The netmap project Luigi Rizzo - Università di Pisa

netmap - the fast packet I/O framework

*** Netmap tutorial at Hot Interconnects - Mountain View, 28 august 2014 ***


netmap / VALE is a framework for high speed packet I/O. Implemented as a kernel module for FreeBSD and Linux, it supports access to network cards (NICs), host stack, virtual ports (the "VALE" switch), and "netmap pipes". netmap can easily do line rate on 10G NICs (14.88 Mpps), moves over 20 Mpps on VALE ports, and over 100 Mpps on netmap pipes.
You can find more information in our papers, and have a look at the Usenix ATC'12 presentation here.

netmap/VALE can be used to build extremely fast traffic generators, monitors, software switches, network middleboxes, interconnect virtual machines or processes, do performance testing of high speed networking apps without the need for expensive hardware. We have full support for libpcap so most pcap clients can use it with no modifications.

netmap, VALE and netmap pipes are implemented as a single, non intrusive kernel module. Native netmap support is available for several NICs through slightly modified drivers; for all other NICs, we provide an emulated mode on top of standard drivers. netmap/VALE are part of standard FreeBSD distributions, and available in source format for Linux too.

Download/Install

The source repository is https://code.google.com/p/netmap.
Additional resouces (including demo OS images for FreeBSD and Linux) are available here:

(FreeBSD releases include the netmap source code). The repository includes instructions for building and installing the kernel module, device drivers and sample applications (a fast packet source/sink, a simple userspace bridge, etc.).
The images also have our improved QEMU/KVM code that supports extremely fast I/O over emulated e1000 devices.

Feedback/Mailing lists

At the moment we prefer to have discussion about netmap on the net@freebsd.org mailing list, or you can email me directly.

Applications

There are a number of systems that have netmap support:
  • FreeBSD has native netmap support since late 2011. The most up to date version is in FreeBSD HEAD
  • The Click modular router has native netmap support since early 2012. You can find the sources on Click on Github
  • 20120813 ipfw meets netmap, a netmap enabled version of ipfw and dummynet that reaches over 6 Mpps (FreeBSD and Linux)
  • Rump is a userspace version of the NetBSD kernel including a full network stack. It can use netmap as a backend

Other code and links

Some older code that might be of interest, although outdated
  • 20120430 openvswitch patches to make openvswitch work on FreeBSD/libpcap and get good performance in userspace.

Papers and presentations

Here are a few papers and submissions describing netmap and applications using it:

Note that we are actively developing and improving the code. The performance data below refers to the July 2011 version, and in many cases we have achieved some significant speedups.

More details

netmap uses a select()-able file descriptor to support blocking I/O, which makes it extremely easy to port applications using, say, raw sockets or libpcap to the netmap API.
netmap achieves extremely high speeds (up to 14.88 Mpps with a single interface using just one core at 900 Mhz); similarly, VALE can switch up to 20 Mpps per core on a virtual port. Other frameworks (e.g. DPDK, DNA) achieve similar speeds but lack the ease of use and portability.
On top of netmap we are bulding features and applications to replace parts of the existing network stack.
netmap is a very efficient framework for line-rate raw packet I/O from user space, which is capable to support 14.88Mpps on an ordinary PC and OS. Netmap integrates some known ideas into a novel, robust and easy to use framework that is available on FreeBSD and Linux without the need of special hardware or proprietary software.
With netmap, it takes as little as 60-65 clock cycles to move one packet between the user program and the wire. As an example, a single core running at 900 MHz can generate the 14.8 Mpps that saturate a 10 GigE interface. This is a 10-20x improvement over the use of a standard device driver.
The rest of this page gives a high level description of the project.

Architecture

 datafiles/a.jpg netmap uses some well known performance-boosting techniques, such as memory-mapping the card's packet buffers, I/O batching, and modeling the send and receive queues as circular buffers to match what the hardware implements. Unlike other systems, applications using netmap cannot crash the OS, because they run in user space and have no direct access to critical resources (device registers, kernel memory pointers, etc.). The programming model is extremely simple (circular rings of fixed size buffers), and applications use only standard system calls: non-blocking ioctl() to synchronize with the hardware, and poll()-able file descriptors to wait for packet receptions or transmissions on individual queues.

Performance

netmap can generate traffic at line rate (14.88Mpps) on a 10GigE link with just a single core running at 900Mhz. This equals to about 60-65 clock cycles per packet, and scales well with cores and clock frequency (with 4 cores, line rate is achieved at less than 450 MHz). Similar rates are reached on the receive side. In the graph below, the two top curves (green and red) indicate the performance of netmap on FreeBSD with 1 and 4 cores, respectively (Intel 82599 10Gbit card). The blue curve is the fastest available packet generator on Linux (pktgen, works entirely in the kernel), while the purple curve on the bottom shows the performance of a user-space generator on FreeBSD using udp sockets.
datafiles/plot-tx-clock-201109.jpg
netmap scales well to multicore systems: individual file descriptors can be associated to different cards or queues of a multi-queue card, and move packets between queues without the need to synchronize with each other.

Operation

netmap implements a special device, /dev/netmap, which is the gateway to switch one or more network cards to netmap mode, where the card's datapath is disconnected from the operating system. open("/dev/netmap") returns a file descriptor that can be used with ioctl(fd, NIOCREG, ...) to switch an interface to netmap mode. A subsequent mmap() exports to userspace a replica of the TX and RX rings of the card, and the actual packet buffers. Each "shadow" ring indicates the number of available buffers, the current read or write index, and the address and length each buffer (buffers have fixed size and are preallocated by the kernel).

Two ioctl() synchronize the state of the rings between kernel and userspace: ioctl(fd, NIOCRXSYNC) tells the kernel which buffers have been read by userspace, and informs userspace of any newly received packets. On the TX side, ioctl(fd, NIOCTXSYNC) tells the kernel about new packets to transmit, and reports to userspace how many free slots are available.
The file descriptor returned by open() can be used to poll() one or all queues of a card, so that blocking operation can be integrated seamlessly in existing programs.

Data movement

Receiving a packet is as simple as reading from the buffer in the mmapped region; eventually, ioctl(fd, NIOCRXSYNC) is used to release one or more buffers at once. Writing to the network requires to fill one or more buffers with data, set the lengths, and eventually invoke the ioctl(fd, NIOCTXSYNC) to issue the appropriate commands to the card.
The memory mapped region contains all rings and buffers of all cards in netmap mode, so it is trivial to implement packet forwarding between interfaces. Zero-copy operation is also possible, by simply writing the address of the received buffer into the in the transmit ring.

Talking to the host stack

In addition to the "hardware" rings, each card in netmap mode exposes two additional rings that connect to the host stack. Packets coming from the stack are put in an RX ring where they can be processed in the same way as those coming from the network. Similarly, packets written to the additional TX ring are passed up to the host stack when the ioctl(fd, NIOCTXSYNC) is invoked. Zero-copy bridging between the host stack and the card is then possible in the same way as between two cards. In terms of performance, using the card in netmap mode and bridging in software is often more efficient than using standard mode, because the driver uses simpler and faster code paths.

Device independence

Programs using netmap do not need any special library or knowledge of the inner details of the network controller. Not only the ring and buffer format is independent of the card itself, but any operation that requires to program the card is done entirely within the kernel.

Example of use

Below is a code snapshot to set a device in netmap mode and read packets from it. Macros are used to assign pointers because the shared memory region contains kernel virtual addresses.

  • struct netmap_if *nifp;
    struct nmreq req;
    int i, len;
    char *buf;
    
    fd = open("/dev/netmap", 0);
    strcpy(req.nr_name, "ix0"); // register the interface
    ioctl(fd, NIOCREG, &req); // offset of the structure
    mem = mmap(NULL, req.nr_memsize, PROT_READ|PROT_WRITE, 0, fd, 0);
    nifp = NETMAP_IF(mem, req.nr_offset);
    for (;;) {
    	struct pollfd x[1];
    	struct netmap_ring *ring = NETMAP_RX_RING(nifp, 0);
    
    	x[0].fd = fd;
    	x[0].events = POLLIN;
    	poll(x, 1, 1000);
    	for ( ; ring->avail > 0 ; ring->avail--) {
    		i = ring->cur;
    		buf = NETMAP_BUF(ring, i);
    		use_data(buf, ring->slot[i].len);
    		ring->cur = NETMAP_NEXT(ring, i);
    	}
    }
    

Detailed performance data

When talking about performance it is important to understand what are the relevant metrics. I won't enter into a long discussion here, please have a look at the papers for a more detailed discussion and up to date numbers.
In short:
  • if you only care about "system" overheads (i.e. the time to move a packet between the wire and the application) then the packet generator and receiver should be as simple as possible (maybe not even touching the data). The program pkt-gen that you find on the distribution implements exactly that -- outgoing packets are prepared once so they can be sent without regenerating all the times, incoming packets are counted but not read;
  • if you have a more complex application, it might need to send and receive at the same time, look at the payload, etc. There are two "bridge" applications in the distribution, transparently passing traffic between interfaces. The one called bridge uses the native netmap API and can do zero-copy forwarding; another one, called testpcap, uses a wrapper library that implements pcap calls on top of netmap. Among other things it does a copy of the packet on the outgoing link, plus it reads a timestamp (in the kernel) at each syscall because certain pcap clients expect packets to be timestamped.

Transmit and receive speed is shown in the previous section, and is relatively uninteresting as we go at line rate even with a severely underclocked CPU.

More interesting is what happens when you touch the data.

netmap can forward packets at line rate (14.88 Mpps) at 1.7 GHz without touching data, and slightly slower with full data copies. As a comparison, native packet forwarding using the in-kernel bridge does about 700Kpps on the same hardware. Though the comparison is a bit unfair because our bridge and testpcap don't do address lookups; however we have some real forwarding code (a modified version of openvswitch) that does almost 3Mpps using netmap.

Index