I agree that we are in quite close agreement, perhaps also because we are discussing something really esoteric 😉 And that is also the crux of my position -- I actually don't do disk spin-down for my primary system, as time-to-data is more important (and that system sees so much use it would come close to your one-minute scenario). So, I deem other things much more important than power savings or disk fatigue in one way or the other.
On the other hand, in the NAS systems I'm directly responsible for (ten of different makes and models), I DO do disk spin-down, because most of them are used as back-up devices active only a few hours per day, where it makes little sense to NOT use power savings. And that's why spin-down is mucho importante in any NAS system.
So yeah, it's totally up to the usage patterns for specific systems.
But the last thing I want to point out -- a fact I completely missed -- is something in that EZRX spec sheet. Or, rather, something that's NOT there. You can see the "Load/unload cycles" figure (do also note that the WD Red/EFRX figure is double that), but I don't see a start/stop cycle count figure...
With regards to hardware RAID spin-down, I have never seen such a system or thought of configuring one, but I'm guessing you are thinking of larger systems with multi-layered approaches? "Near-line data" behind primary storage, the primary perhaps being an SSD one? I agree that this changes the game, at least for big data. For normal-sized businesses, well, maybe a SSD cache can be beneficial there, too. For my line of work, well, let's just say I'm a couple of hundred customers away from that so far 😉 I reckon the next storage solution I buy for any customer will be an all-SSD one.
...but I don't see a start/stop cycle count figure...
I noticed that also when I was gathering info on spin-down. I haven't seen any disk datasheet that has a start/stop cycle spec. Which I think is also interesting.
mangrove wrote:
With regards to hardware RAID spin-down, I have never seen such a system or thought of configuring one, but I'm guessing you are thinking of larger systems with multi-layered approaches?
Large scale for sure. Some were multi-layered.
It is also possible to turn off some drives on data that is read-only - large video repositories for instance. The idea there is that if you have a RAID-6 array, you can "rest" 2 drives at a time in a sequence. With RAID-10, you can "rest" half the drives at any given time. You have to reconstruct the data from the parity blocks if you are using RAID-5 or RAID-6, but with suitable hardware that might not be an issue. In that framework, you might have a partial spindown threshold for disk reads, and a full spindown threshold that is longer for reads/writes.
Some approaches were more fine-grained - a bit like what btrfs is trying to do but on steroids - integrating the redundancy into the file system, instead of underneath it - and allowing an essentially unlimited amount of parity blocks. They were more academic, probably not practical at this point.
As far as I know the 6.1.9 is not final yet. So don't expect 6.2.0 this month (otherwise I don't see the point of the 6.1.9), probably not until christmas as they need additional testing for major versions. I am no Netgear but I suppose this is their roadmap for the best scenario. So either they give us a christmas gift or it may take some more time, I don't think we would have a stable release before that and probably not even an RC.
I think that Netgear has a precise roadmap but just don't want to disclose it, just in case they need more time. People tend to make mistakes under pressure, this is not something we would want for a NAS I guess. I am getting impatient too (don't need spin down but I lack other things that may or may not be implemented) and clearly a release date would help me, but I don't think they will drop a date before everything is already done and getting packaged. So we may have a "there will be a surprise for you tomorrow" but I don't think we will get any more information. I would clearly like to have a date but I don't see it happening.
I just want to add that I am really disappointed by the lack of this feature. I bought ReadyNas 102 to replace one bay Zyxel NAS that is 4 years old now. I didn't even check if readyNas has this feature as this is one of the most obvious features that every other serious NAS has. I'm very disappointed at the moment. I packed the gear back to the box now and I still use the Zyxel. If NetGear don't provide this feature up to a month/2 months time I will just throw it away and I'll make sure to mention NetGear in every discussion I have about NASes. Users have been asking for this feature for a long time now. I don't understand why this has not been implemented yet!? But it's either they just don't care or they are not smart enough to implement this feature. And both explanations really worry me.
I just thought I'd add another request for this feature as this situation is just unacceptable.
I just want to add that I am really disappointed by the lack of this feature.... I'm very disappointed at the moment.... I don't understand why this has not been implemented yet!? ...I just thought I'd add another request for this feature as this situation is just unacceptable.
I agree with all of MarioW's points. Spin down is a basic feature for a NAS.
I understand the philosophy of keeping the drives running (I do this on most of my commercial systems) but for residential boxes, there should be an option to spin them down.
But it's either they just don't care or they are not smart enough to implement this feature.
I would rather guess that it's because OS6 does lots of writes to its sqlite database, or some unforeseen consequence of BTRFS on MD. Anyway it's more of a technical reason I'm sure, but they aren't telling.
And the fact that they aren't willing to communicate technical reasons worries me.
We agree there should be an option to spin the drives down, that's what our customers have been telling us and it is already implemented in the upcoming 6.2.0 release. I think it might be the only feature we've confirmed is coming for 6.2.0.
I have no inside scoop on this, but it seems to me that the hard part of making spin down work is preventing the disk from immediately spinning up again. That requires tracking down everything that writes periodically to the OS partition, and making sure those writes are either timed to prevent frequent spin ups or alternatively suppressed altogether if the disks are spun down. From a programming point of view, its not rocket-science type stuff, but it would be time consuming.
If I'm right, it also involves a lot of small changes in a lot of different subsystems. And once you've committed them to a particular release's code branch, it would be a lot of work to change that decision, since those subsystems are also being modified to support other new features.
Anyway, its been clear for quite a while that Netgear has gotten the spin down message. I don't think anything we say at this point is likely to speed anything up - hopefully we'll see a 6.2 beta soon.
Running some things on a NAS can prevent spin down from working (this is the case on any OS that has tise feature) so there will be troubleshooting needed in some cases and decisions as to whether you really need an app or service or not more than spin down. But once you have it working it should work really well.
I agree that estimating a release date for 6.2.0 could generate more pressure, but waiting too long will make us think Readynas development team is not serious (or we could have very high expectations).
How can such an important update be released in another 3-4 months? This is ridiculous.
Based on past history, they will post a beta when they think it is ready for public use, and they will not pre-announce the date.
If I were in their shoes I'd do the same. They certainly care what we all think. But shipping something that doesn't meet the quality bar won't help them, and pre-announcing dates and features can create revenue recognition issues.
Important ? It could be. Critical ? I don't think so, you already have shutdown and wake on LAN. It consumes more power, I agree, but it is not impairing the NAS usability. I understood that you miss this feature and for good reasons but this is not a absolute necessity to run the NAS. The numbers I gave were my personal expectations, Netgear can prove me wrong and have the 6.1.9 final by monday and 6.2.0 by friday. I just expect them to take a few more months on it. Expecting a 6.2.0 before they even release the 6.1.9 is, I think, unrealistic. Usually there is some time between releases.
As for communication, there is a general consensus that states that : - if you did something wrong, you speak up rapidly to be sure that your story is heard first. It's always easier to defend a castle than an open field (crisis management) - If you want to be remarked you speak up targeting the right audience at the right moment with the right message (marketing). - if you are not in the two situations above, you just shut up. If you speak, what you say could be used against you and presents no real advantage (not to mention repercussions on stock value).
So not communicating may not be a sign of anything, this is just a policy that is there to ensure that everything goes as smoothly as possible. I agree that this policy is sometimes stupid but applying it blindly is the best way for things not to backfire on you, that's why most contracts come with a confidentiality agreement. Telling us things will create more questions (which will slow developers), pressure (which could results in critical bugs) or expectations (which could disappoint us in the end). Actually we know that 6.2.0 will feature spindown. We wouldn't actually have this debate is they said "not supported, act accordingly" and stopped there, there would have been a few rage posts on the net that potential buyers wouldn't even take into account and the thing would have been forgotten. I'm not sure telling that it will released on 6.2.0 actually affected either their sales nor their reputation, but the simple fact of saying this already had repercussions : it created expectations, which also means pressure. Every information they give will create some more. So yes, it is not fun for us (and I believe that MDGM really wants to tell us more but can't due to his contract, so no fun for him either), but this is the best course for Netgear and I think they'll stick with it no matter what we say or do (they care but they can't take chances with their reputation, things can get worse really quickly with a single wrong message). Not to mention that telling us to expect only spin down will make the discovery of the new features more enjoyable and in the end create a better customer satisfaction (I know that 2 months after the release, the wait will be forgotten, but I'll remember the added feature(s) for a while). And I agree with StephenB, I would do exactly the same.
Well I was not suggesting releasing an unfinished software nor to communicate a release date.
What I think is that making people wait too long for an update does mean that the product team has not planned very well, or wants to add too many features at once.
We are talking about an enterprise with more than 1K employees... Developping/testing spin down does not take 6 months. At least I hope so.
Depends how deep the OS redesign goes, on how many features are added, how many people are actively working on it and on how many other projects they work, was is the version roadmap (if the next major upgrade is next year, you better have a solid feature list) , the strategical implications (announcing it before christmas could boost the sales for example...), what is the level of testing required... I don't know how they work, I don't know their philosophy. I just know that Apple tend to beta test from june to september/october for OS X, and I think they test internally before that, so yes they take more than 6months and I'm sure they don't lack manpower. So depending on what is to be done, 6 months can be a reasonable delay. Of course of we just get a service pack that's another story.
I could find them excuses, the fact is I don't know the truth more than you do. Maybe they are slackers, unskilled and have a terrible management... I don't think so but maybe... I feel they are competent and hardworking. Look at MDGM, he spends god knows how many hours on the forum, Skywalker usually tells us the under the hood improvements and actually doesn't seem to have trouble doing so. So I trust they are doing good things, now trust can sometimes be misplaced. Time will tell.
They just announced 6.2.1. Apparently, they skipped over 6.2.0 just to avoid the spin down issue. </sarcasm>
Seriously, though, drive spin down is a relatively minor "feature" for network attached storage. I've been running seagate ES drives non-stop for over 5 years without spin down enabled (as it caused problems when instant access was needed.) The absolutely tiny amount of power savings a person would see from drive spindown over the course of 5 years would pale in comparison to the amount of power that the NAS would chew having to rebuild the array just once when a drive took too long to spin up.
I think they have bigger fish to fry for 6.2.x, and getting those issues resolved could very well be impacting 6.2.x. I could also see many potential problems with spun down drives in a RAID array, so be happy they are holding back releasing that before it's REALLY ready. It would suck if they released it, you installed it, and then you had drives dropping out of the array when they were spun down.
As for how long it's taking (and the lack of an actual release date)... be happy that it's been confirmed as coming in 6.2.0. I've been waiting 5 years for "snapshot resizing" to be re-enabled in the previous product (radiator 4.x.) I have a manual telling me how to resize the snapshot reservation, and release notes going back many years promising that it'll be re-enabled in a future version. (I'm not complaining about the snapshot resizing in this particular thread - I'm just using it as a point of contrast.)
Disk spin-down is not something I've used much as I too value instantaneous access to my data.
However many users do value it, especially if a NAS is used for backups and the disks only need to be accessed for a short amount of time each day, and they'd rather not use a power off/on schedule.
Disk spin-down is an important new feature in 6.2.0.
The 6 is there because of the revolutionary change OS6 is.
The 2 indicates it the second major evolutionary update to OS6.
The 0 indicates the minor update number. It is 0 as it is the first 6.2 release.
With major evolutionary updates there is a greater focus on new features, with minor updates the focus is mainly on bug and security fixes.
As disk spin-down is core functionality in any product involving mechanical drives, it's rather an indication of how unfinished OS6 was when released. I said back then that a real "1.0" version would arrive a year after initial OS6 release, but I was wrong; it will be two years late. Quite remarkable.
As disk spin-down is core functionality in any product involving mechanical drives, it's rather an indication of how unfinished OS6 was when released. I said back then that a real "1.0" version would arrive a year after initial OS6 release, but I was wrong; it will be two years late. Quite remarkable.
Actually, I'm pretty sure my xbox, my TV, my time capsule, my ESX server (not totally sure, I'd have to check advanced configuration) haven't got that feature, so that's not "core" because given 6 devices with mechanical hard drives (2 xbox, one time capsule, 1 ESX, 1TV, 1NAS) none has it... The only one that has this feature is my Mac, which ironically has a SSD built-in... Of course Windows also has it and it is possible to implement on linux (I'm pretty sure stock Debian hasn't it enabled like most linux, though I believe packages are installed, not sure). I've never seen this feature implemented in enterprise and most people I know disable standby and disk spin down right after installing their system. They rely on startup/shutdown mechanism and don't want disk/machine to be put in standby without warning. The only use I can see of this thing is when you leave the house for some days and want remote access to the NAS. Wake on Wan is hard to implement and startup/shutdown will not guess when you will want to connect. But for common backups startup/shutdown should work well and actually it would save you more power I think.
As for the release date, once again it has to do with many factors. If this was that simple, you would have enabled SSH and do it yourself I guess... I used to be Linux certified (those things tend to have a 2 years validity nowadays...) and I didn't do it myself. This would actually be pretty simple to do it. That's one command away. And a few array resync, frequent unneeded spin ups later you end up with corrupted data. You must understand that Netgear tries to give you a good software that keeps your data safe. I value my data safety over anything else. Now, I don't know if you smell what Netgear is cooking (the fish to fry thing made me hungry 😉 ) but I agree that 6.2 will probably feature more than spin down. The best thing we have to do now is wait and see.
Xbox and time capsule have it (of course), but you have possibly disabled it. No idea what your TV is doing but it probably have power saving options too. A server for virtualization should, in my opinion, never shut down but that's a special case for sure.
Just to confirm that this was an advertised feature of the ReadyNAS before I purchased it and the article below only get changed AFTER I raised the question as to why it was not available. http://www.readynas.com/kb/faq/pre-sales/does_the_readynas_support_sleep_mode
You'd think that because of this oversight (I'm not going to say false advertising) that Netgear would put more emphasis on correcting this...
