Western Digital Has Been Developing A New Linux File-System: Zonefs

[starshipeleven]

Which is the "cringy" naming?

You called consumer hardware "desktop peecee hardware".

And no, it doesn't use SMR. We're talking about backup solutions here. *Archiving* solutions (aka WORM data) use SMR, but not backup where the media is regularly updated and re-written.

SMR is fine for incremental backups, it is also fine for full-rewrite backups too (similar to tape drives), the only thing it sucks at is dealing with large amounts of random read/writes. It's basically a better tape drive.

Ding ding ding. It isn't SMR at all because backup solutions don't use SMR.

this stark distinction is only in your head. Anywhere you were using tape you can just as well use SMR and it's going to be better.

[oiaohm]

Sorry, you don't understand how product launches work. Find me a part number for an 18 or 20 TB drive. You cannot, because they don't exist. The industry terminology "shipping to limited customers" does not mean what you think it means.

Sorry I could because I have a 20TB WD sitting on my desk. I work at times with archives I do have that part number but until open list that is under NDA agreement.

Not sure how this helps your argument. So PMR drives are *larger* than SMR drives today? If so, the capacity gap isn't zero, it's a negative number, LMAO.

If you were using the older drivers the number should have been negative capacity gap. There was a WD 15TB but those never went into mass production and do have a product code released.

That's funny. I've been employed in the enterprise data storage market since 1993. Were you even born yet? Serious question.

I have been linked working around do data recovery for archives longer than that. The section of enterprise that gets early access to a lot stuff.

No, that's not how it works. I've tested SMR drives in RAID configurations in the lab at work. There is no failure or error condition. There is a performance penalty, yes, but that's not much different from SSD's and TRIM. The fact is, there are no software or hardware changes *required*.

https://documents.westerndigital.com/content/dam/doc-library/en_us/assets/public/western-digital/collateral/white-paper/white-paper-shingled-magnetic-recording-helioseal-technology.pdf

No you have done a idiot test. You WD harddrives SMR for example are documented above to come in 3 forms/modes as default on power on.

1) Drive Managed SMR In this mode behave like legacy drives can only behave like a legacy drive yes can be horrible slow this is purely the controller..
2) Host Aware SMR Fun mode does support Drive Managed SMR and can operate receiving Host Managed SMR commands.
3) Host Managed SMR this is the one that ruins you day this has no legacy support.

Due to the nature of Host Managed SMR design, it is not a plug-and-play implementation with legacy systems.

This above is a direct quote from the HP white paper.

Something HP white paper does not tell is that you can set mode of a HP SMR drive to any one of those 3.

So I give you a Ultrastar DC HC620 there is no way from what written on the outside of it to know what mode it in. If it in the Host Managed SMR mode all the issues I described where your raid system goes to hell is exactly what happens.

You saw no error condition in legacy hardware because the SMR drives you tried had to be in "Drive Managed SMR" or "Host Aware SMR" mode. Its a completely different game when you by mistake or intentionally for testing throw a "Host Managed SMR" mode drive into a legacy system.

Our policy here is simple we don't take a risk. A modern Raid and OS that can handle "Host Managed SMR" will handle the other 2 modes the drive could be in just fine. But your legacy systems being hit with a drive in Host Managed SMR mode all hell can break loss. So SMR only goes into modern hardware and software setups here why take the risk.

[curfew]

TL;DR:

Your TLDR is wrong and does not answer the question. There is NO BENEFIT per se. Zoned storage is a requirement for having larger capacity hard disk drivers. Zones are a way to allow this but they come with additional restrictions, not benefits. Essentially zones and ZoneFS are a tool to work around a technical issue by minimizing the performance cost it involves while reducing complexity of implementation for other developers.

[ssokolow]

Your TLDR is wrong and does not answer the question. There is NO BENEFIT per se. Zoned storage is a requirement for having larger capacity hard disk drivers. Zones are a way to allow this but they come with additional restrictions, not benefits. Essentially zones and ZoneFS are a tool to work around a technical issue by minimizing the performance cost it involves while reducing complexity of implementation for other developers.

Depending on how you look at it, the benefit is one of the following:

1. Enabling higher drive capacities (If you don't accept the performance loss of drive-managed SMR.)
2. Enabling higher performance at capacities not reachable using technologies capable of truly random-access writes. (If you accept the performance loss of drive-managed SMR as being a valid alternative.)

[oiaohm]

Your TLDR is wrong and does not answer the question. There is NO BENEFIT per se. Zoned storage is a requirement for having larger capacity hard disk drivers. Zones are a way to allow this but they come with additional restrictions, not benefits. Essentially zones and ZoneFS are a tool to work around a technical issue by minimizing the performance cost it involves while reducing complexity of implementation for other developers.

This is not quite true there is a benefit but its not really large enough to write home about in advertising material.

SMR gets rid of two guard spaces. The guard space between your 4kb blocks around the track and the guard space between tracks. Read and Write speed of SMR is slightly higher than conventional due to the more bits per rotation due to guard space between sectors not being there either but this is less than a 0.1 percent uplift in performance. If that covers the modify cost of SMR that is another question unless the workload suits SMR the answer will be no. Does that 0.1 extra read/write performance matter to your workload some really rare cases it might be a important factor.

Zoned equals bigger blocks. Yes the changed of hard-drives from 512 to 4kb sectors was also about increasing storage density and read/write speed. SMR bigger blocks basically max out track density as well.

Really I would love to see a drive that was a mix of SMR, CMR and a new form a max density tracks where you have the option of using complete track around the disc without any guard zones between sectors yet keep the guard zone between tracks.

Some items suite SMR storage some suite CMR and some would be better with smaller zones.

SMR is basically gives you the fastest read/write speed with the type of harddrive head you are using at the price making modification costly but the performance gain being less than 0.1 percent more than CMR means it basically bugger all and not a factor in most cases with the modification cost being a more critical factor to consider.

Read and write speed of harddrives can still increase by increasing density. HAMR and MAMR could see our read/write performances of a harddrive double/tripple without increasing heads with density alone.. This makes a 0.1 performance increase on the very small side.

[aht0]

With big tech giants being so close to various governments there is a 0% chance I would use a filesystem developed by them.

Don 't worry, it's safe. Better places to hide spy software in, like disc controller firmware. Bare handful of global manufacturers, OS independent, hard to find..