IPsec ciphers, we need some more acronyms

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > IPsec ciphers, we need some more acronyms

Prescription Combs: Apr 20, 2005; 6

Most of my co-workers hate AnyConnect. I love setting it up, it's so easy...

It's always something hurf durf ipsec blah blah old ways. The thing that stinks about it, which I think may be changing, is the licensing. Especially if you want mobile phones to work on Anyconnect you need the 'for mobile' licence.

# ¿ Apr 17, 2015 22:15

Adbot: ADBOT LOVES YOU

# ¿ Apr 25, 2024 10:54

Prescription Combs: Apr 20, 2005; 6

MrMoo posted:

IKEv1 only permits tunneling of a single VLAN unless you are using a proprietary Cisco extension. IKEv2 permits multiple networks in leftsubnet= and rightsubnet= parameters. The typical workaround for this is to use multiple connections or to add L2TP above IPsec, i.e. a level 2 tunneling protocol above a single IPsec connection. On Linux this adds two additional processes I believe - xl2tpd and pppd. You could use GRE or any other IP-in-IP encapsulation protocol but IPsec + L2TP is the standard combination.

ChromeOS only supports OpenVPN or IPsec + L2TP (StrongSwan + xl2tpd).

iOS supports IPsec (Racoon), PPTP (GRE), IPsec + L2TP, and OpenVPN (via a special app). IKEv2 requires usage of Apple's enterprise system prep tool. A lot of the tools are new C++ rewrites by Apple so expect less feature parity and compatibility.

I think you're just used to awful linux IPSec implementations.

Why the hell does strongswan use 'leftsubnet' / 'rightsubnet'? What is the frame of reference? That is confusing as gently caress.

# ¿ May 2, 2015 07:51

Prescription Combs: Apr 20, 2005; 6

MrMoo posted:

Per RFC 4306 it appears the enhancement in IKEv2 is multiple traffic selectors (and narrowing) within a single child SA.

Ya but you can have multiple SAs to the same peer as long as you aren't using a terrible IPSec implementation device.

# ¿ May 3, 2015 05:55

Prescription Combs: Apr 20, 2005; 6

Tyren posted:

Side note - MS (and others) plan to deprecate SHA1 very soon (circa 2016). While we don't see SHA2 often, I'm guessing we'll see more alternatives to SHA1 in the near future.

https://www.schneier.com/blog/archives/2013/11/microsoft_retir.html

SHA1 has pretty much been deprecated by the popular browsers. Most certs have been re-keyed with SHA256.

Edit: I'm legitimately excited to see software solutions starting to utilize the AES/AES-NI extensions :monar:

Prescription Combs fucked around with this message at 01:22 on May 5, 2015

# ¿ May 5, 2015 01:18

Prescription Combs: Apr 20, 2005; 6

Are those numbers packets per second? If so that's fast as hell.

Edit: Oh I see, openSSL 'speed' benchmarks.

On my i5-2500K

code:

OpenSSL> speed -evp aes-128-cbc aes-256-cbc camellia-128-cbc
Doing aes-256 cbc for 3s on 16 size blocks: 17380960 aes-256 cbc's in 3.00s
Doing aes-256 cbc for 3s on 64 size blocks: 4710371 aes-256 cbc's in 3.00s
Doing aes-256 cbc for 3s on 256 size blocks: 1191494 aes-256 cbc's in 3.00s
Doing aes-256 cbc for 3s on 1024 size blocks: 299249 aes-256 cbc's in 3.00s
Doing aes-256 cbc for 3s on 8192 size blocks: 37441 aes-256 cbc's in 3.00s
Doing camellia-128 cbc for 3s on 16 size blocks: 22247823 camellia-128 cbc's in 3.00s
Doing camellia-128 cbc for 3s on 64 size blocks: 8415460 camellia-128 cbc's in 3.00s
Doing camellia-128 cbc for 3s on 256 size blocks: 2383579 camellia-128 cbc's in 3.00s
Doing camellia-128 cbc for 3s on 1024 size blocks: 615498 camellia-128 cbc's in 3.00s
Doing camellia-128 cbc for 3s on 8192 size blocks: 77781 camellia-128 cbc's in 2.99s
Doing aes-128-cbc for 3s on 16 size blocks: 146106823 aes-128-cbc's in 2.99s
Doing aes-128-cbc for 3s on 64 size blocks: 38846021 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 256 size blocks: 9868240 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 1024 size blocks: 2477179 aes-128-cbc's in 3.00s
Doing aes-128-cbc for 3s on 8192 size blocks: 309895 aes-128-cbc's in 3.00s
OpenSSL 1.0.1k-fips 8 Jan 2015
built on: Thu Mar 19 17:34:17 2015

The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
aes-256 cbc      92698.45k   100487.91k   101674.15k   102143.66k   102238.89k
camellia-128 cbc   118655.06k   179529.81k   203398.74k   210089.98k   213104.33k
aes-128-cbc     781842.53k   828715.11k   842089.81k   845543.77k   846219.95k

Neat.

Prescription Combs fucked around with this message at 04:27 on May 12, 2015

# ¿ May 12, 2015 03:57

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > IPsec ciphers, we need some more acronyms