As even a single SATA III drive can hit 600MB/sec which is enough to saturate a 5Gbit/sec link, why was 10GbE not introduced years back on NAS systems ? Note that the NVMe drives will be crippled even by a 10GbE. (NVMe data rates are upwards of 2000MB/sec, 10GbE is limited to under 1200MB/sec)
Cost, power, and infrastructure. Yes, some people went used market for switching, but realistically, the amount of home users that have the wiring or even a want or need for 10, is a very tiny amount. (Most people rent and can't just add cat 6 wherever they want). So it's a risk for a company like Synology to create a $1600 storage device when most will build their own. nvme is used for cache. Saturating 5Gb or 10Gb is rare for any home user, doesn't matter what hard drive speed is.
The reason why 10 gigabit is expensive is because demand for it is low in the consumer space. Demand for it is super high in the enterprise environment but is still incredibly overpriced. It is copper based and requires the weakest ARM based processor and a fan. Companies are just milking it and will continue to do so. We are lucky that it is possible to get a proper 4 port 10 gigabit ethernet + 1 SFP port switch for $350. You can buy a 4 port SFP 10 gigabit for $150 but then you need compatible SFP modules after that.
Most consumers don;t really want cables at all anymore. Despite the unreliable nature of WiFi the convenience it provides is undeniable. So the market for home wired connections in general isn't that attractive. The market for home 10G copper connections is minute. You only get the benefits once the whole chain supports 10G (network card, cables, switch, whatever is inside the endpoints) and very few people transfer so much data inside the network to bother. In the end they get a laptop that's more expensive, uses more power, a more expensive switch and cables and very little to show for it.
I keep seeing people around here being completely ignorant of the reality outside their browser window, talking about how common home labs are and how they *need* 10G, etc. Home labs are actually rarer then hens' teeth in the grand scheme of things, most don't realistically benefit from 10G (I mean faster is always better but law of diminishing returns and all that), and manufacturers have no reason to bump up the price for 99.9% of the people so some guy can move his porn and warez collection from one place to another a bit faster.
I said it once and I'll say it again, whoever *needs* 10G already has it and even for reasonable prices. Like a switch with 8x1G and 2x10G copper can be had for under $200. Even less if used. And if you have a home lab going for SFP is actually a blessing since it's overall cheaper. If you need such a setup you should be qualified enough to configure it yourself and don't mind the medium used.
It's great to have newer and better, no objection there. I'd also like all that enterprise grade equipment to somehow be at consumer price levels. But the whining that people who don't *need* it should somehow subsidize it for the few that *want* it is pathetic.
I agree, there isn't a large awareness for actual consumer benefits when it comes to >GbE, largely because there isn't much actually driving that on the consumer side. I would very much like to run an all-fiber SFP+/SFP28 or faster network around the house, but I also have only one PC actually capable of saturating that. Even if I had two, what would I care so much about, that it has to transport across the network at a higher bitrate? Movies aren't going to magically run with more visual quality. Games aren't going to become better. Maybe if I was an editing/studio house with 3-4 editors working on raw high bitrate 8k HDR feeds while mastering on 8k LG OLED panels, I'd have an actual need-driven upgrade path. But for.... a few devices browsing the internet and some streaming media? GbE will have to do.
Normally you can get the network part of the 10G (cards, switch, cables, ideally used) as almost a rounding error to how much the costs of the rest of your infrastructure that would *need* it. Very fast machines (clients, NAS) filled with large, fast SSDs. People dish out thousands of $ for fast PCs, fast, loaded NAS, and then snub their noses at a couple of hundreds for the network part?
I'm one with a home lab who splurged on some 10G stuff (new & used, all short run copper) and come nowhere near utilizing the bandwidth! More than 1G yes, but 2.5 or 5 NbaseT would do just as well for less cash. Most people seem to have little clue about bandwidth and the difference between megabits & megabytes, let alone protocol overhead!
I agree but it's not even an issue if you can saturate it. Say you have 2 fast SSDs on each side but how much data do you have to copy and how often for those savings to make a difference?
100GB gets copied in ~15 minutes on a 1G connection. I booted up 100 VMs over 1G and even that boot storm wasn't unbearable. So unless you make a living out of moving data fast and your income depends on saving that time, *demanding* 10G is an epeen contest.
And the pathetic part is that some people are basically shouting that they *need* 10G because they do such important stuff but not important enough to take a few hundred $ out of their pocket. Nothing says "important" like saying it has to cost $15 bucks or it's too much. Kind of shrivels the epeen.
I would like to respectfully disagree with your comment because conversing about causal consumer electronics in respect to a $1600 16 bay NAS. Very few causal consumers would even consider such a machine for their home usage. That is not even the target purchaser for this piece of equipment so your analogy fails.
The other issue with 10GbE over copper is that the requirements for cable/bend radius/etc. are different from Cat5E and it requires a re-write unless you were originally cabled for it.
And to be honest the vast majority of end user desktops don't even see significant improvement going from 100 meg to gigabit.
Most office workers simply don't do that much data. Even streaming 4k video is easily done over 100 megabit with modern codecs and as things move to the cloud, having 10 gig on your desk makes no differences if you're communicating over a slow speed WAN link anyway.
and ? How does that allow me to quickly move 8TB of data from my Laptop to my NAS ? I just run a USB C hub connected to my Laptop and a couple SSDs but even that is painfully slow.
I don't understand a NAS (lots of and fast storage) at all unless you have wired points that people use. If you want to fileshare, WiFi is so slow you can just plug a storage device into your router. I have a 12 TB drive plugged into my router for filesharing via SAMBA across the house. I can plug another in if I want more storage. No WiFi is faster then the spinning platters of my slow HDD. I own my own place and there are even a couple wired points but they are never where I need them to be. So the TV and Android Box connected to it is about the only thing wired in, and I use Kodi on the little Android box from the Samba share anyway.
I keep thinking what am I missing out on by not having a NAS and I can never figure it out.
My Synology NAS syncs pictures from my Onedrive and my wifes's Google Drive (both free tiers and constantly full).
It's also used for backing up the household computers and I used to use it for storage for my (now broken) media player. The kids still watch the saved films and TV shows using their own laptops and phones.
My router has 4x4 802.11ac + Gbit ethernet and the slowest link is the NAS (budget model, can't quite saturate the Gbit interface), and the pair of 5400 rpm drives w/ random access.
Except for the RAID mirroring, cloud sync and SMTP alerts - I could use my router's USB port for sharing files like you do, although my router's Samba implementation is 'anonymous/full access for everyone' - which just doesn't cut it for me. External USB drives are available with RAID as well, but you can't get alerts for it.
With enough technical skills there's nothing on a home NAS that you can't achieve with alternative solutions. But a NAS really lowers the bar for the skill you need to configure a share, backups, security, a Plex server, a photo library, cloud sync, a web server, etc. and server that to the whole house. And it all comes in a neat compact package with low power and noise. For this it's all but guaranteed 10G is not needed.
A 6-bay 10G NAS is probably less for the home and more for a small business but it achieves kind of the same. Very few people would *need* this in the home.
Because you don't have a need for it is why you can't think of a reason. Despite that, people who edit movies/photos. Home developers that store assets on them, etc
I know a guy that has like 1TB of raw sound files he uses for a game he is working on. lol Lots of reasons people need lots of storage.
the problem with this logic is that most of the developed world now has gigabit internet in the 60$/month range.
why would you buy a NAS, when a mega or one drive sub is much cheaper and has the same perf. These products are utter garbage after 20 years of recycling gigabit ethernet with the "most ppl dont need more" slogan.
"why would you buy a NAS, when a mega or one drive sub is much cheaper and has the same perf. " why? maybe because some one may not want that type of storage, and would prefer to use a nas. what happens IF your internet access happens to go down ? then what ? cloud storage is good for some, and NAS storage is good for others. its called personal preference.
Citation needed on the gigabit internet claim. I'm based in the UK, where the most widely-available service is 516Mb and around 44% of households can get it. A whopping 8% of us have FTTP.
Jokes that aren't really jokes about the UK no longer being part of the "developed world" aside, I feel like you're overstating your claim.
I don't know where this equivalency between availability of gigabit internet = developed world leaves us in Australia. August 2020 figures (from Speedtest Global Index rankings) out OZ at average DL of 64.55 Mbps. Which puts us at number 61 (just above Uruguay and Grenada) and I had always considered that Australia ranks as a 'developed' country. We are getting a 'countrywide' rollout of 'fast' broadband through govt organised NBN (cable, fixed wireless, satellite) and gigabit is only going to be potentially available for the lucky few (maybe 20%) who are getting connected by FTTP. I was lucky and got connected by FTTC (fibre to curb - getting 48Mbps on a 50 Mbps plan good day), the greatest majority are FTTN (to the node) who might be lucky to see 50 Mbps if that. Still better than my 12 Mbps capped 4G connection tho .
We had 2.5G though that can be ran on short cat5e which is usually the case of small appartments, and its an limit where most of us will draw the line.
It was, the DS1817 had it back in early 2017. The reason it doesn't tend to be fitted as standard to Synology NAS boxes is twofold, firstly it raises the cost/price of the NAS and anyone who wants it can add it via the expansion slot anyway. Secondly there are multiple ways to connect 10GbE, and RJ45 isn't currently the most popular for home NAS setups, this extra RJ45 port is useless for people with SFP+ switches/cards in their home setup (currently the cheapest/most popular way to do it).
Probably because it takes a lot more hardware than is available in the vast majority of NAS systems to actually use 10Gb in the real world. For one, looking at peaks read speeds with an SSD for NAS system performance is pretty off. Realistically, you are going to be using mech drives with peak read speeds of ~200MB/s but realistic sustained speeds of 10s of MB/s at best. Maxing out much beyond Gb enet generally takes more than a single raid set.
only if you are doing purely sequential and are the only user. at which point, why are you using a multi-bay NAS when you can get the same performance with direct attach over usb...
also given the sizes of modern spinning disks, using any form of striped raid is pretty counter productive. If you need the redundancy, go to straight mirror sets, esp in the consumer nas space with its severely limited bays. For rust, anything less than +2/+3 redundancy has been counter productive for a while.
The use cases for locally attached vs. network are quite different. Performance is just one aspect. The most important one to make a difference is that one is available over the whole network (at whatever performance point) and the other one is not. And that's one hell of a difference for most use cases.
>> why was 10GbE not introduced years back on NAS systems ?
In the consumer space, setting up a NAS that needs or even can use that bandwidth is extremely unlikely. If they're using mechanical drives, your read and write speeds will tank the moment you do anything that doesn't involve very large sequential transfers. Most consumers won't be putting SSDs into something like this as well. Not to mention the complete lack of routers and switches with anything more than gigabit ethernet in the consumer space. I bet 99.99% of home users do not have anything more.
As much as it sucks for advanced users, 1Gb ethernet is probably more than plenty for the vast majority of people purchasing this type of device. Those who want more have plenty of options for higher bandwidth devices or know how to roll their own NAS.
It's not that I can't saturate a 10GbE connection, it's that even as a semi-pro user with a massively overbuilt infrastructure at home the times when it would actually occur are few and far between. Yeah I move a TB around sometimes but it's not so often that it justifies spending a small fortune and having continual higher power draw given that it's still reasonably quick even on 1Gig.
Almost no home users are doing tasks daily, weekly or even monthly that justify such connections. I'd rather spend the money on a faster internet connection and better wifi, and in fact I did.
Cost vs. the reality of the situation that most network shared storage rarely sees the ideal sequential reads that a Sata3 drive will saturate 5 gigabit with.
Most end user desktops don't have 10 GbE, most SMB/SME do not have 10 GbE switching for distribution and many are still even on 100 megabit.
In the space Synology are playing in 1 gigabit to the end user is the norm. Even in a 300 user office, our file/print server here is rarely seeing the large amount of throughput that would require 10 GbE.
Sure, there are edge cases but those are typically people who need more throughput than even 10 gig ethernet will provide, and are using local NVME or direct-attached storage with the NAS functioning purely as archive storage.
The focus on 10GbE is a red herring, for consumers and small businesses 2.5/5GbE is cheaper and more cost effective for the best future (including cheaper wiring costs), and Synology integrating 2.5/5GbE into more models (like the still 1GbE limited x20+ series) could have helped drive 2.5/5GbE adoption. That would still potentially saturate the NAS, but at 2.5-5x the performance and at minimal cost difference to end users. They could likely keep their same wiring and just need a 2.5 or 5 GbE capable switch to start realizing performance gains as they add 2.5 or 5 GbE capable devices to their network (which might just require new NICs that are also cheaper than many 10GbE NICs).
The QNAP x53D series, which are equivalent to the Synology x20+ in CPU and features, offer 2.5GbE integrated, and QNAP has an official USB 2.5/5GbE adapter to add faster Ethernet to many of their models.
There’s no reason Synology can’t be keeping up by now. They just don’t want to.
I still think Synology should sell its OS it uses by itself for NAS users. It could simply focus on improving hardware support and encourage software devs for its packages. Many would buy it for making own NAS. It could be a good money stream, if offered stand alone, subscription mode, let devs charge for packages for it. Open up lots of potential for them.
Synology seems to spend way to much on its hardware to justify to anyone to get one outside some very niche areas anymore. I love my synology don't get me wrong. Just like many if i every need more storage, i doubt i'm going to just get another one.
I totally agree. Cause the new DS1621xs is a strange device for $1500. Using 2015 CPU with no GPU, no official way to use SSD for volumes, just one basic copper 10G (no SFP+, no 2.5/5G) I like my 918+ but I too doubt that with current HW approach my next NAS will be synology
I think I'd rather DIY something like this for a home given the price and suffer with less capable or friendly software (not that FreeNAS is all that bad really). Not everyone will want that sort of solution. A one stop appliance with a warranty has its perks but for me the ticket price is pretty high given what you get back.
I think it's more interesting to read the comments than the article itself!
I've been fiddling with FreeNAS for the last 6 months, mainly on old hardware (circa 2010). Superb stuff, but only if you're a tech tinkerer and effectively have the time to get involved in a new hobby.
Kinda cool, but I would take a decent freenas any day, especially a cheaper one. I currently have one set up with an X58 board, Xeon X5660, and 18GB of RAM. I can always add a 10Gb NIC if I want as well for under $150.
Synology, QNAP, TerraMaster are good for end users who need turn key solutions where the most difficult part is to plug in the cables.
I you want DIY 10G solutions on the cheap, go with 10G SFP+ used switches, NICs and cables from eBay. You can for instance find SolarFlare 10G cards for $20. Note that 2.5G and 5G are recent protocols. 10G used hardware from Enterprise/Data Centers on eBay will not know about these protocols and will not autonegotiate down from 10G. So if you are planning to mix 2.5, 5 and 10G devices you're better with the consumer 10G hardware but for $$$.
Throughout forums it is common to read "Who needs 10G at home?", "SFP+ has no role to play at home!", etc. Well, I'm remember people saying "Who needs 1G at home?" in the 90s :-)
The real point is that once you "tasted" 10G or more at home, you love it and do not want to go back to 1G.
Then again it depends what you want to do with it.
On my side I was tired to wait for 10G consumer at "honnest" prices so 1+ year ago I dug for info and used hardware on eBay.
I ended up with InfiniBand :-) FDR 36-port switch for $200, ConnectX-3 or ConnectX-3 Pro NIC for $50 to $75, $25 per FDR 3 m DAC cable. So in all, still less expensive than today's 10G consumer...
...and I enjoy 35G to 44G IPoIB network (Internet Protocol stack over 56Gbps InfiniBand), take that 10G! Very happy about it. The trick is that you need recent PC hardware components (CPU, Memory, PCIe 3 or 4, good disks) to face the onslaught of such network throughput.
You still lose bandwidth due to the slowness of consumer local components (mostly disk IO) but with 35+ G your remote drive basically behave like your local disks or even faster if the "server" is more powerful than the client. With Linux, using NFS in async mode, our remote disk IO speed is basically the speed of the network :-)
Calling 2.5 and 5 GbE “recent” is kind of laughable at this point. The 802.3bz (NBASE-T) spec was finalized in 2016. There are second and even third generation chipsets from Aquantia and Realtek to enable it in devices like an x86-powered NAS. The QNAP x53D series, which are equivalent to the Synology x20+ in CPU and features, offer 2.5GbE integrated, and QNAP has an official USB 2.5/5GbE adapter to add faster Ethernet to many of their models.
There’s no reason Synology can’t be keeping up by now. They just don’t want to.
From the manufacturer's web site: "DiskStation Manager (DSM) is an intuitive web-based operating system for every Synology NAS, designed to help you manage your digital assets across home and office"
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
52 Comments
Back to Article
Duncan Macdonald - Friday, September 11, 2020 - link
As even a single SATA III drive can hit 600MB/sec which is enough to saturate a 5Gbit/sec link, why was 10GbE not introduced years back on NAS systems ?Note that the NVMe drives will be crippled even by a 10GbE. (NVMe data rates are upwards of 2000MB/sec, 10GbE is limited to under 1200MB/sec)
Dug - Friday, September 11, 2020 - link
Cost, power, and infrastructure.Yes, some people went used market for switching, but realistically, the amount of home users that have the wiring or even a want or need for 10, is a very tiny amount. (Most people rent and can't just add cat 6 wherever they want). So it's a risk for a company like Synology to create a $1600 storage device when most will build their own.
nvme is used for cache.
Saturating 5Gb or 10Gb is rare for any home user, doesn't matter what hard drive speed is.
quiksilvr - Friday, September 11, 2020 - link
The reason why 10 gigabit is expensive is because demand for it is low in the consumer space. Demand for it is super high in the enterprise environment but is still incredibly overpriced. It is copper based and requires the weakest ARM based processor and a fan. Companies are just milking it and will continue to do so. We are lucky that it is possible to get a proper 4 port 10 gigabit ethernet + 1 SFP port switch for $350. You can buy a 4 port SFP 10 gigabit for $150 but then you need compatible SFP modules after that.close - Friday, September 11, 2020 - link
Most consumers don;t really want cables at all anymore. Despite the unreliable nature of WiFi the convenience it provides is undeniable. So the market for home wired connections in general isn't that attractive. The market for home 10G copper connections is minute. You only get the benefits once the whole chain supports 10G (network card, cables, switch, whatever is inside the endpoints) and very few people transfer so much data inside the network to bother. In the end they get a laptop that's more expensive, uses more power, a more expensive switch and cables and very little to show for it.I keep seeing people around here being completely ignorant of the reality outside their browser window, talking about how common home labs are and how they *need* 10G, etc. Home labs are actually rarer then hens' teeth in the grand scheme of things, most don't realistically benefit from 10G (I mean faster is always better but law of diminishing returns and all that), and manufacturers have no reason to bump up the price for 99.9% of the people so some guy can move his porn and warez collection from one place to another a bit faster.
I said it once and I'll say it again, whoever *needs* 10G already has it and even for reasonable prices. Like a switch with 8x1G and 2x10G copper can be had for under $200. Even less if used. And if you have a home lab going for SFP is actually a blessing since it's overall cheaper. If you need such a setup you should be qualified enough to configure it yourself and don't mind the medium used.
It's great to have newer and better, no objection there. I'd also like all that enterprise grade equipment to somehow be at consumer price levels. But the whining that people who don't *need* it should somehow subsidize it for the few that *want* it is pathetic.
jeremyshaw - Friday, September 11, 2020 - link
I agree, there isn't a large awareness for actual consumer benefits when it comes to >GbE, largely because there isn't much actually driving that on the consumer side. I would very much like to run an all-fiber SFP+/SFP28 or faster network around the house, but I also have only one PC actually capable of saturating that. Even if I had two, what would I care so much about, that it has to transport across the network at a higher bitrate? Movies aren't going to magically run with more visual quality. Games aren't going to become better. Maybe if I was an editing/studio house with 3-4 editors working on raw high bitrate 8k HDR feeds while mastering on 8k LG OLED panels, I'd have an actual need-driven upgrade path. But for.... a few devices browsing the internet and some streaming media? GbE will have to do.close - Saturday, September 12, 2020 - link
Normally you can get the network part of the 10G (cards, switch, cables, ideally used) as almost a rounding error to how much the costs of the rest of your infrastructure that would *need* it. Very fast machines (clients, NAS) filled with large, fast SSDs. People dish out thousands of $ for fast PCs, fast, loaded NAS, and then snub their noses at a couple of hundreds for the network part?CaptainChaos - Friday, September 11, 2020 - link
I'm one with a home lab who splurged on some 10G stuff (new & used, all short run copper) and come nowhere near utilizing the bandwidth! More than 1G yes, but 2.5 or 5 NbaseT would do just as well for less cash. Most people seem to have little clue about bandwidth and the difference between megabits & megabytes, let alone protocol overhead!close - Saturday, September 12, 2020 - link
I agree but it's not even an issue if you can saturate it. Say you have 2 fast SSDs on each side but how much data do you have to copy and how often for those savings to make a difference?100GB gets copied in ~15 minutes on a 1G connection. I booted up 100 VMs over 1G and even that boot storm wasn't unbearable. So unless you make a living out of moving data fast and your income depends on saving that time, *demanding* 10G is an epeen contest.
And the pathetic part is that some people are basically shouting that they *need* 10G because they do such important stuff but not important enough to take a few hundred $ out of their pocket. Nothing says "important" like saying it has to cost $15 bucks or it's too much. Kind of shrivels the epeen.
Spunjji - Monday, September 14, 2020 - link
2.5Gb is really where it should be at for high-end home usage, and it's finally starting to filter into that market, which is good to see.phatboye - Sunday, September 13, 2020 - link
I would like to respectfully disagree with your comment because conversing about causal consumer electronics in respect to a $1600 16 bay NAS. Very few causal consumers would even consider such a machine for their home usage. That is not even the target purchaser for this piece of equipment so your analogy fails.Gigaplex - Saturday, September 12, 2020 - link
It's also hard to implement in a reasonable power envelope, which adds significantly to the cost.throAU - Tuesday, September 22, 2020 - link
The other issue with 10GbE over copper is that the requirements for cable/bend radius/etc. are different from Cat5E and it requires a re-write unless you were originally cabled for it.And to be honest the vast majority of end user desktops don't even see significant improvement going from 100 meg to gigabit.
Most office workers simply don't do that much data. Even streaming 4k video is easily done over 100 megabit with modern codecs and as things move to the cloud, having 10 gig on your desk makes no differences if you're communicating over a slow speed WAN link anyway.
TheinsanegamerN - Friday, September 11, 2020 - link
You dont need to run wire through the walls. You can just run a cable between your NAS and your desktop/wifi router.Ubiqutious - Friday, September 11, 2020 - link
and ? How does that allow me to quickly move 8TB of data from my Laptop to my NAS ? I just run a USB C hub connected to my Laptop and a couple SSDs but even that is painfully slow.I don't understand a NAS (lots of and fast storage) at all unless you have wired points that people use. If you want to fileshare, WiFi is so slow you can just plug a storage device into your router. I have a 12 TB drive plugged into my router for filesharing via SAMBA across the house. I can plug another in if I want more storage. No WiFi is faster then the spinning platters of my slow HDD. I own my own place and there are even a couple wired points but they are never where I need them to be. So the TV and Android Box connected to it is about the only thing wired in, and I use Kodi on the little Android box from the Samba share anyway.
I keep thinking what am I missing out on by not having a NAS and I can never figure it out.
sandtitz - Saturday, September 12, 2020 - link
My Synology NAS syncs pictures from my Onedrive and my wifes's Google Drive (both free tiers and constantly full).It's also used for backing up the household computers and I used to use it for storage for my (now broken) media player. The kids still watch the saved films and TV shows using their own laptops and phones.
My router has 4x4 802.11ac + Gbit ethernet and the slowest link is the NAS (budget model, can't quite saturate the Gbit interface), and the pair of 5400 rpm drives w/ random access.
Except for the RAID mirroring, cloud sync and SMTP alerts - I could use my router's USB port for sharing files like you do, although my router's Samba implementation is 'anonymous/full access for everyone' - which just doesn't cut it for me. External USB drives are available with RAID as well, but you can't get alerts for it.
close - Sunday, September 13, 2020 - link
With enough technical skills there's nothing on a home NAS that you can't achieve with alternative solutions. But a NAS really lowers the bar for the skill you need to configure a share, backups, security, a Plex server, a photo library, cloud sync, a web server, etc. and server that to the whole house. And it all comes in a neat compact package with low power and noise. For this it's all but guaranteed 10G is not needed.A 6-bay 10G NAS is probably less for the home and more for a small business but it achieves kind of the same. Very few people would *need* this in the home.
imaheadcase - Monday, September 14, 2020 - link
Because you don't have a need for it is why you can't think of a reason. Despite that, people who edit movies/photos. Home developers that store assets on them, etcI know a guy that has like 1TB of raw sound files he uses for a game he is working on. lol Lots of reasons people need lots of storage.
Gigaplex - Saturday, September 12, 2020 - link
Not if there's a wall in the way.azfacea - Friday, September 11, 2020 - link
the problem with this logic is that most of the developed world now has gigabit internet in the 60$/month range.why would you buy a NAS, when a mega or one drive sub is much cheaper and has the same perf. These products are utter garbage after 20 years of recycling gigabit ethernet with the "most ppl dont need more" slogan.
Qasar - Monday, September 14, 2020 - link
"why would you buy a NAS, when a mega or one drive sub is much cheaper and has the same perf. " why? maybe because some one may not want that type of storage, and would prefer to use a nas. what happens IF your internet access happens to go down ? then what ? cloud storage is good for some, and NAS storage is good for others. its called personal preference.Spunjji - Monday, September 14, 2020 - link
Citation needed on the gigabit internet claim. I'm based in the UK, where the most widely-available service is 516Mb and around 44% of households can get it. A whopping 8% of us have FTTP.Jokes that aren't really jokes about the UK no longer being part of the "developed world" aside, I feel like you're overstating your claim.
BearInBrisbane - Saturday, September 19, 2020 - link
I don't know where this equivalency between availability of gigabit internet = developed world leaves us in Australia. August 2020 figures (from Speedtest Global Index rankings) out OZ at average DL of 64.55 Mbps. Which puts us at number 61 (just above Uruguay and Grenada) and I had always considered that Australia ranks as a 'developed' country. We are getting a 'countrywide' rollout of 'fast' broadband through govt organised NBN (cable, fixed wireless, satellite) and gigabit is only going to be potentially available for the lucky few (maybe 20%) who are getting connected by FTTP. I was lucky and got connected by FTTC (fibre to curb - getting 48Mbps on a 50 Mbps plan good day), the greatest majority are FTTN (to the node) who might be lucky to see 50 Mbps if that. Still better than my 12 Mbps capped 4G connection tho.
deil - Monday, September 14, 2020 - link
We had 2.5G though that can be ran on short cat5e which is usually the case of small appartments, and its an limit where most of us will draw the line.James_Edge - Friday, September 11, 2020 - link
It was, the DS1817 had it back in early 2017. The reason it doesn't tend to be fitted as standard to Synology NAS boxes is twofold, firstly it raises the cost/price of the NAS and anyone who wants it can add it via the expansion slot anyway. Secondly there are multiple ways to connect 10GbE, and RJ45 isn't currently the most popular for home NAS setups, this extra RJ45 port is useless for people with SFP+ switches/cards in their home setup (currently the cheapest/most popular way to do it).ats - Friday, September 11, 2020 - link
Probably because it takes a lot more hardware than is available in the vast majority of NAS systems to actually use 10Gb in the real world. For one, looking at peaks read speeds with an SSD for NAS system performance is pretty off. Realistically, you are going to be using mech drives with peak read speeds of ~200MB/s but realistic sustained speeds of 10s of MB/s at best. Maxing out much beyond Gb enet generally takes more than a single raid set.Gigaplex - Saturday, September 12, 2020 - link
RAID5 speeds on spinning platter can exceed 500MB/s too, which also saturates 5Gbit.ats - Sunday, September 13, 2020 - link
only if you are doing purely sequential and are the only user. at which point, why are you using a multi-bay NAS when you can get the same performance with direct attach over usb...also given the sizes of modern spinning disks, using any form of striped raid is pretty counter productive. If you need the redundancy, go to straight mirror sets, esp in the consumer nas space with its severely limited bays. For rust, anything less than +2/+3 redundancy has been counter productive for a while.
close - Monday, September 14, 2020 - link
The use cases for locally attached vs. network are quite different. Performance is just one aspect. The most important one to make a difference is that one is available over the whole network (at whatever performance point) and the other one is not. And that's one hell of a difference for most use cases.inighthawki - Friday, September 11, 2020 - link
>> why was 10GbE not introduced years back on NAS systems ?In the consumer space, setting up a NAS that needs or even can use that bandwidth is extremely unlikely. If they're using mechanical drives, your read and write speeds will tank the moment you do anything that doesn't involve very large sequential transfers. Most consumers won't be putting SSDs into something like this as well. Not to mention the complete lack of routers and switches with anything more than gigabit ethernet in the consumer space. I bet 99.99% of home users do not have anything more.
As much as it sucks for advanced users, 1Gb ethernet is probably more than plenty for the vast majority of people purchasing this type of device. Those who want more have plenty of options for higher bandwidth devices or know how to roll their own NAS.
Jake13942 - Saturday, September 12, 2020 - link
Intel Xeon D-1527 was released in Q4 2015, Synology is charging $1600 for this LOLReflex - Sunday, September 13, 2020 - link
It's not that I can't saturate a 10GbE connection, it's that even as a semi-pro user with a massively overbuilt infrastructure at home the times when it would actually occur are few and far between. Yeah I move a TB around sometimes but it's not so often that it justifies spending a small fortune and having continual higher power draw given that it's still reasonably quick even on 1Gig.Almost no home users are doing tasks daily, weekly or even monthly that justify such connections. I'd rather spend the money on a faster internet connection and better wifi, and in fact I did.
throAU - Tuesday, September 22, 2020 - link
Cost vs. the reality of the situation that most network shared storage rarely sees the ideal sequential reads that a Sata3 drive will saturate 5 gigabit with.Most end user desktops don't have 10 GbE, most SMB/SME do not have 10 GbE switching for distribution and many are still even on 100 megabit.
In the space Synology are playing in 1 gigabit to the end user is the norm. Even in a 300 user office, our file/print server here is rarely seeing the large amount of throughput that would require 10 GbE.
Sure, there are edge cases but those are typically people who need more throughput than even 10 gig ethernet will provide, and are using local NVME or direct-attached storage with the NAS functioning purely as archive storage.
shelbystripes - Friday, January 1, 2021 - link
The focus on 10GbE is a red herring, for consumers and small businesses 2.5/5GbE is cheaper and more cost effective for the best future (including cheaper wiring costs), and Synology integrating 2.5/5GbE into more models (like the still 1GbE limited x20+ series) could have helped drive 2.5/5GbE adoption. That would still potentially saturate the NAS, but at 2.5-5x the performance and at minimal cost difference to end users. They could likely keep their same wiring and just need a 2.5 or 5 GbE capable switch to start realizing performance gains as they add 2.5 or 5 GbE capable devices to their network (which might just require new NICs that are also cheaper than many 10GbE NICs).The QNAP x53D series, which are equivalent to the Synology x20+ in CPU and features, offer 2.5GbE integrated, and QNAP has an official USB 2.5/5GbE adapter to add faster Ethernet to many of their models.
There’s no reason Synology can’t be keeping up by now. They just don’t want to.
imaheadcase - Friday, September 11, 2020 - link
I still think Synology should sell its OS it uses by itself for NAS users. It could simply focus on improving hardware support and encourage software devs for its packages. Many would buy it for making own NAS. It could be a good money stream, if offered stand alone, subscription mode, let devs charge for packages for it. Open up lots of potential for them.Synology seems to spend way to much on its hardware to justify to anyone to get one outside some very niche areas anymore. I love my synology don't get me wrong. Just like many if i every need more storage, i doubt i'm going to just get another one.
Qasar - Friday, September 11, 2020 - link
i currently have a 5 bay, and and looking at one of their 8 bays.mmm200 - Friday, September 11, 2020 - link
I totally agree.Cause the new DS1621xs is a strange device for $1500.
Using 2015 CPU with no GPU, no official way to use SSD for volumes, just one basic copper 10G (no SFP+, no 2.5/5G)
I like my 918+ but I too doubt that with current HW approach my next NAS will be synology
Aegrum - Friday, September 11, 2020 - link
Amen to that. I want to run their OS, but their hardware is *way* too dated for my interest. Paying $1600 for a 5 year old processor is kinda bonkers.PeachNCream - Friday, September 11, 2020 - link
I think I'd rather DIY something like this for a home given the price and suffer with less capable or friendly software (not that FreeNAS is all that bad really). Not everyone will want that sort of solution. A one stop appliance with a warranty has its perks but for me the ticket price is pretty high given what you get back.esoel_ - Friday, September 11, 2020 - link
Can I install freenas on it?Jake13942 - Friday, September 11, 2020 - link
Intel Xeon D-1527 was released in Q4 2015, Synology is charging $1600 for this LOLtyger11 - Friday, September 11, 2020 - link
Is anyone making an NVMe SSD-only NAS or DAS yet? They woyld be so much smaller.Even a 2.5” SATA SSD-one would be nice.
Bikerchris - Saturday, September 12, 2020 - link
I think it's more interesting to read the comments than the article itself!I've been fiddling with FreeNAS for the last 6 months, mainly on old hardware (circa 2010). Superb stuff, but only if you're a tech tinkerer and effectively have the time to get involved in a new hobby.
Shmee - Saturday, September 12, 2020 - link
Kinda cool, but I would take a decent freenas any day, especially a cheaper one. I currently have one set up with an X58 board, Xeon X5660, and 18GB of RAM. I can always add a 10Gb NIC if I want as well for under $150.domih - Saturday, September 12, 2020 - link
Synology, QNAP, TerraMaster are good for end users who need turn key solutions where the most difficult part is to plug in the cables.I you want DIY 10G solutions on the cheap, go with 10G SFP+ used switches, NICs and cables from eBay. You can for instance find SolarFlare 10G cards for $20. Note that 2.5G and 5G are recent protocols. 10G used hardware from Enterprise/Data Centers on eBay will not know about these protocols and will not autonegotiate down from 10G. So if you are planning to mix 2.5, 5 and 10G devices you're better with the consumer 10G hardware but for $$$.
Throughout forums it is common to read "Who needs 10G at home?", "SFP+ has no role to play at home!", etc. Well, I'm remember people saying "Who needs 1G at home?" in the 90s :-)
The real point is that once you "tasted" 10G or more at home, you love it and do not want to go back to 1G.
Then again it depends what you want to do with it.
On my side I was tired to wait for 10G consumer at "honnest" prices so 1+ year ago I dug for info and used hardware on eBay.
I ended up with InfiniBand :-) FDR 36-port switch for $200, ConnectX-3 or ConnectX-3 Pro NIC for $50 to $75, $25 per FDR 3 m DAC cable. So in all, still less expensive than today's 10G consumer...
...and I enjoy 35G to 44G IPoIB network (Internet Protocol stack over 56Gbps InfiniBand), take that 10G! Very happy about it. The trick is that you need recent PC hardware components (CPU, Memory, PCIe 3 or 4, good disks) to face the onslaught of such network throughput.
You still lose bandwidth due to the slowness of consumer local components (mostly disk IO) but with 35+ G your remote drive basically behave like your local disks or even faster if the "server" is more powerful than the client. With Linux, using NFS in async mode, our remote disk IO speed is basically the speed of the network :-)
shelbystripes - Friday, January 1, 2021 - link
Calling 2.5 and 5 GbE “recent” is kind of laughable at this point. The 802.3bz (NBASE-T) spec was finalized in 2016. There are second and even third generation chipsets from Aquantia and Realtek to enable it in devices like an x86-powered NAS. The QNAP x53D series, which are equivalent to the Synology x20+ in CPU and features, offer 2.5GbE integrated, and QNAP has an official USB 2.5/5GbE adapter to add faster Ethernet to many of their models.There’s no reason Synology can’t be keeping up by now. They just don’t want to.
RecycledElectrons - Sunday, September 13, 2020 - link
From the manufacturer's web site: "DiskStation Manager (DSM) is an intuitive web-based operating system for every Synology NAS, designed to help you manage your digital assets across home and office"Next time, define your abnreviations!