SynoGate and walk down memory lane

[White Bear]

For what it is worth... We reluctantly chose all Synology HDD and NVMe for a major NAS upgrade last August. As @Rusty mentioned, our decision was based on concern that drive restrictions or functionality limitations might be imposed in the future. We are long-time Synology NAS users. And I felt at the time that our NAS investment should be all-Synology to streamline any contact I or (more importantly) others, here, may have with Synology support in the future. We have had no technical issues with 30+ Synology drives so far.

 

[Jan Janowski]

I can only post what works here, but as an End User who has dealt with hard drives since mid 1967 (and it fit in the bottom of a rack!)…
I have history with a manufacturer. Over many professional jobs. They have treated me fairly - dealing with professional and even personal concerns - in the past. And with ~ 40K+ hours of use, on 2 of 3 SHR systems now… from this manufacturer: I see no reason to change, unless someone has lots of proof indicating otherwise.

I’m more picky than Synology! I dealt with old raid rules: saying only: identical drives with identical firmware can go in a raid. I believe in that rule: It works!

From Synology I see only: edicts, lockout statements, backed up by ???? Technical data?

I can fully understand having SE ‘only work’ on Synology drives.
Yes, Really, I can!
It’s ‘their system’ and drives’. They might have ‘found something’ No one else has! This sort of thing has happened before!

But a lockout?. Shame!

Be careful in your demands, it might have reactions you won’t like!

Things are way too expensive right now! I know of multiple companies that are withholding new expenses awaiting for prices to go back to a year or two ago. I, too, as an end user, have plans on what to do, should a computer NAS or HDD/SSD Die & can’t be repaired. Ain’t Pretty, but it’s do-able! We too have withheld all future purchases do to insane high prices!

The combination of Price and reaction to a lockout - could cause a double whammy to anyone’s bottom line!

NO ONE is going to tell me I can’t use my manufacturer of choice for drives that were specifically designed for that operation! NO ONE!

 

[DSfuchs]

The first hard drives didn't appear until 1980, did they? Or did you get your hands on a decommissioned RAMAC?
That's my first one:
Seagate: ST225
First cust. shipment: 1981
Interface: ST412
Recording method: MFM
Capacity unformatted: 19.14 MByte
Capacity formatted: 15 MByte

 

[Jan Janowski]

Had a HDD On a Data General Nova 2 ??? From company that became Seagate or Seagate as part of a CDL PEC 120a in 1967.
It was used via parallel interface and modified boards to control Ampex 2” Quadraflex Broadcast VTR’s. It was the first fully computer controlled videotape editing system. With it I made Thousands of TV Commercials, many pilots of TV Shows some still on now. That system was first developed in Canada by CDL with hard drive as PEC 120 in late 1965.
1970-ish. Did much work with Fujitsu and Quantel on drives for the “Paintbox” at TV Stations, and later used on Robert Bosch FGS4000 the first computer Graphics system.
These were all floor or rack mounted systems, way before any pc.
1980 or so was playing with custom raid arrays of as many as 40-48 PATA drives that was an attempt to do both HD (yeah, it was around in infancy then), and SD as a Hard Drive video recorder/player for edit. That company died a horrible death.

 

[DSfuchs]

Professionally, I started in 1986 as an operator on a System/38, working with external IBM 3370 Direct Access Storage Device (DASD).
Your Data General Nova 2 was launched in 1973!

 

[Jan Janowski]

mid-late 80’s HDD’s (Winchesters as they were called), the drives themselves were being physically modified by Abekas, not drive mfg, to become stand alone SD digital video layering edit systems. Do a search on Abekas A62 to see what they could do! Forgot drive mfg name they used - is no more, as that mfg modified the drive plate disc coating and there were so many changed before it was realized it was NG, and since each drive was physically modified by Abekas, and enough drives died (minimum of 4 per A62, 2 in A64, 1 in other A hardware… that pretty much enough modified drives failed & the drive company disappeared before fix was found and drives got modified (it’s modified disc buss was also crazy, couldn’t be field modified. The Drive Bad Frames-Error Map of each drive was stored on EProm outside of drive! I used to edit Prom files and write new ones to fix (hide) data errors based on video time code of data error, while waiting for drive replacements to come, so they could continue using it for production work.

HDD Disc buss changes has been the deathnell of so many good drive companies. Then there was the SCSI Standard: that had as many combinations as the RS-232 ‘Standard’ did for serial! & pretty much all dissimilar!

 

[DSfuchs]

The Winchester drive was installed in the system unit of my System/38; it was the IBM 3340 from the year 1973.
Perhaps you meant Watford Electronics, who also sourced the drive from IBM.

 

[Jan Janowski]

Spent last hour searching defunct SCSI-1/2 drive mfg. found nothing. I Gave my A62 engineering class notes away pre 2000, and I forgot who to, but drive wasn’t that mfg.
Drive was about 6” by 8” by 12-14”, with an Abekas drive R/W board & Abekas interface board mounted above it, and weighed 20-25lbs. 2 drives per 6/8 RU chassis. 2x drive chassis and an 8/10 RU PROCESSOR CHASSIS.
Regarding drives; If I remember correctly, spindle speed was decreased slightly; to a multiple of 59.94fps, so it locked to multiple of vertical sync of NTSC. Both drives moved arms identically so they both worked in tandem.
The 8 bit video was stored on plates in a 4:2:2 format:
NTSC Composite was decoded to D2: Y C Y C Y C Y C 8 bit format 4 samples of 8 bit Luminance were stored samples on front drive, and 4x samples of color signal samples of 8 bit chroma are stored on rear drive. 50 seconds of video on first drive enclosure. 50 to 1:40 was saved on second chassis. You bounced back and forth between the two keying new video as a layer over the previous edit, making the full result first generation due to digital cloning. NTSC Never looked so good! The A62 was NTSC ONLY. A64 could do NTSC or PAL as well, but it did SMPTE 601 Color Space 4x luminance, 2x R/Y & 2x B/Y (stored as Y RY Y BY Y RY Y BY not D2, But only a SMPTE video disc recorder/player.

All this is just proof that folks who build hardware have in past modified drives to do something new & unusual, that others have not done.

 

[White Bear]

The Abekas example is an interesting historical case that demonstrates how storage vendors can change firmware to suit a particular application. However, I’d still like to see much more objective validation of Synology’s claims regarding the benefits of firmware changes in their branded drives. Specifically, I’d be interested in independent studies showing measurable improvements in reliability, performance, error handling (particularly during RAID rebuild), or data integrity compared with equivalent non-Synology enterprise drives operating in the same NAS environment.

 

[Jan Janowski]

Agreed! They received an Emmy for it, too! (I have 2!)

Now though I have no Qualms about Synology Releasing Feature's only THEIR SYSTEMS AND THEIR DRIVES can take advantage of...... (Really!)

I DISAGREE with Synology's 3rd Party 'Lockout' of Drive Media... That was 'Designed' to be used in a NAS...

On a similar Vein, With Synology Looking for $$, I've always been wondering why they didn't come out with their own (Or: Branded from someone else if necessary - it would not be a new direction for Synology to rebrand something!) 2.5/5GB USB/Ethernet adapter and take advantage of the potential windfall from all their existing user base?? It would do 2 things: Show their user base they are still being 'thought of', which would promote great positive results... and bring in $$ too... While new products are also rolling out ???

What's that famous saying: "You'll catch more fly's with Honey, than Vinegar??"

 

[DSfuchs]

The Abekas example is an interesting historical case that demonstrates how storage vendors can change firmware to suit a particular application. However, I’d still like to see much more objective validation of Synology’s claims regarding the benefits of firmware changes in their branded drives. Specifically, I’d be interested in independent studies showing measurable improvements in reliability, performance, error handling (particularly during RAID rebuild), or data integrity compared with equivalent non-Synology enterprise drives operating in the same NAS environment.

Before Synology drives may appear in statistics such as those from Backblaze, it must be determined whether these drives behave neutrally and can be used in non-Synology environments.
However, I suspect that the firmware is forced to remain silent and that the drives are little more than electronic hazardous waste without a Synology unit. The OEM would otherwise act differently.

 

[White Bear]

Do you have a source indicating that a Synology-branded HDD loses its warranty when used outside a Synology NAS? I haven’t found anything in the published warranty information suggesting that the drive must be operated exclusively in Synology hardware. We've used spare Synology drives in a Mac/Thunderbolt/SoftRAID environment for temporary storage or data transfer with no issues. However, I understand that's just one anecdotal bit of evidence. Safe to say we would all like to see more robust studies.

 

[Jan Janowski]

I seem to remember a post somewhere about a year ago indicating not to use them outside of NAS. Can’t remember who posted or where……..
That would seem to be a good question for Support

 

[White Bear]

That would seem to be a good question for Support

Asked and answered in the affirmative; but that was almost a year ago. Our willingness to commit to a Synology NAS with drive restrictions was on condition of a positive response - that the drives could be used elsewhere.

 

[DSfuchs]

However, for standard systems, it is of no use to report a healthy status if the hard drive is going to fail in the short to medium term.
Why else would Synology disable all extended SMART analysis for its only drives starting with DSM 7.2 ... The goal is simply to make it through the warranty period without a request for replacement, right? As the sole supplier for its own NAS systems, the company doesn't need positive headlines—it’s a sure-fire success.
I would have liked to see a fair technical competition, one that would make customers happy to consider purchasing Synology drives. It comes down to the question: Which one lasts longer, an He10 with a 50k-hour penalty or a brand-new HAT?

 

[White Bear]

Synology’s marketing materials have mentioned benefits such as:

• Faster RAID rebuilds/resyncs through tuning of drive firmware and DSM behavior.
• Reduced risk of drive dropouts during rebuilds by coordinating error-recovery behavior between the drive firmware and the NAS.
• More predictable handling of bad sectors and read errors during rebuild operations.
• Better validation of drive firmware with Synology’s storage stack, resulting in fewer unexpected compatibility issues.

It appears that the technical basis for these claims is not unique to Synology. Many enterprise vendors tune HDD firmware. The idea is that a drive in a RAID should more quickly report a problem rather than spending minutes trying to recover a sector on its own. During a RAID rebuild, this can help prevent a drive from being marked failed unnecessarily.

The key question is whether Synology-branded drives are materially better than NAS-oriented drives from major vendors. Particularly in this circumstance.

 

[Jan Janowski]

Agree with your post.

I’ve seen many companies shy away from “shoot-outs” between manufacturer’s on identical product’s comparison, due to fact that there have been times where factory folk had ‘pre-visited’ the test facility and pre-installed something that would tilt the playing field in their direction during the shoot-out! Or, didn’t want to contribute because they knew theirs was inferior.
Not kidding.

What you end up with is then each manufacturer quoting specs of their own making…. Independant from a single test facility…….

And the result is a bunch of unsubstantiated claims filtered by corporate legal minds: effectively diluting all potential gains to vague references that are hard to put your finger on…. So you would want a single test location to bring sanity to the results!

And we all scramble to determine what we are to do: Because of that!!!

 

[White Bear]

Due to the complexity and diversity of RAID environments, the best evidence we could hope for is some combination of independent testing and anecdotal experiences over time. A benchmark that's meaningful to one type of user may be totally irrelevant to another. And it is unlikely that we would ever see large-sample statistics as with Backblaze.

Synology made a decision about storage media that has created confusion and concern in its traditional user base. It’s entirely reasonable to disagree with the decision, question the rationale, or even decide to buy from another vendor. But I think it’s more productive to debate the merits and consequences of the policy rather than assume bad motives on the part of the people who made it. As customers, we share a vested interest in Synology's success.

 

[DSfuchs]

• Faster RAID rebuilds/resyncs through tuning of drive firmware and DSM behavior.
• Reduced risk of drive dropouts during rebuilds by coordinating error-recovery behavior between the drive firmware and the NAS

So, Synology's Quick Repair essentially handles the top two points, and it works on all the other drives, too. The one thing missing to fulfill that promise is the ZFS file system, which Synology does not offer us. Perhaps something is on the verge of being announced.

• More predictable handling of bad sectors and read errors during rebuild operations.

The simple trick of merely shortening the wait time for a read error won't improve matters. But it will certainly lead to higher sales figures.

• Better validation of drive firmware with Synology’s storage stack, resulting in fewer unexpected compatibility issues.

That sounds logical, assuming the NAS operates exclusively using its own hard drive firmware. Arrangements were made for that.


The aim is to suggest to the customer that the fear of total data loss during recovery can be eliminated simply by using the manufacturer's own drives.

However, this works much better with ZFS, and ideally only with a mirrored drive setup. With Btrfs, the RAID metadata is simply located too close together. Fragmenting the data across many hard drives in a RAID 5 configuration is absolutely counterproductive, not only in terms of the time required but also regarding the level of risk. The more drives involved in the recovery process, the greater the risk of total failure, and the longer the period of anxious suspense. Simply using hard drives with perceptibly different runtimes helps, too.

If redundancy data is corrupted during hardware recovery, recovery is certainly impossible; even specialized drive firmware won't help in that case.
Only ZFS can handle advanced self-healing and also address bit-flipping issues.