Hello gents,
i received an UTM220 v5 manufactures in february 2013!(latest ver afaik, for a customer) and was very intrigued on what hardware is inside as the performance specs are quite "high"(8 gigabit ports..., ~200 users with all the jazz) -not to mention the steep cost of the unit!! [:O]-
so when i check the hardware i'm... -for lack of a better word- very underwhelmed on what's under the hood...., look:
A C2D Celeron E3500(a 2.7GHz, 1MB L2, 65W part): ¿WHAAAAT?1?!!, are you teeling me the latest version MADE IN 2013 is using an outdated +5yr old platform?!?!?, not to mention an already EOL processor model introduced in 2010... [:O]
i could have accepted a minimum of s1155 platform SNB Pentium(G850 series), an IVB Pentium would in fact have been an acceptable option(G20x0 series), but a wolfdale celeron????, at this price?, nope.. no sir...
With a G2010(that is 2.9Ghz, 3MB cache, IMC, 16 lanes of DIRECT TO CPU LOW LATENCY pcie 3.0 AND 55W even) that is proven to use FAR LESS idle power than any C2D platform the UTM220 would be modern and more believable, what's more it saves board/power space as it only requires ONE companion chip -which also uses far less power than the old ones-)
And of course this platform comes hand in hand with outdated core logic without PCIe 3.0 and an outdated ICH7 southbridge(802801GB), which means no AHCI either! (which with using a B75 or similar modern platform you would already have)
On the bright side the ram is DDR3, on the wrong side, 2GB ONLY!(and 2x1GB, at least dual channel) for 200+users banging hard on the web proxy?........
HDD is 250GB(looks like WD, couldn't check if it was an enterprise ABYS drive but i highly doubt it...., will edit later)... a little anhemic for today's standard capacities.. was hoping for 500GB(not like it will go out of space anytime soon but again... at this price point....)
What i don't know if the 8 individual NICs are PCI or PCIe (i'll have to check DMESG), i guess they're pcie as i did spot what looks like a pcie switch but i wonder from where they're taking the lanes....., are they taking the 16 from the NB(or maybe 8) or -god forbid- 2 out of he southbridge?.
A good test would be to saturate traffic 1 to 2, 2 to 3, 4 to 5, and so on and see where it stalls, a GOOD DESIGN for an 8 port gigabit would require a PLX tech switch at least in 8x8 9 port config(which means 1:1 lane to the CPU ports), that would guarantee full non-blocking BW and minimal port to port latency. In fact using PCIe 3.0 would only require 2 lanes(Pcie 3.0 is 1GB/s per lane per stream) for the full 8 ports to be saturated, thus potentially saving on the switch chip costs
i'm sorry if this sound naggy, but the hardware IS outdated and i can'0t understand the decision behind it other than penny-pinching, which should never happen on a device of this price (and i don't feel like promoting the 4xx and 5xx series, god knows what outdated hardware lurks there, at those levels i'd rather buy a state of the art Dell server -which will cost less most likely- and have an order of magnitued better everything inside)....
Certainly i wasn't expecting a CELERON on a 8 port big border security box...., and if it was a Celeron, why an outdated platform and not an ivy bridge one?
This thread was automatically locked due to age.
