Performance Without Encryption
In the world of desktop CPUs, back in the days before AES-NI and similar acceleration hardware, encrypting data was a big, heavy, hairy deal. The difference between encrypting data in hardware compared to software (without hardware acceleration) was an order of magnitude or more. Similarly, routers used to suffer a performance penalty when running with encryption enabled, which led some speed fanatics to forgo encryption and run open. We wanted to see whether there was still a notable gain to be had in running unencrypted. For convenience, we’ll show the earlier encrypted results alongside their unencrypted counterparts.
The general results on our 2GB folder transfers seem to be inconsistent. Amped actually drops in performance without encryption, especially on the client to server test. Asus also nudges downward. The Nighthawk does a little give and take, but all three of these are more or less in the zone of test-to-test variances. The exception is the R6300, which takes a 10% to 20% jump. This is a strong enough leap to propel it into first place in our test. Could it be that the R6300 is in fact a far more capable speedster being held back by encryption processing? Or is it just luck and variable ambient test conditions? Let’s see...
Yup, chalk one up to luck and wobbly test factors. This time, the R6300 takes a good 10% to 20% dive without encryption. Ouch. Similarly, the Nighthawk takes a big hit on TCP performance. On the other hand, Amped and Asus both take broad jumps ahead without encryption. Feeling confused yet?
More of the same here with IxChariot’s TCP results. Three out of four average throughput results go down without encryption. Amped has the only router to show improvement—so much improvement that it went from last place to first.
Seriously, we did double-check these numbers. Asus and the Nighthawk show little UDP performance change when encryption is removed. Amped goes from exhibiting massive fluctuation to falling right in line (and actually winning the no-encryption test). The Netgear R6300 displays the exact opposite response, as if whatever madness had gripped the RTA15 with encryption suddenly jumped to the R6300 when running without encryption. It’s baffling.
Very quickly, let’s spin through the 2.4GHz encryption comparisons.
Keep in mind that we want to see lower numbers on our 2GB folder transfer. Instead, Amped and Asus actually get slower without encryption. The Nighthawk gets faster, but the R6300 gets much faster–enough to barely edge out the Nighthawk on this test.
And just when you might have started to think that encryption was somehow helping throughput, this time we see eliminating it provide the gains we expected in three out of four cases. Interestingly, these same three routers primarily saw their TCP numbers rise more than their UDP without encryption. Coincidence?
Once again, three routers decline without encryption, and yes, this negates our prior thought that there might be a correlation with TCP improvement. Only the R6300 bucks the trend and gains substantially as the Amped falls from its perch.
UDP results mimic those of TCP here. The only thing remarkable we see in this last encryption chart is a strangely uniform set of results across all four routers.
All in all, it seems safe to say that there is no statistically significant penalty to be paid in applying encryption to router traffic.