Teledyne LeCroy：800G高速以太网一站式自动化测试解决方案（网络打流测试+物理层加压干扰+协议分析）

devtools/2024/10/20 3:07:45/

LinkExpert一站式测试解决方案

LinkExpert 是一款软件应用程序，可对Teledyne LeCroy的协议分析仪和训练器进行自动化硬件控制和管理。除了作为合规性、一致性和验证测试的便捷接口外，它还能轻松地将这些测试添加到自动回归测试流程中。

现在，新的 LinkExpert集成可实现800G以太网连接的自动协商和链路训练 (AN/LT)。该集成使工程师能够轻松利用Xena以太网流量发生器的AN/LT测试功能和SierraNet协议分析仪。

该解决方案为800G以太网提供了一站式集成自动化测试，可将端口的物理层性能测试与分析和Layer23层流量压力测试无缝衔接，提供了跨以太网协议层级的Layer123完整验证，提供了端口一致性验证、端口性能压力测试、协议可视化分析一站式测试服务。

端口一致性测试

可作为Golden Sample，对接DUT进行端口一致性测试校验

端口兼容性测试

可调节Pre-Coding，PPM Adjustment将端口进行劣化，以模拟特定DUT的性能

端口性能测试

线速和超线速的极限压力测试，微突发场景等

真实业务压力测试

RDMA,PFC,ECQCN,ECN,QP业务模型仿真

速率支持

OSFP接口和QSFP-DD原生接口

56G PAM4- 50G/100G/200G/400G

112G PAM4 - 100G/200G/400G/800G

功能举例

800G ANLT测试界面

  • 作为端节点，进行以太网协议的握手适配，对接
  • 作为监控节点，串行接入在链路中，进行物理层链路和数据链路层的报文捕获和分析，即Layer1，Layer2，Layer3等全覆盖
  • 负面压力环境仿真，如FEC Error Injection测试验证，Link Flap验证，Packet Loss验证 

硬件构成：Xena Freya Z800+SierraNet M1288

<a class=800G以太网测试与分析仪表" height="625" src="https://i-blog.csdnimg.cn/direct/ab9eee1e6efa4f35b2768781fe210a48.png" width="870" />

Xena Ethernet Tester

  • 10M-800G以太网端口速率支持

  • 支持全线速的以太网流量生成和收发统计

  • Layer1、2、3层测试和自动化支持

  • 固定式1U桌面便携型和模块化4U机架式

  • 多种不同速率测试板卡和报文损伤仿真板卡

  • 高可扩展性、多业务槽位后期扩展支持

  • RFC2544，RFC2889，RFC3918，Y.1564等基准性能测试套件支持

SierraNet Protocol Analyzer/Jammer

  • 10G至800G以太网和FC光纤通道支持
  • 链路可视化分析和干扰注入
  • 高阶多态触发和过滤功能
  • 全栈（Layer1-7）线速捕获分析和错误注入
  • Infusion Jammer干扰注入支持
  • 业内唯一的L1层协议捕获和ANLT分析仪
  • 支持脚本引擎VSE用于自动化后期分析

仪表典型应用场景

XENA以太网测试仪

XENA 800G Freya以太网测试仪

  • 灵活生成以太网数据流量流进行收发包测试，以评估网络设备（交换机、路由器、防火墙）在压力条件下的运行情况
  • 模拟多个用户或设备同时访问网络，了解网络负载均衡性能
  • 测量网络设备或链路可处理的最大数据传输速率，以测试网络吞吐量
  • 利用 Chimera 网络仿真的各种故障（延迟/抖动/丢包）在实验室中模拟真实生产环境
  • 创建具有不同优先级的不同类型流量，以测试 PFC、QoS机制
  • 模拟分布式拒绝服务 (DDoS) 等攻击，或创建模仿恶意行为的流量模式
  • 有助于在受控环境中创建逼真的网络条件，作为网络模拟和建模的一部分

SierraNet 协议分析仪

力科800G M1288协议分析仪

  • 监控和捕获所有网络流量，以便进行详细分析
  • 加快故障排除速度并识别故障的链接
  • 验证数据包是否符合所需的协议标准和规范
  • 检测和分析互操作性问题，确保不同的网络设备和系统正常工作
  • 以太网和光纤通道的质量保证和验证
  • 产品基准和竞争分析
  • 可全面查看整个网络的所有通信情况
  • 检测和分析安全威胁，如未经授权的访问和可疑的流量模式
  • 监控带宽使用情况，识别大流量用户和应用程序
  • 故障排除和现场支持，以诊断故障链路、互操作性问题或网络性能问题

典型待测物

网卡，线材类

AI HPC高速网卡，SmartNIC，DPU等，用以服务器之间的连接或服务器与交换机的连接

AEC，ACC 线材，验证其ANLT的生效情况

交换机类

端口性能验证

业务模型搭建测试极限并发性能

负面压力模拟，将真实的用户场景搬进实验室进行测试验证，并可定量精准，可重复的测试

规模组网验证

RDMA ，QP业务模型搭建

友好的API二次开发接口

signal_integrity_hist_plot.py

################################################################
#
#              SIGNAL INTEGRITY HISTOGRAM PLOTTING
#
# What this script example does:
# 1. Connect to a tester
# 2. Reserve a port. 
# 3. Collecting SIV data from all 8 lanes
# 4. Plot the live histogram 
#
################################################################import asynciofrom xoa_driver import testers
from xoa_driver import modules
from xoa_driver import ports
from xoa_driver import enums
from xoa_driver import utils
from xoa_driver.hlfuncs import mgmt
from xoa_driver.misc import Hex
import logging
import math
from typing import Listimport matplotlib.pyplot as plt
from collections import deque#---------------------------
# GLOBAL PARAMS
#---------------------------
CHASSIS_IP = "10.165.136.60"
USERNAME = "xoa"
PORT = "6/0"
FIGURE_TITLE = "Z800 Freya Signal Integrity Histogram Plot"
DENSITY = 1 # how many batches of siv data to show on the plot. A higher density means more data and slower plotting.
LANES = [0,1,2,3,4,5,6,7] # select lanes to display, ranging from 0 to 7
PLOTTING_INTERVAL = 1 # plots refreshed every n second
PLOTTING_DURATION = 120 # number of seconds for plottingasync def siv_plot(chassis: str, username: str, port_str: str,figure_title: str,density: int,plotting_interval: int,plotting_duration: int,lanes: List[int],):# configure basic loggerlogging.basicConfig(format="%(asctime)s  %(message)s",level=logging.DEBUG,handlers=[logging.FileHandler(filename="siv_plot.log", mode="a"),logging.StreamHandler()])# disable matplotlib.font_manager logginglogging.getLogger('matplotlib.font_manager').disabled = True# remove duplicates and sort listlanes = list(set(lanes))lanes.sort()async with testers.L23Tester(host=chassis, username=username, password="xena", port=22606, enable_logging=False) as tester:logging.info(f"#####################################################################")logging.info(f"Chassis:                 {chassis}")logging.info(f"Username:                {username}")logging.info(f"Port:                    {port_str}")logging.info(f"Lanes:                   {lanes}")logging.info(f"Figure Title:            {figure_title}")logging.info(f" Data Density:            {density}")logging.info(f" Plot Refresh Interval:   {plotting_interval} s")logging.info(f" Plot Duration:           {plotting_duration} s")logging.info(f"#####################################################################")# Access module on the tester_mid = int(port_str.split("/")[0])_pid = int(port_str.split("/")[1])module_obj = tester.modules.obtain(_mid)if not isinstance(module_obj, modules.Z800FreyaModule):logging.info(f"Module {_mid} is not Z800 Freya module. Abort.")return Noneport_obj = module_obj.ports.obtain(_pid)await mgmt.free_module(module=module_obj, should_free_ports=False)await mgmt.reserve_port(port_obj)resp = await port_obj.capabilities.get()max_serdes = resp.serdes_countserdes_cnt_to_show = len(lanes)if max(lanes) > 7:logging.warning(f"Exceed max serdes index. Abort.")return Noneif serdes_cnt_to_show > max_serdes:logging.warning(f"Exceed max serdes count. Abort.")return Noneif serdes_cnt_to_show == 0:logging.warning(f"Nothing to show Abort.")return None# figure configplt.ion()fig = plt.figure(constrained_layout=True)fig.suptitle(f"{figure_title}\nChassis {chassis}, Port {port_str}, L={lanes}, D={density}")# grid specif serdes_cnt_to_show == 1:gs = fig.add_gridspec(nrows=1, ncols=1)if serdes_cnt_to_show > 1:gs = fig.add_gridspec(nrows=math.ceil(serdes_cnt_to_show/2), ncols=2)# add subplotssiv_subplots = []for i in range(serdes_cnt_to_show):siv_subplots.append(fig.add_subplot(gs[i%gs.nrows, int(i/gs.nrows)]))# data dequeue for each serdes lane. queue depth = density*2000INT_CNT_PER_DATA = 2000data_queue = []for _ in range(serdes_cnt_to_show):data_queue.append(deque((), maxlen=density*INT_CNT_PER_DATA))# set x and y label for each subplotfor i in range(serdes_cnt_to_show):siv_subplots[i].set(xlabel=f"Value", ylabel=f"Lane {lanes[i]}")# group control commands for each serdes lane together to later send it as a command group.control_cmd_group = []for i in range(serdes_cnt_to_show):control_cmd_group.append(port_obj.l1.serdes[lanes[i]].medium.siv.control.set(opcode=enums.Layer1Opcode.START_SCAN))# get commands for each serdes lane together to later send it as a command group.get_cmd_group = []for i in range(serdes_cnt_to_show):get_cmd_group.append(port_obj.l1.serdes[lanes[i]].medium.siv.data.get())resp_group = ()plot_count = math.ceil(plotting_duration/plotting_interval)for _ in range(plot_count):await utils.apply(*control_cmd_group)while True:# get responses from all lanesresp_group = await utils.apply(*get_cmd_group)result_flags = [x.result for x in resp_group]if 0 in result_flags:# if not all lanes are ready in data, query again.continueelse:for i in range(serdes_cnt_to_show):siv_raw_levels = resp_group[i].value[0:12]siv_raw_values = resp_group[i].value[12:]# convert from 12 raw bytes into 6 signed intsiv_int_levels = []for x in zip(siv_raw_levels[0::2], siv_raw_levels[1::2]):siv_int_levels.append(int.from_bytes(bytes(x), byteorder='big', signed=True))# Please note: only the first slicer data is used here.# convert from 4000 bytes into 2000 signed intsiv_int_values = []for x in zip(siv_raw_values[0::2], siv_raw_values[1::2]):siv_int_values.append(int.from_bytes(bytes(x), byteorder='big', signed=True))# put value data in queuedata_queue[i].extend(tuple(siv_int_values))# siv data ranges from -64 to 63, thus 128 bins in total.siv_subplots[i].cla()siv_subplots[i].relim()siv_subplots[i].autoscale_view()siv_subplots[i].set(xlabel=f"Value", ylabel=f"Lane {lanes[i]}")siv_subplots[i].hist(x=[*data_queue[i]], bins=128, range=(-64, 63), density=False, color="blue", orientation="horizontal")# levels contains 6 values, 4 average pam4 levels and 2 slicers, (<p1> <p2> <p3> <m1> <m2> <m3>)# add base slicer (this is always at 0)y = 0siv_subplots[i].axhline(y, color='black', linestyle='-', linewidth=0.5)siv_subplots[i].text(siv_subplots[i].get_xlim()[1] + 0.1, y, f'base={y}', fontsize="small")# add upper slicer <p2>y = siv_int_levels[1]siv_subplots[i].axhline(y, color='green', linestyle='dashed', linewidth=0.5)siv_subplots[i].text(siv_subplots[i].get_xlim()[1] + 0.1, y, f'slicer={y}', fontsize="small")# add lower slicer <m2>y = siv_int_levels[4]siv_subplots[i].axhline(y, color='green', linestyle='dashed', linewidth=0.5)siv_subplots[i].text(siv_subplots[i].get_xlim()[1] + 0.1, y, f'slicer={y}', fontsize="small")# add average level 3 <p3>y = siv_int_levels[2]siv_subplots[i].axhline(y, color='black', linestyle='dashed', linewidth=0.1)siv_subplots[i].text(siv_subplots[i].get_xlim()[1] + 0.1, y, f'level3={y}', fontsize="small")# add average level 2 <p1>y = siv_int_levels[0]siv_subplots[i].axhline(y, color='black', linestyle='dashed', linewidth=0.1)siv_subplots[i].text(siv_subplots[i].get_xlim()[1] + 0.1, y, f'level2={y}', fontsize="small")# add average level 1 <m3>y = siv_int_levels[5]siv_subplots[i].axhline(y, color='black', linestyle='dashed', linewidth=0.1)siv_subplots[i].text(siv_subplots[i].get_xlim()[1] + 0.1, y, f'level1={y}', fontsize="small")# add average level 0 <m1>y = siv_int_levels[3]siv_subplots[i].axhline(y, color='black', linestyle='dashed', linewidth=0.1)siv_subplots[i].text(siv_subplots[i].get_xlim()[1] + 0.1, y, f'level0={y}', fontsize="small")plt.show()plt.pause(plotting_interval)breakawait mgmt.free_port(port_obj)logging.info(f"Bye!")async def main():stop_event = asyncio.Event()try:await siv_plot(chassis=CHASSIS_IP,username=USERNAME,port_str=PORT,figure_title=FIGURE_TITLE,density=DENSITY,plotting_interval=PLOTTING_INTERVAL,plotting_duration = PLOTTING_DURATION,lanes=LANES)except KeyboardInterrupt:stop_event.set()if __name__ == "__main__":asyncio.run(main())

fec_error_dist_plot.py 

################################################################
#
#                   PRE-FEC ERROR DIST PLOT
#
# What this script example does:
# 1. Connect to a tester
# 2. Reserve a all ports on a module
# 3. Set the port FEC mode on
# 4. Clear FEC stats
# 5. Query FEC Blocks (symbol error) and FEC stats
# 6. Plot the Pre-FEC error distribution for all ports
#
################################################################import asynciofrom xoa_driver import testers
from xoa_driver import modules
from xoa_driver import ports
from xoa_driver import enums
from xoa_driver import utils
from xoa_driver.hlfuncs import mgmt
from xoa_driver.misc import Hex
import logging
import mathimport matplotlib.pyplot as plt
import numpy as np
from collections import deque#---------------------------
# GLOBAL PARAMS
#---------------------------
CHASSIS_IP = "10.165.136.66"
USERNAME = "xoa"
MODULE = "4"
FIGURE_TITLE = "Pre-FEC Error Distribution Plot (log10)"
PLOTTING_INTERVAL = 1 # plots refreshed every n second
PLOTTING_DURATION = 120 # number of seconds for plotting# Enable the FEC mode you want
FEC_MODE = enums.FECMode.ON # either RS FEC KR or KP. Determined by the port automatically
# FEC_MODE = enums.FECMode.FC_FEC
# FEC_MODE = enums.FECMode.RS_FEC_INT#---------------------------
# pre_fec_error_dist_plot
#---------------------------
async def pre_fec_error_dist_plot(chassis: str,username: str,module_str: str,figure_title: str,plotting_interval: int,plotting_duration: int,fec_mode: enums.FECMode):# configure basic loggerlogging.basicConfig(format="%(asctime)s  %(message)s",level=logging.DEBUG,handlers=[logging.FileHandler(filename="fec_plot.log", mode="a"),logging.StreamHandler()])# disable matplotlib.font_manager logginglogging.getLogger('matplotlib.font_manager').disabled = True# Establish connection to a Valkyrie tester using Python context manager# The connection will be automatically terminated when it is out of the blockasync with testers.L23Tester(host=chassis, username=username, password="xena", port=22606, enable_logging=False) as tester:logging.info(f"#####################################################################")logging.info(f"Chassis:                 {chassis}")logging.info(f"Username:                {username}")logging.info(f"Port:                    {module_str}")logging.info(f"Figure Title:            {figure_title}")logging.info(f" Plot Refresh Interval:   {plotting_interval} s")logging.info(f" Plot Duration:           {plotting_duration} s")logging.info(f"#####################################################################")# Access module on the tester_mid = int(module_str)module_obj = tester.modules.obtain(_mid)if isinstance(module_obj, modules.E100ChimeraModule):logging.info(f"FEC not supported on E100 Chimera modules")return Noneif isinstance(module_obj, modules.Z10OdinModule):logging.info(f"FEC not supported on Z10 Odin modules")return None # reserve all ports on a moduleport_objs = [x for x in module_obj.ports]port_cnt = len(port_objs)# Forcibly reserve the portawait mgmt.free_module(module=module_obj, should_free_ports=False)resp = await module_obj.revision.get()module_module_name = resp.revisionfor p in port_objs:await mgmt.reserve_port(p)await asyncio.sleep(1)# figure configplt.ion()fig = plt.figure(constrained_layout=True)fig.suptitle(f"{figure_title}\nChassis {chassis}, Module {module_str}, {module_module_name}")# grid specif port_cnt == 1:gs = fig.add_gridspec(nrows=1, ncols=1)if port_cnt > 1:gs = fig.add_gridspec(nrows=math.ceil(port_cnt/2), ncols=2)# add subplotspre_fec_subplots = []for i in range(port_cnt):pre_fec_subplots.append(fig.add_subplot(gs[i%gs.nrows, int(i/gs.nrows)]))# set x and y label for each subplotfor i in range(port_cnt):pre_fec_subplots[i].set(xlabel=f"Symbol Errors", ylabel=f"FEC Codewords ({module_str}/{i}) (log10)")# set FEC mode onlogging.info(f"Set FEC Mode = {fec_mode.name}")for p in port_objs:await p.fec_mode.set(mode=fec_mode)# clear FEC counterlogging.info(f"Clear FEC counter")for p in port_objs:await p.pcs_pma.rx.clear.set()# query FEC Totals and Pre-FEC Error Distributionplot_count = math.ceil(plotting_duration/plotting_interval)pre_fec_error_dist_data = [[0]*17]*port_cntprint(len(pre_fec_subplots))for _ in range(plot_count):logging.info(f"PRE-FEC ERROR DISTRIBUTION")for i in range(port_cnt):port_obj = port_objs[i]logging.info(f"Port {port_obj.kind.module_id}/{port_obj.kind.port_id}")# await port_obj.pcs_pma.rx.clear.set()_total_status, _fec_status = await utils.apply(port_obj.pcs_pma.rx.total_status.get(),port_obj.pcs_pma.rx.fec_status.get())n = _fec_status.data_count - 2for j in range(n):logging.info(f"  FEC Blocks (Symbol Errors = {j}): {_fec_status.stats[j]}")logging.info(f"  FEC Blocks (Symbol Errors > {n-1}): {_fec_status.stats[n]}")x_axis = [str(x) for x in range(n)]x_axis.append(f"> {n-1}")color_array = ['y']*(n-1)color_array.insert(0, 'g')color_array.append('r')pre_fec_error_dist_data[i] = [x + y for x, y in zip(pre_fec_error_dist_data[i], _fec_status.stats[0:n+1])]pre_fec_ber_str = ""if _total_status.total_pre_fec_ber == 0:pre_fec_ber_str = f"Pre-FEC BER = 0"else:pre_fec_ber_str = f"Pre-FEC BER = {abs(1/_total_status.total_pre_fec_ber)}"pre_fec_subplots[i].cla()pre_fec_subplots[i].relim()pre_fec_subplots[i].autoscale_view()pre_fec_subplots[i].set(xlabel=f"Symbol Errors", ylabel=f"FEC Codewords ({module_str}/{i})")pre_fec_error_dist_data_log10 = []for x in pre_fec_error_dist_data[i]:if x > 0:pre_fec_error_dist_data_log10.append(np.log10(x))else:pre_fec_error_dist_data_log10.append(0)# print(pre_fec_error_dist_data_log10)tmp = pre_fec_subplots[i].bar(x=x_axis, height=pre_fec_error_dist_data_log10, color=color_array,)pre_fec_subplots[i].bar_label(container=tmp, fmt='%.1f')x0, xmax = pre_fec_subplots[i].get_xbound()y0, ymax = pre_fec_subplots[i].get_ybound()pre_fec_subplots[i].text((x0+xmax)*0.7, (y0+ymax)*0.9, pre_fec_ber_str, fontsize="small")plt.show()logging.info(f"Clear FEC counter")for p in port_objs:await p.pcs_pma.rx.clear.set()plt.pause(plotting_interval)async def main():stop_event = asyncio.Event()try:await pre_fec_error_dist_plot(chassis=CHASSIS_IP,username=USERNAME,module_str=MODULE,figure_title=FIGURE_TITLE,plotting_interval=PLOTTING_INTERVAL,plotting_duration=PLOTTING_DURATION,fec_mode=FEC_MODE)except KeyboardInterrupt:stop_event.set()if __name__ == "__main__":asyncio.run(main())


http://www.ppmy.cn/devtools/127177.html

相关文章

Chapter 2 - 7. Understanding Congestion in Fibre Channel Fabrics

Chapter 2 - 7. Understanding Congestion in Fibre Channel Fabrics

The Culprit Host In this case, is Host-1 really a culprit? The answer depends on whether to look from Host-1’s perspective or fabric’s perspective. If looking from Host-1’s perspective, calling it a culprit may not be entirely correct because it’s no…
阅读更多...
MEMC功能详解

MEMC功能详解

文章目录 MEMC的工作原理&#xff1a;优点&#xff1a;缺点&#xff1a;适用场景&#xff1a;1. Deblur&#xff08;去模糊&#xff09;2. Dejudder&#xff08;去抖动&#xff09;总结两者区别&#xff1a; MEMC&#xff08;Motion Estimation and Motion Compensation&#x…
阅读更多...
【数字图像处理】第5章 图像空域增强方法

【数字图像处理】第5章 图像空域增强方法

上理考研周导师的哔哩哔哩频道 我在频道里讲课哦 目录 5.1 图像噪声 相关概念 ①图像噪声的产生 ② 图像噪声分类 ③ 图像噪声特点 5.2 图像增强方法分类 ①图像增强概念 ②图像增强目的 ③图像增强技术方法: 5.3 基于灰度变换的图像增强 1. 概述: 2. 灰度变换…
阅读更多...
解决uniapp中全局代码实现微信小程序的分享到朋友圈和分享到群聊功能

解决uniapp中全局代码实现微信小程序的分享到朋友圈和分享到群聊功能

如果在用uniapp的Vue3开发微信小程序时&#xff0c;需要实现将微信小程序可以分享到群和朋友圈功能&#xff0c;如果没有这个功能&#xff0c;应该怎么实现所有页面都能分享呢&#xff1f; 那么介绍一个方法&#xff0c;可以实现全局加上代码后&#xff0c;小程序的每个页面都能…
阅读更多...
Java--集合框架

Java--集合框架

目录 前言 一 、Collection接口 1.1 Collection常用方法 1.1.1 添加 1.1.2 判断 1.1.3 删除 1.1.4 其它 1.1.5 iterator() 注意 1.2 子接口&#xff1a;List 1.2.1 List接口常用的方法 1.2.2 List实现不同类的对比 1.3 子接口&#xff1a;Set 1.3.1 Set及其实现类…
阅读更多...
scala 类的继承

scala 类的继承

继承的定义 idea实例 语法 重写 重写&#xff1a;在子类中重新定义父类的同名方法 idea实例 多态 多态&#xff1a;传入的对象不同&#xff0c;调用的方法的效果就不同&#xff01; 原理&#xff1a;参数是父类类型 idea实例 构造器
阅读更多...
【基于docker的深度学习训练环境】关键步骤记录

【基于docker的深度学习训练环境】关键步骤记录

最近给公司搭建了一个小型的深度学习环境&#xff0c;实现了多人通过SSH对GPU资源的利用&#xff0c;下面对一些关键架构和易用性部分进行记录。 一、整体软硬件框架 1、硬件配置&#xff0c;采用的双GPU的方案&#xff0c;两块消费级显卡。 2、应用层架构 宿主机系统为ubunt…
阅读更多...
Android OpenGL天空盒

Android OpenGL天空盒

在我们的项目学习过程中&#xff0c;我们从一片漆黑的虚空开始构建。为了给这个世界增添一些色彩&#xff0c;我们加入了三个粒子喷泉&#xff0c;但即便如此&#xff0c;我们的世界依然大部分被黑暗和虚无所笼罩。这些喷泉仿佛悬浮在无尽的黑暗之中&#xff0c;没有边界&#…
阅读更多...
最新文章

Copyright @ 2022~2023