I have just now re-tested this with the Netgear GS308 connected between the NUC and HP 2810-24G.
My initial test was erroneous as the link from the ISP DOCSIS modem to my firewall must have been 100 megabit, not Gig. This is the only connection where I can not easily see the link speed.
Below is the speed through the Netgear switch:
Speedtest by Ookla
Latency: 17.63 ms
Download: 504.75 Mbps
Upload: 25.32 Mbps
Sorry for the confusion!
It appears that the Netgear 8 port Gigabit switch GS308v3 does not have a switching engine to push high speeds. The Netgear switches are often mentioned as well functioning in combination with Roon. There are several other cheap Gigabit switches which may be different, I have only used the GS308v3.
I recently got my internet bandwith upgraded to 500/25 Mbps. Below is the speeds my Debian NUC sees through the Netgear (all devices connected to it displays Gigabit fdx) and through the HP switch that replaced it. 93-97 Mbps is “wirespeed” i.e. maximum physical capacity (payload) for a 100 Mbps connection even if the Netgear was connected to a Gigabit port. With the NUC connected to the HP switch there is no limitation and I can easily reach 500 Mbps.
There is no jumbo packets or any other differences in how the two switches were connected.
It is important to note that I never had any problems with the Netgear, but I only use a single RAAT enpoint. With many endpoints or a multi channel/multi way setup one might exceed throughput capacity.
With Netgear GS308v3
Speedtest by Ookla
Latency: 17.37 ms
Download: 93.81 Mbps
Upload: 25.27 Mbps
With HP 2810-24G
Speedtest by Ookla
Latency: 17.33 ms
Download: 508.05 Mbps
Upload: 25.29 Mbps
Well, that HP switch is managed and as such, not a suggested switch to use since managed switches can cause their own set of issues.
That aside, are you basing your statement solely on your test, or on the actual spec sheet for the GS308?
Yes, the HP switch is managed. I have two managed HP switches in my LAN and this has worked flawlessly with Roon for years. I installed the new one because I wanted more Gigabit ports.
The numbers are actual speedtest figures. I’ve read the spec sheet for the Netgear when I bought it, which is some months ago.
I’ve used a 308 for awhile, and I don’t remember it being capped at 98. Whilst my internet is still sub 100 Mbps in bound, when I get home tonight I’ll test my speeds between my desktop and NAS, both of which have Gigabyte speed NICs, and are connected on the same Netgear.
Can’t dispute what you say. You seem to know what you’re talking about.
Doubtful. Even at 100 Mbps., there’s gobs of bandwidth for Roon’s needs.
It would probably be enough. I am thinking of, say, 8-10 feeds of 24/192. That will approach 90 Megabit and may give problems. If you think you have Gigabit throughput it would be difficult to troubleshoot something like that.
As I said, I never had any problems with the Netgear switch in my system (no port errors or packet loss of any kind), but I only use a single endpoint.
Who would have that many endpoints capable of 24192?
Theoretically maybe but not in real life.
Just a few that have many endpoints and even fewer that play all endpoints at the same time, but there are some that have active crossover systems via convolution. Four or five way with subwoofers for just stereo.
This sounds like a 100BASE-T handshake. Did you at least check for a 1000BASE-T handshake on the port(s) in question?
Yes speed looks like 100BaseT, but the handshake was 1000Base to the managed switch. I had one NUC and three NAS boxes connected. All these where also1000 full duplex. The only thing that could have capped this is if one of the two firewall ports in the chain had handshaked at 100BaseT. Possible, but a bit unlikely. I’ll have to check again.
Something is wrong with your environment. I don’t know what it is and replacing a $28 switch with something managed is an OK way to “fix” it. But, there is something wrong with the environment and either that switch needs to go back for replacement or its not the switch.
I agree something must be off. I’ll request mods to hide the thread until I’ve checked this again.
All you cables are at least CAT6 or short 5E I assume.
my connection is via CAT6 connected Unifi 8 port switches and im cascaded through 3 of them to get my router USG3P Unifi too…I have a 1GB u/d fibre connection.
I gave up on Apple’s APEs and other cheap dlink/tplink/netgear/linksys switches years ago and never looked back
I’ve done a new test, see the updated first post. The little Netgear handles high throughput just fine!
Just for info, my network is composed of mainly GS308 and 316 Gbit switches. The oldest cable in it will be around 20 years old (Cat 5E probably, 15-20m) and forms the main backbone over which most wired traffic in the house flows to the ISP router and between floors. The network has grown over the years to meet the family needs, but I have always preferred wired over wireless connections where possible.
A while back I was curious as to how healthy the network was in terms of throughput and error rate, having never given any consideration to purchasing anything other than standard Cat 5E cable from Belkin or the Amazon Basics range when I needed to extend. I found a tool called iPerf, which was a command line tool then and probably still is now. I don’t recall the exact figures, but I was able to push just over 900Mbps from any part of the house to any other part, testing at multiple ports and from the most physically distant parts (garage where Roon lives) into different rooms in the house. There were no errors or retransmits either.
In reality, the devices you put on the end of the Ethernet cable will be the limiting factor for throughput, especially if they are not properly configured for Gbit speeds, or have slightly dodgy old drivers for their NICs (kept for compatibility because Realtek can’t be bothered to update their drivers for WIN 10).
Use exactly the same cables in your test and check the lights on the switch to make sure it’s a 1gbps link on all connections and not 100Mbps. There’s nothing wrong with Netgear switches generally, it might be your particular switch but I’d bet on the cabling.
I use the same cabling, but now I connect the ISP modem to a Gbit port on the closest HP switch (where I earlier had direct connection from modem to firewall). I can see that the speed is 1000 fdx with no error counters on the port. I use a Gbit vlan trunk between the two switches so the firewall (connected to the second switch) has Gbit through to the modem. Clearly the firewall and modem had (for some strange reason) handshaked at 100mbit.