Alekz said Otherwise you will be bound to one or two companies that write apps for one or two systems, and any can disappear or abandon the project.Or speaking in the "MQA terms" - "vendor lock-in". (In the general IT industry you can think of Oracle and Adobe).
magister saidI fully agree with the Playlist functionality - on the DMP it is intermittantly dysfunctional. Discs are recognized on the DMP but not the Playlist making them uneditable while others are recognized by neither - and not obscure discs - Paul Simon. Unless the Playlist functionality is improved - Octave is going to be a struggle.Paul McGowan said But the magic I seek here is for the user to fall in love from the first second. Instantly and without a learning curve. To never upset; like ignoring user installed tags; like not finding proper meta data; like doing lame things. . . . It will just work and work exactly the way you want.I agree with Elk that such a product would be amazing. I also think you may be underestimating the work required in the area of software and metadata, which has not been PSA's strong suit. In fact, I'm not even sure it's possible, at least without devoting the kind of effort that Roon does to collecting and analyzing metadata. Does PSA have the resources for that?Hereās a serious suggestion: devote some resources to getting the Playlist function working better. Others and I have posted our frustrations about the Playlist; I will be happy to send my suggestions again if needed. Doing so would make your existing/future PWT and DMP owners happy. It would also give you more experience with manipulating metadata, experience that will be useful when developing the software for Octave ā which will be orders of magnitude more complex than the Playlist requires.
No UPnP system is any good.None has the visual beauty of Roon, true. In my house one can use either BubbleUPnP on an Android or Linn Kazoo on an iPad to run a UPnP setup. It does require one person (me) with a modicum of computer skills to set up, but once up and running anyone can use it easily; it is reliable; it is inexpensive (assuming one already owns a tablet); it runs on low-power NAS or almost any computer (unlike Roon which requires a SSD and multiple cores if one has anything beyond a small library); and looks OK when in use, if not fancy. I call that reasonably good. Not ideal, not perfect, but reasonably good.
Also - as Paul focuses on the 3 big ones
- A great user interface
- Output impervious to the vagaries of the computer and hardware inside
- Simple to use, operate and implement
Time Machine is merely an incremental backup utility. software for which has been around literally for decades. Why is this exciting/a must have?
And a 2TB SSD remains expensive, where a quality 2TB hard drive is well under $100. SSD speed is not required for a backup drive.
All a music server needs is the ability to directly attach a USB drive, and an option for installing an internal SSD for those who want one A utility to perform backups to an extra attached USB drive is a plus. It should also be able to communicate with a NAS for those who want to go this route.
Michael saidAs a 4th Iād also bring up the storage medium - hard drive versus SSD. I do not have a server but many of the early ones had mirrored drives and Iād imagine a reliable backup is mandatory given the time and investment to rip discs. Starting off by building the unit around a 2TB SSD with a similarly sized backup a la Appleās Time Machine would go a long way to making the Octave SoA. And a 2TB internal SSD is not that expensive - an HD would not be the way to save monet.
Do not mirror (aka RAID-1) average consumer-grade SSDs. Youāll only wear out the cells in two SSDs simultaneously and a lot more quickly than without the mirroring. All SSD manufacturers vehemently recommend against using any kind of RAID, but most of all against RAID-1. (FusionIO and other purpose-built appliances being exceptions but these are not normal SSD hardware.) I know this because of my job where we work directly with the vendors and their engineers, not just my own personal experience. The grudging exception here is RAID-5 or -6, if you absolutely cannot avoid it.
Run periodic differential backups, preferably at the block level when possible. There are myriad ways to do this which could be built-in to the system (and probably should be, if itās there at all). Windows and Linux both have had internal methods for doing this for quite a long time.
Incoming tangentā¦
There are deeply technical concerns about monitoring the health of the internal storage in a āset it and forget itā system like this, specifically in the area of making the user aware that a problem is looming and what simple thing they can do to alleviate it before they lose any/all of their data, at least some of which we must presume exists nowhere else unless we (the system itself) ensures thereās at least one other pristine copy at any given time (RAID-1 will not do this).
Neither HDDs nor SSDs have infinite lifespan and no filesystem that might practically be used here is immune to silent bitrot/corruption (although ZFS is pretty close). The standard SMART monitoring is utter crap and canāt be relied upon to alert before an issue is already a problem; however, the SMART hardware-health data can be polled from the HDD/SSD independently ā rather than relying on the OS or BIOS to do it for you ā but even then itās only one data source and more are needed to make an accurate determination of the state of the storage device. File-level and overall filesystem health has to regularly be determined separately; SMART wonāt do that at all.
Rather than [only] USB ā and this would have to be USB-3 to actually be reasonably future-resilient and useful ā Iād prefer to see an additional (for backups or additional data storage; userās choice) 2.5-inch SATA bay, hidden until needed of course, into which one could drop an HDD or SSD. The system confirms the size of the disk, asks the user if they want to (re)initialize the disk or try to use it as-is (could be for rotating offline backup disks, or for importing data in the fastest way possible). Having that external USB device sitting around being ugly, in one way or another, and RFI/EMI-noisier is undesirable. Just brainstorming here, not describing a full technical solution in any way. 
A lot of that is probably caffeine-fueled so take it or leave it as you wish.
-Jim
Elk said Time Machine is merely an incremental backup utility. software for which has been around literally for decades. Why is this exciting/a must have?And a 2TB SSD remains expensive, where a quality 2TB hard drive is well under $100. SSD speed is not required for a backup drive.
All a music server needs is the ability to directly attach a USB drive, and an option for installing an internal SSD for those who want one A utility to perform backups to an extra attached USB drive is a plus. It should also be able to communicate with a NAS for those who want to go this route.
Pretty sure Time Machine was meant as an ease-of-use example, not a literal suggestion. But as far as that goes, yes, level 1+ backups have been around for ages and none of them were or are anywhere near as easy to use/manage as TM, so pointing out ābeen around foreverā serves no practical point.
Viable/valid data backups are always useful, no exceptions, especially if/when this appliance becomes the central point of origin/storage for some/most/all of a personās digital versions/copies of music. (That said, itās not the backup thatās most important; itās being able to restore the data intact thatās most important and that means regular consistency/validity checks of both the primary and backup copies of data.)
As for SSD expense, this seems an odd argument to make given the general expensiveness of PSAās hardware products. Iād class this as an item for the buyer of such an appliance to concern themselves with, wouldnāt you? PS Audio should certainly provide the option for either HDD or SSD, either at purchase or in-home user-performed upgrade/downgrade ā it should matter not at all to the internals of the system, other than perhaps the filesystem getting some special flags used at creation time to ensure TRIM is enabled (in the case of SSD). (Depending on the OS being used, there may be other SSD-specific kernel settings to be made but that can all be invisibly automated.)
The perceived need for the speed of an SSD is an awfully subjective thing and Iām surprised to see you apparently brushing it off as universally insignificant (youāre usually more aware of the potential needs/wants of others, Iāve observed). Given how much faster SSDs are than HDDs, and that it may be necessary at some point to restore from the backup on that SSD or HDD, it may be of large value to the user to have an all-SSD system. Restoring a TB or more at 300+MB/s is a whole lot more tolerable than what youāll get from any HDD. That and, of course, SSDs are silent and can emit less EMI/RFI than an HDD. As I said in another post, Iād prefer to see a user-accessible 2nd SATA bay for this purpose, rather than only an external USB-3 connector (how gauche when itās perfectly avoidable), but Iād be surprised to see that happen.
As a neutral observation, your reply in general seems oddly curt and dismissive compared to your usual thoughfulness and consideration. Not intended as a put-down, but more as āhope everythingās all rightā.
-Jim
Paul
What is the targeted price for this server, wonāt hold you to it just trying to get an idea of cost.
magicknow
Jim said . . . yes, level 1+ backups have been around for ages and none of them were or are anywhere near as easy to use/manage as TM . . .What does Time Machine do which is impressive? All incremental backup software I have seen/used is near brainless.
As for SSD expense, this seems an odd argument to make given the general expensiveness of PSA's hardware products. I'd class this as an item for the buyer of such an appliance to concern themselves with, wouldn't you?Does not point two make your point one irrelevant? :) And yes, as I initially noted, PSA should just provide the ability to integrate whatever one prefers.
Given how much faster SSDs are than HDDs, and that it may be necessary at some point to restore from the backup on that SSD or HDD, it may be of large value to the user to have an all-SSD system.A point, although I would not incur the additional expense of an all SSD system merely because it is possible I'll need to restore from backup. I run such things unobserved.
I can appreciate the idea of not being dependent on others but I think a proprietary development will be a ādauntingā task. I hope PS Audio will base the music server kernel on Linux in order for Roon Core (ALSA) to be installed as plan B in case it fails. It would also be nice to see a ādirect connectā solution to a storage device in case a SSD catridge is not built in. A networked store is probably catered for, but not uPnP if I understand correctly (FTP/SMB?).
Elk said All a music server needs is the ability to directly attach a USB drive,If Paul's goal for Octave is to make it as simple as possible to use, it needs internal storage and internal backup. For most people on this forum buying and plugging in a USB drive is not a big deal. But that's not true for everyone. USB drives are still slow, as I am reminded every time I backup my NAS. There are also people who object (or, perhaps more often, whose significant others object) to having lots of cables and extra stuff sitting around in the room. People don't do backups, even though they should; having an internal backup seems to be a very desirable thing -- almost a necessity -- for a 'set it and forget it' kind of unit. (Jim's point about how to monitor the health of a backup unit is certainly worth pondering.)
A utility to perform backups to an extra attached USB drive is a plus.Again, if the goal is simplicity and reliability, there should be a way (a pushbutton on the front or a very easily accessible on-screen command) to backup to an external drive. If it's easy to use, more people (no, not everyone) will take advantage of it. I'd call it a requirement for the kind of unit Paul is describing, unlike some servers that are aimed at those more computer-savvy.
If my understanding is correct, Octave is the name of the control app running on the iDevice?
I suppose this would be powerful enough to manage any storage commands residing on the music server or network device.
Yes, correct. The Octave Music Server system is based on Linux. At its core is MPD used as the player. The majority of work has gone into its user interface.
Many have asked me why we didnāt just go with Roon. Others have waved their fingers in my face and said āremember eLyric!ā like the Alamo, where disaster once nearly killed us. Still others have pointed out that software and UIs havenāt been our strong suit and they are all correct.
That notwithstanding, weāre marching forward and for a number of reasons that may not be crystal clear.
To why we donāt use Roon. We love Roon and the guys behind it too. But they are too expensive for manufacturers to purchase. Our long term plans include taking the Octave user experience downscale into lower and lower cost products PS Audio makes. Imagine someday a seriously under $1,000 music server from us. Hard to do if we have to pay Roon $500 for the firmware. And then thereās longevity. How long will Roon be around? Maybe forever, maybe not. But I am unwilling to risk turning our hardware products into bricks if they go away. Weāve been around 40+ years and plan on another 40 to go under the helm of my son Scott.
As to our expertise in software. We have new blood, weāre learning, and the fact we failed before only spurs me on to learn from our experiences and work harder at getting it right in the future. I failed to make good sounding gear for several years before starting PS Audio - but I learned from my mistakes and eventually won. Weāre making great progress, I love where the UI is going. Weāll be gathering a private group of beta testers for the UI itself sometime late in the first quarter.
The first product out the door will be priced similar to DirectStream, perhaps DSJ.
Jim
Appreciate your informed insights on the SSD/HD backup suggestion especially the point that SSDs are not infallible. Like many users of PSA products, Iād put myself in the computer naive camp. I have a Sony DAC recorder and I know despite all the redundancies on the tapes in protecting data it can self corrupt and leave, in essence, a blank tape. Iād hate to invest in an Octave-like device and face the possibility that it too was labile especially with the continuing idiosyncrasies of Playlist.
TM was indeed an example of theoretical ease of use - although after spending the past month talking with Apple engineers on a continuing issue with my wifeās iMac - not so sure.
Thanks
Mike
Jim saidDo not mirror (aka RAID-1) average consumer-grade SSDs. Youāll only wear out the cells in two SSDs simultaneously and a lot more quickly than without the mirroring. All SSD manufacturers vehemently recommend against using any kind of RAID, but most of all against RAID-1.
This is interesting, what you are saying. Do you have any links? From my understanding, RAID1 should not put any additional wear on disks, both should wear at the same rate as one disk of the same size, since the same information is being written to both disks. Many motherboards have two m.2 slots, what is perfect for mirroring.
As of SSD vs HDD. Since the music server is supposed to be placed in the āaudio roomā, there must be no rotating disks, and no fans. 1TB SSD are relatively cheap, for example, Crucial MX300 1,05TB disk costs 272euros in NL. Having super-fast disks is not necessary, even HDDs are fast enough. It would be nice to have 4 slots, so it would be possible to choose RAID5 if required, or create two mirrors, or use smaller and cheaper disks. But to cut the costs down, two slots is also not that bad. Actually itās a balance between more expensive backplane or more expensive disks. Lifespan of modern SSDs is big enough to withstand occasional CD ripping or FLAC uploads for many-many years. E.g. for the MX300 itās 200GB a day for 5 years!
Backup software can be as simple and effective as a UNIX rsync command (or grsync with a primitive GUI), no fancy stuff is needed.
As I said Iām very interested in the standards that are/will be chosen. For example, MPD/ALSA is a fantastic combination, that works perfectly in a real-time setup, and has a well-known and widely supported protocol. But what is planned for the server-DAC connectivity? USB, Ethernet, I2S, TOSLINK, etc, or all of them? From what I understand, I2S is the preferred way, right?
Thanks for the update Paul. I mentioned the eLyric experience only as an example of how hard this stuff is to do right. With all its faults eLyric was actually a great program and had some features I still miss. There are plenty of counter-examples of PSA successesāthe firmware that drives the DS and DS Jr and Bridge II readily come to mind. I fully expect you will be able to reach your goals here and produce a product that stands out in an increasingly crowded field. And even with all those alternative products I still see a need for a product that pulls it all together and can be used by non-technical folks.
To add to what Steve said: I was very pleased to read Paulās comments in post #51 ā sounds like theyāve got things going in the right direction and I admire Paulās can-do attitude. I agree that thereās āa need for a product that pulls it all together and can be used by non-technical folks.ā
I personally can handle some technical stuff, but am very interested in what the GUI will look like. My current setup (NAS + control point on an iPad) works reliably and provides excellent SQ, but there certainly is room for improvement in the user interface. If PSA can construct a GUI that works well for newcomers and more experienced folks and (unlike Roon) takes advantage of existing tags (FLAC/Vorbis and ID3-4) I would jump on board, even if I had to buy some new hardware. If the GUI could be made available on an iPad to control my existing setup, I would be one very happy camper!
Paul are you planning to do some sort integration with Tidal?
magicknow
Thanks for the update @adminpaul . Interesting perspective. Certainly looking forward to the Octave. I understand the decision not to use Roon as the OS/interface. Yes, its expense can certainly change the economics at the lower end of the scale, and there is a risk associated with any partnership or vendor relationship. Much as I love Roon, I would welcome another superior interface to the landscape. Like others here, I liked eLyric and still miss some of its features compared to others.
I am waiting for my Mivera Superserver right now; it will run Roon 1.3 when released, but the license expense will be mine to incur [or in my case, already incurred]. Another option with the Mivera stuff is running Daphile [I know some folks really like it, but I am not a fan]. The Mivera biz model is of course different from PSA, so I donāt mention this as a competitive comparison, only to point out that there will likely be many ways to skin the network cat for quite a while to come. I expect the Mivera server to improve on my network experience by segregating the Roon and server functions, but I wonāt know until I try it.
I like the target price range for the Octave. There are some boxes out there that are ridiculously priced for what they offer, and some of them have what we might politely call underwhelming innards.
If and when yāall are seeking Octave testers, count me in!
Alekz said As of SSD vs HDD. Since the music server is supposed to be placed in the "audio room", there must be no rotating disks . . .On a knee-jerk level, this is appealing. However. many modern hard drives are silent or so quiet they cannot be heard a foot or so away. And there are those who claim SSD emits various potentially problematic electrical fields and you thus do not want them inside the device.
Probably best is to provide a couple of options such as SATA slots for your drives of choice and a number of USB connections, both powered and powered. The user can then indulge in his bias de jour.
Of course, if PS Audio determines certain types of drives and/or locations of drives sound better they should go with this.
Alekz said Backup software can be as simple and effective as a UNIX rsync command (or grsync with a primitive GUI), no fancy stuff is needed.If my understanding is correct Linux Rsync mimics Robocopy in the Windows world, so this should do the trick on a command line level.