I want to provide some feedback for the Sophos UTM 9.1 home version.
I have it running on the following hardware:
GIGABYTE - Motherboard - Socket 1155 - GA-Z77-D3H (rev. 1.0)
CPU: Intel Core i7 (8 Core may be overkill, but performance is incredible)
32 GB RAM
VMWare ESXi 5.1
120 GB SSD
It actually performs better in a VM environment than running it on hardware (I believe I know why .may not be what Sophos tested it with), so thats why I sent my hardware information.
After measuring performance on 8 cores, I reduced the system to 2 cores and 1 GHz of power. It performed nearly the same, but now I get less bottlenecks from the application. I then added the third core (3 cores) and do not see a difference from the performance of 8 cores. I allocated 16 GB of RAM and noticed that it does not use 16, but 11.8 GB. Im going to reduce the RAM use to 13 GB.
So I forgot to explain why I added the third core. I am also using the Link Aggregation Group (I havent checked to measure NIC performance yet) and I have 3 Broadcom NICs as Uplinks to the Cable Modem. I also turned down the power of the CPU to 800 MHz (after learning that my Netgear R6300 is a dual-core router and was shocked a bit). The router runs at 800 MHz, so I deduced that if the UTM is running at 3.7 GHz with 8 cores, network performance may be impacted (no real data to show). What would occur is the Uplinks would show an error and I have to reset the Cable Modem, release the IPs, renew the IPs, restart the UTM several times before all is well, so something was actually going on at the Data Link layer for some strange reason. The R6300 would not have any impact, but the Cable Modem (Motorola SBG6580 SURFboard) would reset.
I plan to run some media streaming tests across 3 different computers today and measure that data.
I have it running on the following hardware:
GIGABYTE - Motherboard - Socket 1155 - GA-Z77-D3H (rev. 1.0)
CPU: Intel Core i7 (8 Core may be overkill, but performance is incredible)
32 GB RAM
VMWare ESXi 5.1
120 GB SSD
It actually performs better in a VM environment than running it on hardware (I believe I know why .may not be what Sophos tested it with), so thats why I sent my hardware information.
After measuring performance on 8 cores, I reduced the system to 2 cores and 1 GHz of power. It performed nearly the same, but now I get less bottlenecks from the application. I then added the third core (3 cores) and do not see a difference from the performance of 8 cores. I allocated 16 GB of RAM and noticed that it does not use 16, but 11.8 GB. Im going to reduce the RAM use to 13 GB.
So I forgot to explain why I added the third core. I am also using the Link Aggregation Group (I havent checked to measure NIC performance yet) and I have 3 Broadcom NICs as Uplinks to the Cable Modem. I also turned down the power of the CPU to 800 MHz (after learning that my Netgear R6300 is a dual-core router and was shocked a bit). The router runs at 800 MHz, so I deduced that if the UTM is running at 3.7 GHz with 8 cores, network performance may be impacted (no real data to show). What would occur is the Uplinks would show an error and I have to reset the Cable Modem, release the IPs, renew the IPs, restart the UTM several times before all is well, so something was actually going on at the Data Link layer for some strange reason. The R6300 would not have any impact, but the Cable Modem (Motorola SBG6580 SURFboard) would reset.
I plan to run some media streaming tests across 3 different computers today and measure that data.