Hi I just purchased a miniComputer from the chinese vendor Qotom. It seems like a pretty nice little device but I've run into a very significant problem in that the UTM ISO won't recognize more than a single NIC of the 4 available.
Can anyone assist?
Thanks,
Doug
SPECS:
Intel Core i7 4500U Haswell
Intel 4400 Graphics
1x HDMI 1x RS-232 1x mini-PCIe 1x mSATA
4x Intel I211-AT Gigabit Network
No, that is not what I'm interested in doing. It's possible to remove an entire VM from an ESXi implementation. I'm looking for a bare iron installation not VM.
Note: I'm running a VM on ESXi at the present time. This is not secure enough for me.
just wanted to give a quick update after a little more than 2 weeks of usage:
performance is great like i said, IPsec Site2Site via AES is up to 850MBit/s over LAN, OpenVPN via CPU is 200Mbit/s per Core and about the same for IPS (200Mbit/s per Core for IPS)
---
Sophos UTM 9.3 Certified Engineer
dougga said:I don't think the number of snort processes is configurable. Did you figure out how to do that?
This works easily:
console> show ips-settings -------------IPS Instances------------ IPS CPU 1 0
console>set ips ips-instance add IPS cpu 1 0 1 console> set ips ips-instance apply
Finally proof, you have assigned a second CPU to the second IPS instance:
console> show ips-settings -------------IPS Instances------------ IPS CPU 1 0 2 1
Please send me Spam gueselkuebel@sg-utm.also-solutions.ch
Hey Ben,
So I can't do a thing to challenge this Qotom box with only 40Mb/s download.
I started doing power and cost analysis and am running into something strange.
My UPS has a sensor that tells me how much power various devices are using.
Well, the Qotom while booting doesn't register any power at all. I'm assuming this is a bug in the UPS.
Do you have the ability to test power consumption (watts)?
I've been doing calculations to see if or how much running on AWS would save me but it's possible these very lower power implementations might match the savings of Amazon and adding the simplicity & security of onsite has some value vs. remote/hosted.
Note: I think this Qotom is fairly close to the M4 Large AWS instance which is $0.10/hr of processing of 2cores @ 2.4GHz. Half the cores but twice the speed. [?] The break even for power cost for me in Seattle would be 24 watts. Given my UPS doesn't register any power, I'm assuming I'm beating this.
All of this is based on my extremely low utilization which probably means this is only a guestimate. I'd be very interested in plugging in your numbers to see what your break even would be based on your more robust workload.
Cheers,
~Doug
An interesting thought occurred to me. If you have similarly low (negligible) power usage, I'm wondering if these are knock-off intel cpus.
I'll wait before investigating this red herring.
Examination of a fake Intel Core i7: https://www.youtube.com/watch?v=92uflDSYUHw
New Egg dealing with counterfeit merchandise: https://gizmodo.com/5488699/newegg-confirms-that-counterfeit-intel-core-i7-processors-were-shipped
Chinese Knock-off Intel CPUs: https://www.quora.com/Are-some-Chinese-manufacturers-duplicating-Intel-CPUs-as-well-like-the-Core-i7-i5-i3-etc
Intels testing software: https://downloadcenter.intel.com/download/19792/Intel-Processor-Diagnostic-Tool
After reading these article, I'm tempted to install Windows and test the thing. The power consumption seems just too good to be true. I'm wondering if we've been scammed - though for this use-case, fewer CPU features for better power consumption is exactly what I'd choose. Diving further down the paranoid rabbit-hole, there might be some features we may not have bargained for that could be "problematic" for firewall usage.
The Intel specifications show a 15 W general power consumption when all cores are active. That's very low... so who knows? Perhaps everything's fine and these are exceptional little machines. Given my power break-even number is 24 watts it appears these Qotoms are as good or better than Amazon AWS... unless they're knock-offs in which case I'd never trust 'em.
Regarding power. My ups reads 32 watts for all the network equipment during idle. This includes modem, r7000 router in AP mode, a 10 yr old 3com 8port gigabit switch, obi200, the qotom box (w/ esxi, utm, freepbx, and a light centos install for monitoring the ups). The problem with this is this is a converted value. Specifically ups reports a percentage of rated power (in my case 810 watts). 4% is roughly 32 watts. The values reported are integers only. When saturating the pipe (350mbps), power will spike up to 50 watts per the ups.
Sometime tomorrow I'll shut the qotom down but leave everything else up to get an idea how much draw there is from the other devices. Then a more accurate power consumption can be evaluated.
CPU:
Looks pretty genuine to me.
Ran the validator using win10 installed on an external usb 3 drive. Makes for a handy tool without having to swap drives.
--------------------------------------------------------------------
CPU1
Genuine Intel CPU Test
Module Version: 1.0.12.64b.W
Start Time: Sat Nov 4 08:29:22 2017
Test Result - PASS
Expected: GenuineIntel
Detected: GenuineIntel
End Time: Sat Nov 4 08:29:22 2017
Total Time: seconds: 0
--------------------------------------------------------------------
CPU1
CPU Brand String Test
Module Version: 1.0.14.64b.W
Start Time: Sat Nov 4 08:29:22 2017
Test Result - PASS
Expected: Intel(R) Core(TM) CPU @
Detected: Intel(R) Core(TM) i5-5250U CPU @ 1.60GHz
Intel(R) Core(TM) processor detected..!!..
End Time: Sat Nov 4 08:29:22 2017
Total Time: seconds: 0
--------------------------------------------------------------------
----------------
UPDATE:
Powered everything down, then restarted just the qotom box. UPS wattage jumped between 0 and 16 watts. I believe the smallest resolution is 8 watts. So considering there's a 5400 rpm drive inside, I'd say probably upwards of 20 watts is the max it'll draw under full load. Booting esxi and the vm's is a load but not 100% load. I think based on the results above this thing is the real deal.
Good info.
The Genuine Intel question appears to be resolved. I think I'll do the same thing just for practice.
Regarding the power consumption, however, the comparison is to the Qotom mini PC running Sophos UTM directly on bare metal.
The number of variables of virtualized scenarios is hard to track: there are just too many settings that impact load.
Having said that, though, one can safely say that running bare metal would be considerably less power / more efficient that via a hypervisor & vm so 20+ watts max is astonishingly good news!
Security is a principle concern as this is a security gateway product. The whole VM scenario for running security gateways seems suspect as people have contests to see how quickly a hacker can retrieve AN ENTIRE ESXi VM from a hypervisor without any knowledge let alone authorization from the VM administrator. The notion of running Sophos UTM on a hypervisor has outlived its usefulness (IMHO). Admittedly, I did it for years. No longer.
Great data, thanks!
Cheers,
~D
