expected polling time under smartctl -c. Start it, leave it, and check in the morning. If a recert is going to fail early, this is where it fails, and it fails outside your array instead of inside it.

Evidence ledger

Last updated
Methodology
This homelab guide was written and edited by Lowell K. Wood IV in St. Louis County, MO. Specs and prices verified against vendor and project documentation current on the date above. Full editorial standard: methodology.
Update log
  • 2026-09-07 — Last reviewed and updated.
Corrections
Spotted an error or a stale number? Email hello@techfuelhq.com. Confirmed corrections are added to the update log above.

About the author

Written by Lowell K. Wood IV, who builds and runs TechFuelHQ from St. Louis, Missouri.