Menu
I do have some experience with DPDK but currently I'm reading many blogs about XDP. I am trying to compare both technologies and understand the differences between DPDK and XDP. This raises some questions. I hope someone can help me with the following questions:
Thank you in advance for your help!
927
What is eBPFeBPFThe Berkeley Packet Filter (BPF) is a technology used in certain computer operating systems for programs that need to, among other things, analyze network traffic. It provides a raw interface to data link layers, permitting raw link-layer packets to be sent and received.https://en.wikipedia.org › wiki › Berkeley_Packet_FilterBerkeley Packet Filter - Wikipedia XDP? eBPF is an extended version of the Berkeley Packet Filter (BPF). It is an abstract virtual machine (VM) that runs within the Linux kernel, much like the Java Virtual Machine (JVM) can run applications in a controlled environment.
XDP or eXpress Data Path provides a high performance, programmable network data path in the Linux kernel as part of the IO Visor Project. XDP provides bare metal packet processing at the lowest point in the software stack which makes it ideal for speed without compromising programmability.
With DPDK I can map workers to CPU cores and isolate CPU cores which will be used by DPDK. In case of eBPF / XDP, which CPU cores are used?
Answer: XDP with eBPF runs in kernel space, unlike DPDK user space.
Are all available CPU cores used?
Answer: Yes, but normally irqbalance or interrupt pinning will put the RX queue of the port on to specific core.
Would it be possible to isolate CPU cores meant for eBPF / XDP programs?
Answer: You are referring to KERNEL_CMD_LINE option isol, the understanding is incorrect. As mentioned above you can pin the interrupt of RX queue forcing to run eBPF XDP on that core.
When I test the throughput from a DPDK application, I'm able to check whether ring buffers (mempools) are full so packets will be lost. But how can I check whether an eBPF / XDP program causes packet drops because the throughput is too high?
Answer: you have use a mix of NIC and eBPF counters to achieve the same
I assume when an eBPF / XDP program takes too much time to process a packet, eventually, you will see packet drops? (especially when sending 64B packets on a high rate to find the maximum number of packets that can be send)
Answer: not necessary true, best performance of XDP is with zero-copy Driver to user sapce. Running application thread on a separate core gives an almost comparable performance as DPDK (tested with 2 * 10Gbps - 95% of DPDK performance).
130
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With
