Why I’m not happy with Syno branded HDDs or SSDs strategy

Currently reading
Why I’m not happy with Syno branded HDDs or SSDs strategy

jeyare

Subscriber
1,876
623
@DataNAS
base of creating of each strategy must be:
- an achievable target definition
- in defined time
- in defined market

Then you need perform a deep dive analysis of As-Is vs To-Be to able tune:
- the vision to achievable and measurable (checkpoints) scope = a plan
- in defined sequence/hierarchy = no one builds a house from the roof

So when you will take into account your point:
namely, enhancing product reliability
I perceive it as part of right vision.
However, as you can see, vision is just a group of words w/o achievable, sequential plan.
Specially when you would like to step in the really different market from retail = the enterprise market.
Create the product reliability from locking of the product to strictly branded (their) HDD - you can try it only once. Do it as first step is really deadly. Ofc - negative campaign is also campaign. You can do it in retail. But no one will forgive you anything in the enterprise market.
TBS: all big enterprise NAS vendors have locked also disk drives (tested). But no one for single vendor or their own brand. No one.
Just Synology, from 0.000000001% NAS enterprise market share.
As was written here. Syno needs for the enterprise market rebuild their point of view to enhancing product reliability in most important way:
- DSM recovery in case of primary partition damage
- enterprise SLA based support
- native block based iSCSI storage (no fork based)
- SAS support
- business partner network for standard or heavy enterprise projects
- ...
this is a foundation = must be done.
Then their can build a walls, e.g. reliable disk drives.
 

jeyare

Subscriber
1,876
623
In enterprise market is absolutely normal, that you can’t ask a support when you will use a part or spare part uncertified. No doubt. When vendor needs to guarantee a support even SLA, then a strictly rules must be defined.
Then the enterprise market can make a choice:
- vendor lock to literally two drives (HDD, SSD)
- more choices.

The RAM modules have different point of view in the enterprise market - you can’t install other models as was defined by vendor. In enterprise operation it’s right rule. Because you need a guarantee of the operation (up to your paid support service).

Then Syno needs to open a simple calculator:
- do I need invest a lot to the enterprise market?
- or I will stay as honoured vendor in retail and SMB?
With single DSM it’s not achievable. dtto with same obstacles for retail/enterprise customers.
 

DataNAS

Subscriber
22
4
Last edited:
Synology coding concerns me... If their firmware is as reliable as their software, interesting times are ahead.
Hopefully, Synology's testing will mitigate the risk: "The HAT5300 has been tested for well over 300,000 hours on every storage platform we have released since 2017. Intensive tests and thousands of test cycles are designed to replicate datacenter environments and punishing 24/7 availability requirements. Tests range from operations as simple as rebooting the system under different temperatures to complex file-copy tests that can magnify deficiencies in how each drive responds to non-ideal conditions." (Source)
-- post merged: --

"Synology will not provide technical support if your device is not on the Synology Products Compatibility List."
It appears that Synology's decision to require hard disk drive compatibility as a prerequisite for technical support applies to both RackStation and DiskStation products.
 

jeyare

Subscriber
1,876
623
Last edited:
300,000 hours / 1,000 HDDs / 20 NASes = 15 hours per single disk = 0.5 day only for each of them .... but they don't have such big-bay NASes
or
300,000 hours / 100 HDDs / 20 NASes = 150 hours per single disk = 6.25 days only for each of them .... changed in low time range or just 100 HDDs tested?
or
300,000 hours / 10 HDDs / 20 NASes = 1,500 hours per single disk = 62.5 days only for each of them .... not enough

or
they spent 300,000 hours for every single HDD in the test = 34 years for each ?

statistic outcome is a powerful weapon, until the moment when people start thinking about the math :geek:
-- post merged: --

I think, they need in Synology change someone who is responsible for such official statements, because it's great for the retail dummies, but not for experienced from the enterprise.

Every competitive vendor would enjoy this "300k hours advantage" on one nice big slide for their customers - look how they proposed a trustworthiness.
 

fredbert

Moderator
NAS Support
Subscriber
2,158
871
NAS
DS1520+, DS218+, DS215j
Router
  1. RT2600ac
  2. MR2200ac
Operating system
  1. macOS
Mobile operating system
  1. iOS
So you don't think that one HAT5300 drive has been tested for 300,000 hours?

Surely this plan has been under wraps for 34 years and only now, when the time is ripe, do they announce the results of their testing of the drive. It will soon embark on a ticker tape parade of all major global cities and then retire to the Hall of Fame Museum for Rigorously Tested Devices.
 

Telos

Subscriber
1,432
491
NAS
DS418play, DS213j, DS3622+, DSM 7.1.4-11091
So you don't think that one HAT5300 drive has been tested for 300,000 hours?
3.png
 

DataNAS

Subscriber
22
4
Yes, it would be beneficial for Synology to provide more information on its HAT5300 testing regimen – indeed! :) One additional detail reported by Synology is that the 300,000 hours of testing consisted of “1000+” cycles of “stress testing” to “ensure that HAT5300 drives can withstand even the most demanding enterprise environments" (see here).

The drive is rated for 2.5 million hours MTTF, which equates to an Annual Failure Rate of about 0.3% - meaning that a datacenter with 1,000 drives would experience 3 failures per year. That’s competitive with other enterprise drives, such as the Seagate Exos.
 

Telos

Subscriber
1,432
491
NAS
DS418play, DS213j, DS3622+, DSM 7.1.4-11091
Most of these specs appear consistent to Toshiba's MG line. I suspect Synology's is just parroting Toshiba performance data. Who knows?
 

jeyare

Subscriber
1,876
623
they can’t beat even the Exos 7E8 (performance, price, ...)
never Exos X line (performance)
maybe IronWolf Pro, bud it’s midrange, no enterprise

and 1000+ stress cycles means nothing real for me ... even 1001 cycles
 

jeyare

Subscriber
1,876
623
Follow Syno official statement:
Multi-client sequential read throughput. Performance testing was conducted by Synology using 12 drives on an SA3600, configured using RAID 5, against similar capacity (12TB) and class (Enterprise) drives with IOMeter (64KB blocks).

First:
Multi-client sequential read throughput is really low level load for the systems.
No defined test samples there.

Second:
So Iometer , OK.
I remember for the Iometer in 2001, when it was interesting product from Intel. Yes 2001.
Just downloaded and checked again.
Nothing was changed from 2001.

Some insights:
- testing of RAW drives provides better results
- there isn't know setup of the test workers - performance load, just 64K block
- when there was used just 64K block setup (what is great for large files = media storage), why they don't provide 512bytes block results (DB driven purpose)?
- then we don't know if there was setup of the IO Data Pattern by Total random set or Pseudo random (same random set reused for each write sequence) or just by Repeating bytes (one of the mandatory setup from Iometer).
- also there are plenty of parameters for the test scenario(s) what make(s) heavy impact for the results.

So, Syno is hiding details. Maybe there is a reason. For me it's out of trustworthiness.
For the retail it's useful. Not for the enterprise.
 

DataNAS

Subscriber
22
4
@jeyare, I believe we agree that Synology could do a better job in communicating and documenting their testing of the HAT5300. To be fair, though, this criticism is equally applicable to others, such Seagate or WD.

One thing to keep in mind, however, is that Synology does not need to test the underlying hardware, since that has already been done by Toshiba – plus, these drives have already been in use for a few years. (Interestingly, the 2020 Backblaze drive statistics for the HAT5300-16T [Toshiba MG08] report zero failures across 1,014 drives so far.)

In summary, I am incredulous that Synology would risk their brand reputation by releasing the HAT5300 without thorough testing – but, I remain open to the possibility that I am mistaken, and reserve the right to change my opinion as more data become available. Time will tell….

Peace.
 

Telos

Subscriber
1,432
491
NAS
DS418play, DS213j, DS3622+, DSM 7.1.4-11091
I am incredulous that Synology would risk their brand reputation by releasing the HAT5300 without thorough testing –
... and then tie that to user technical support. Life w/Synology... an unending "beta" (alpha?).
 

jeyare

Subscriber
1,876
623
@DataNAS
I understand your point. And it’s absolutely right.

In case of the facts, we can manage of them only when we can get them or understand them.
In case of assumptions, there is a high chance that we will get sooner or later an understanding. Then during such debate we need to keep in mind:
Assumption is the mother of all f*ck-ups.

Some here know, how many ideas was proposed directly to Syno from this forum through tickets or direct discussion with Syno representatives.

Regarding the BB 2020 experiences with these Toshiba HDDs vs HAT5300:
- 3 P/N in operation from the MG07ACA series (14TA, 14TEY, 16TEY)
- the 14TA is different, because uses native 4k sector size, when 14/16TEY just 512e
- all HAT5300 are based on 512 emulation sector size
- Toshiba 14/16TEY shares just 0.7% from all the BB reported drives
- more important for the Drive days number. Seems to be they don’t use them in average more than 16 days in 2020 (14TB PN) or 33 days (16TB PN). Then here is possible reason of the perfect zero failure ratio.
- here is also written evidence from the BB report for these PNs:
While each drive model has only been installed for about two months, they are off to a great start.
- you can find there also, that these Toshiba HDDs wasn’t operated during 2018-2020 period
- last point - Toshiba has also in the main family this P/N: MG07ACP14TE with 512e sector, what is different from mentioned above. Then no one from us can be 100% sure, what is base of the HAT5300.

the report is here:
 

jeyare

Subscriber
1,876
623
final stage to the assumptions:
- there is also german “unboxed” review of the HAT5300:
- not an exact proof there, but useful for the complex considerations
- there is a statement - “The HAT5300-8T is technically identical to the Toshiba MG06ACA800E, whereas the HAT5300-12T is identical to the Toshiba MG07ACA12TE”

My points:
- MG07ACA line doesn’t have 8TB or 16TB capacity (only 12 or 14TB). Seems to be no way for the capacity downgrade for Synos’ HAT5300.
- MG07ACA ( from 2018) line is based on helium sealed design (20db) and 720g weight (9/18 platters/heads design)
- MG06ACA ( from 2017) isn’t based on the helium sealed design (34db) and there is capacity range 6-10TB, weight 770g, (6/12 platters/heads design, 1.33GB/platter)
- there is known a “story”, that Syno changes only the Toshiba firmware to “achieve” more performance
- and what is the base of Syno 16TB capacity - based on logic and weight (720g), max transfer speed (262MiB/s), 512e sector design, power consumption(4W), there is one and only candidate- Toshiba MG08 with single capacity (16TB).
- I can guarantee, that no one will change OEM process to recreate disk drive design/geometry (number of plates, actuators, heads,...) because it would be totally new development from a scratch. Really, really costly way. Then final weight of such upgraded HDD will be different. But 8TB Syno has exact weight as the MG06ACA. Same is valid for the 12TB Syno vs MG07ACA and 16TB vs MG08.
- Tune of the firmware is standard process for all storage vendors, then they use specific PNs. But don’t expect a rebuild of the entire firmware- no way.

What I miss in every single HAT5300 reviews, incl. Syno official pages:
- Acoustic performance in dB (depends on Helium sealed design)
- Helium sealed design mention or proof for the capacity >=10TB
... every single vendor uses this kind of important technology design identification in announcements and technical specs. Include Toshiba in the mentioned product line.
- Toshiba 16TB MG08 has 512MiB buffer size, but Syno specification contains only 256MiB. No weight change impact??
Syno doesn’t mention the helium sealed design. They forgot? Really? For the heavy capacity?
They forgot also the buffer size change? Without impact to the weight of the disk? Just they locked the buffer by firmware upgrade? Really?

For me its another big mystery behind the HAT5300 development and market communication.

How Toshiba mentioned Helium sealed technology in their HDD product info:

I hope it can help you in such “fast era” get more insight.
 
165
88
NAS
RS1221+, DS1517+, RS819, RS217
Operating system
  1. macOS
Mobile operating system
  1. iOS
I understand (from some industry sources) that the raw cache / SLC buffer size remains unchanged on the Synology version but as the only actual firmware change is to the way this bit of flash integrates directly with a Synology RAID and it gets worked much harder. The lost capacity is used as a reserve / wear-leveling pool to compensate for the increased usage. In all other regards the Synology drives are identical to the original manufacturer's offerings.

I also understand that the new firmware is not particularly strapped down and you can flash an original drive with the tweaked firmware. I wonder how long that option will remain, unless there is nothing they can do about it with Toshiba's existing design.
 

Create an account or login to comment

You must be a member in order to leave a comment

Create account

Create an account on our community. It's easy!

Log in

Already have an account? Log in here.

Welcome to SynoForum.com!

SynoForum.com is an unofficial Synology forum for NAS owners and enthusiasts.

Registration is free, easy and fast!

Trending threads

Top