NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
GS308E Access Management Site
Hello, I currently used two GS305e at different locations at home. Between the switches I used a trunc-connection with VLAN 101 (192.168.2.0/24) and VLAN 200 (192.168.10.0/24) on Port 5. Port 5 i...
I'm a network software engineer. I also write firmware for microcontrollers.
schumaku wrote:This could even be microcode issue on that switch chip. The firmware people (which just set some bits to the behavior of the switch like on an embedded device don't have control over it. To what I can see, the switch code is not a loaded module, probably burned-in with the manufacturing.
It's hard to make any assumptions without knowing what exact switch chip is used in the GS308E, but according to this video https://youtu.be/HuQ3iv4iLbw?t=322 it's clearly seen that the GS108 (a simple unmanaged switch) uses Broadcom BCM53118. According to the chip's description (https://html.alldatasheet.com/html-pdf/1074559/BOARDCOM/BCM53118/448/1/BCM53118.html) this switch chip is highly versatile, supports VLAN tagging and has one GMII/RGMII interface along with 10/100/1000 MAC for connection with a CPU or a wireless/DLS/cable router chip. That interface is obviously not used in the GS108 due to its unmanaged nature. Thinking that the chip can't tag/untag traffic received/sent on that internal interface renders it absolutely useless for what it's designed for.
Simple logic leads me to a conclusion that something clearly more advanced like the GS308E uses at least something similar as a switch controller partly because the GS108 does not even use managed capabilities of its own chip. It's hard to believe that the GS308E uses an inferior chip provided it needs a CPU for management tasks.
waxar wrote:
Thinking that the chip can't tag/untag traffic received/sent on that internal interface renders it absolutely useless for what it's designed for.
This switch chipset class is implemented in almost any CPE, small router, and also on non-managed and configurable small standalone switches.
waxar wrote:
Simple logic leads me to a conclusion that something clearly more advanced like the GS308E uses at least something similar as a switch controller partly because the GS108 does not even use managed capabilities of its own chip.
The GS108 is the typical design example for a design for non-managed switches. Newer unmanaged switches still have got the addition of a controller adapted to the GMII/RGMII to add some features desired, similar to the GS308E. It happened to more than one vendor that this added code was faulty, so complete series of unmanaged switches showed problems on operations.
Scratch the simple logic. These switches are a their heart unmanaged switches, built on non-managed switch cores.
Don't want to dispute if the add on is an additional microcontroller or a CPU, the boundary is unsharp nowadays,. And the chip added does take care of various tasks the simple switch chip can't do on its own, and for several functions it requires the full access to the data stream. This processor does not just handle the tiny IP stack, the web server (on very early E models the NSDP protocol), it also takes care of IGMP snooping for example. And yes that chip is linked to the GMII/RGMII. And because there is no VLAN tagging support on the switch chip, and the tiny IP stack can't handle the VLAN tagging, can't handle tls, but it serves both the switch runtime and the management functions like http, sftp, nsdp.
Certain larger GSxxxE and XSxxxE type models are indeed built on managed cores - there Netgear has made the management VLAN capability available. Beyond, the admin feature set was kept intentionally in line to the configurable, but unmanaged functionality.
This is all I can say, everything based on observations done while Beta testing some of these Plus switches over the years.
waxar When reading into the release notes, you might understand my non-reverse-engineering design info can't be to much off:
===
To support the device UI, all HTTP packets are forwarded to the embedded CPU. With Flow Control enabled, throughput can be affected if the rate is higher than the CPU can handle.
...IGMP dynamic router port does not timeout exactly at 300 seconds. It could be anywhere from 300 to 600 seconds.
IGMP host timeout is not exactly 260 seconds. It could be anywhere from 260 to 520 seconds.===
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!
