| blob | ed7dba88643370c08a900247f3ea0dd92bd7623c |
1 Work that relates to netsniff-ng and how we differ from it:
2 ///////////////////////////////////////////////////////////
4 ntop
5 * W: http://www.ntop.org/
7 The ntop projects offers zero-copy for network packets. Is this approach
8 significantly different from the already built-in from the Linux kernel?
9 High likely not. In both cases packets are memory mapped between both address
10 spaces. The biggest difference is that you get this for free, without
11 modifying your kernel with netsniff-ng since it uses the kernel's RX_RING
12 and TX_RING functionality. Unfortunately this is not really mentioned on the
13 ntop's website. Surely for promotional reasons. For many years the ntop
14 projects lives on next to the Linux kernel, attempts have been made to
15 integrate it [1] but discussions got stuck and both sides seem to have no
16 interest in it anymore, e.g. [2]. Therefore, if you want to use ntop, you are
17 dependent on ntop's modified drivers that are maintained out of the Linux
18 kernel's mainline tree. Thus, this will not provide you with the latest
19 improvements. Also, the Linux kernel's PF_PACKET is maintained by a much bigger
20 audience, probably better reviewed and optimized. Therefore, also we decided
21 to go with the Linux kernel's variant. So to keep it short: both approaches
22 are zero-copy, both have similar performance (if someone tells you something
23 different, he would lie due to their technical similarities) and we are using
24 the kernel's built-in variant to reach a broader audience.
26 [1] http://lists.openwall.net/netdev/2009/10/14/37
27 [2] http://www.spinics.net/lists/netfilter-devel/msg20212.html
29 tcpdump
30 * W: http://www.tcpdump.org/
32 tcpdump is probably the oldest and most famous packet analyzer. It is based on
33 libpcap and in fact the MIT team that maintains tcpdump also maintains libpcap.
34 It has been ported to much more architectures and operating systems than
35 netsniff-ng. However, we don't aim to rebuild or clone tcpdump. We rather focus
36 on achieving a higher capturing speed by carefully tuning and optimizing our
37 code. That said doesn't mean that tcpdump people do not take care of it. It
38 just means that we don't have additional layers of abstractions for being as
39 portable as possible. This already gives us a smaller code footprint. Also, on
40 default we perform some system tuning such as remapping the NIC's IRQ affinity
41 that tcpdump probably would never do due to its generic nature. By generic, we
42 mean to serve as many different user groups as possible. We rather aim at
43 serving users for high-speed needs. By that, they have less manual work to do
44 since it's already performed in the background. Next to this, we also aim at
45 being a useful networking toolkit rather than only an analyzer. So many other
46 tools are provided such as trafgen for traffic generation.
48 Wireshark/tshark
49 * W: http://www.wireshark.org/
51 Probably we could tell you the same as in the previous section. I guess it is
52 safe to say that Wireshark might have the best protocol dissector out there.
53 However, this is not a free lunch. You pay for it with a performance
54 degradation, which is quite expensive. It is also based on libpcap (we are not)
55 and it comes with a graphical user interface, whereas we rather aim at being
56 used somewhere on a server or middle-box site where you only have access to a
57 shell, for instance. Again, offline analysis of /large/ pcap files might even
58 let it hang for a long time. Here netsniff-ng has a better performance also in
59 capturing pcaps. Again, we furthermore aim at being a toolkit rather than only
60 an analyzer.
62 libpcap
63 * W: http://www.tcpdump.org/
65 Price question: why don't you rely on libpcap? The answer is quite simple. We
66 started developing netsniff-ng with its zero-copy capabilities back in 2009
67 when libpcap was still doing packet copies between address spaces. Since the
68 API to the Linux kernel was quite simple, we felt more comfortable using it
69 directly and bypassing this additional layer of libpcap code. Today we feel
70 good about this decision, because since the TX_RING functionality was added to
71 the Linux kernel we have a clean integration of both, RX_RING and TX_RING.
72 libpcap on the other hand was designed for capturing and not for transmission
73 of network packets. Therefore, it only uses RX_RING on systems where it's
74 available but no TX_RING functionality. This would have resulted in a mess in
75 our code. Additionally, with netsniff-ng, one is able to a more fine grained
76 tuning of those rings. Why didn't you wrap netsniff-ng around your own library
77 just like tcpdump and libpcap? Because we are ignorant. If you design a library
78 than you have to design it well right at the beginning. A library would be a
79 crappy one if it changes its API ever. Or, if it changes its API, than it has
80 to keep its old one for the sake of being backwards compatible. Otherwise no
81 trust in its user or developer base can be achieved. Further, by keeping this
82 long tail of deprecated functions you will become a code bloat over time. We
83 wanted to keep this freedom of large-scale refactoring our code and not having
84 to maintain a stable API to the outer world. This is the whole story behind it.
85 If you desperately need our internal functionality, you still can feel free to
86 copy our code as long as your derived code complies with the GPL version 2.0.
87 So no need to whine. ;-)