EoRPR技术白皮书

合集下载

华为Edge OTN解决方案技术白皮书V1.1说明书

保留一切权利。

非经本公司书面许可，任何单位和个人不得擅自摘抄、复制本文档内容的部分或全部，并不得以任何形式传播。

商标声明和其他华为商标均为华为技术有限公司的商标。

本文档提及的其他所有商标或注册商标，由各自的所有人拥有。

注意您购买的产品、服务或特性等应受华为公司商业合同和条款的约束，本文档中描述的全部或部分产品、服务或特性可能不在您的购买或使用范围之内。

除非合同另有约定，华为公司对本文档内容不做任何明示或默示的声明或保证。

由于产品版本升级或其他原因，本文档内容会不定期进行更新。

除非另有约定，本文档仅作为使用指导，本文档中的所有陈述、信息和建议不构成任何明示或暗示的担保。

华为技术有限公司地址：深圳市龙岗区坂田华为总部办公楼邮编：518129网址：https://客户服务邮箱：******************客户服务电话：4008302118文档版本V1.1 (2021-03-20) 版权所有© 华为技术有限公司第 2 共29目录1 FMEC网络融合的趋势与挑战 (4)1.1 品质业务需求快速增长 (4)1.2 融合业务成为趋势 (6)1.3 FMEC网络建设面临的挑战 (7)1.4 总结 (8)2 Edge OTN方案是FMEC融合建网的最佳选择 (9)2.1 Edge OTN架构 (9)2.2 基于价值区域的精准布局建网方式 (10)2.3 总结 (12)3 Edge OTN关键技术 (13)3.1 环境适应性增强技术 (13)3.2 灰光彩光混合传输 (13)3.3 Liquid OTN技术 (14)3.4 高精度时间同步 (15)4 华为Edge OTN解决方案 (16)4.1 精准规划工具 (16)4.2 全场景部署能力 (17)4.3 光层电层创新方案 (19)4.3.1 极简光层 (19)4.3.2 X+Y分布式电层 (20)4.3.3 创新线路速率 (22)4.3.4 平滑演进典型方案 (22)4.4 智慧运维 (23)4.4.1 NCE智能管控 (23)4.4.2 光层自动调测 (24)4.4.3 智能光纤管理 (24)4.4.4 智慧光性能管理 (24)5 总结 (26)A 缩略语 (27)1 FMEC网络融合的趋势与挑战1.1 品质业务需求快速增长宽带成为人们生产、生活必需的基础资源。

ECO云开放平台白皮书

乙E C O云开放平台2.0为教育提供数据大脑公司介绍COMPANY INTRODUCTION夭闻数媒科技（北京）有限公司成立千2010年，是中南传媒（证券代码：601098)旗下子公司，注册资本3.2亿元，是国有控股的国家高新技术企业、中关村高新技术企业。

从电子书包平板互动教学的开拓者，成长为国内领先的互联网教音平台商，天闻数媒不断创新，以“建立生态“吸聚合作伙伴，以“实现智能”作为研发目标，以满足人本”提升用户体验，致力千为中国教育信息化提供领先的全域全流程解决方案，向K-12领域提供教、学、管、评、测系列产品及服务。

基千夭闻ECO云开放平台，天闻数媒致力打造教育行业新生态。

公司智慧教育核心产品簇包括天闻ECO云开放平台AiCloud教音云、AiSchool智慧校园AiClass云课堂三通两平台等。

截至2018年末，天闻数媒智慧教育产品已覆盖全国25个省（直辖市、自治区）、135个市、496个区（县），服务约8000所学校，1800万用户。

其中，AiClass云课堂服务约400万用户，AiCloud教音云覆盖2个市、近百个区（县），校比邻APP融合用户约400万人。

引领教音文化”走出去”，天闻数媒以教音咨询、教材编印、教育培训、教育信息化建设为核心的海外业务已走进欧洲、非洲、东南亚等十数个国家。

夭闻数媒还提供优质的数字公共文化和融媒体服务，并设立了智能媒体服务高科技企业——夭江智媒科技（北京）有限公司。

夭闻数媒秉承“使知识更丰富、让文化更精彩”的公司愿景，坚持＇以客户为中心、以奋斗者为本、坚持艰苦创业”的公司核心价值观，持续为客户提供全方位、高质量的建设、运营、运维等服务。

平台介绍ECO云开放平台是国内基础教音行业首个商用开放平台，致力千消除烟囡“应用和数据孤岛，真正实现跨系统的教育大数据采集和共享。

平台已接入第三方厂商30+,接入应用60+,开放微服务100+,覆盖学校5000+, 累积用户量超过400万。

eth白皮书

eth白皮书
以太坊的白皮书是一份官方文档，介绍了以太坊的原理、技术细节和实现目标。

白皮书的主要内容包括：
1.以太坊的原理：白皮书介绍了以太坊的基本原理，包括其基于区块链技术的去中心
化特性和智能合约的应用。

它解释了以太坊如何通过构建图灵完备的编程语言，使得开发人员能够创建各种去中心化应用（DApps）。

2.技术细节：白皮书深入探讨了以太坊的技术细节，包括其账户系统、交易模型、Merkle
树、以太币等。

它还解释了以太坊如何通过其独特的共识算法（如工作量证明）确保区块链的安全性和可靠性。

3.实现目标：白皮书明确地提出了以太坊的实现目标，即提供一个能够支持大规模去
中心化应用的平台。

它强调了以太坊的开放性、可扩展性和灵活性，并讨论了如何通过分片、零知识证明等技术实现这些目标。

总的来说，以太坊的白皮书是一份详尽的官方文档，旨在向开发人员和用户介绍以太坊的原理、技术细节和实现目标。

它为以太坊的发展奠定了基础，并为其未来的扩展和改进提供了指导。

戴尔易安信技术白皮书：PowerEdge YX5X 服务器中改进的内存可靠性、可用性和可服务性 (R

Technical White PaperImproved Memory Reliability, Availability, and Serviceability (RAS) in PowerEdge YX5X serversAbstractThis technical white paper describes enhancements made to the memory RASfeatures in PowerEdge YX5X servers that use Intel 3rd Generation (‘Ice Lake’)Xeon Scalable Processors. System configurations for achieving the maximummemory up time are called out.September 2021RevisionsAcknowledgementsAuthor(s): Jordan Chin (Memory Systems Engineering)Support: Mark Dykstra (Escalation Engineering), Kevin Cross (Memory Systems Engineering), Dave Chalfant(BIOS Engineering), Rene Franco (Memory Systems Engineering), and Trent Bates (Product Management)Others: Sheshadri PR Rao (InfoDev)The information in this publication is provided “as is.” Dell Inc. makes no representations or warranties of any kind with re spect to the information in this publication, and specifically disclaims implied warranties of merchantability or fitness for a particular purpose.Use, copying, and distribution of any software described in this publication requires an applicable software license.Copyright © 2021 Dell Inc. or its subsidiaries. All Rights Reserved. Dell Technologies, Dell, EMC, Dell EMC and other trademarks are trademarks of Dell Inc. or its subsidiaries. Other trademarks may be trademarks of their respective owners. [9/13/2021] [Technical White Paper] [Manager]Contents Revisions (2)Acknowledgements (2)Contents (3)Executive Summary (4)1RAS improvements in the Chipset (5)1.1Intel Partial Cache Line Sparing (PCLS) (5)1.2Improvements to Advanced ECC (or Single Device Data Correction) (5)2Smarter Self-healing (6)2.1Memory Health Check (6)2.2Self-healing during Full-Boot Memory tests (7)3Other Memory Serviceability improvements (9)3.1Redesigned Memory Event messaging (9)3.1.1Standard Memory Event Message severity levels (9)3.1.2Prescriptive Remediation (9)3.2User-directed Memory Map-out (10)4Legacy RAS features (12)5Achieving Maximum Memory Uptime (13)6Applicable PowerEdge YX5X servers (14)Executive SummaryExecutive SummaryIn our Memory Errors and Dell EMC PowerEdge YX4X Server Memory RAS Features technical white paper,we provided a primer on memory errors and the Memory Reliability, Availability, and Serviceability (RAS)features on our PowerEdge YX4X products. Most of the principles and RAS feature information from thatwhitepaper continue to remain intact for our latest YX5X line of PowerEdge products. However, as with everygeneration of Dell EMC PowerEdge server product, we continue to innovate and improve. With thisdocument, we would like to share with our customers the next generation of improvements to the memoryRAS capabilities in our latest YX5X line of PowerEdge products. At a high-level, this includes the followingnew features or processes:The content covered in this technical white paper applies only to DRAM based memory modules and does not apply to Intel Optane Persistent Memory. Also, this content is specific to Dell EMC PowerEdge YX5X serverswith 3rd generation Intel Xeon Scalable Processors.RAS improvements in the Chipset1 RAS improvements in the ChipsetIntel introduced the following improvements in their third generation of Xeon Scalable Processors:1.1 Intel Partial Cache Line Sparing (PCLS)The third generation of Xeon Scalable Processors contains 16 segments of spare cachelines per DDR channel in their integrated Memory Controllers (iMC). When BIOS detectsthat DRAM on a DIMM is experiencing single-bit errors (SBEs) for a particular cache line,it may request the iMC to re-map the offending portion of the cache-line to use its internalspare cache instead. Single-bit errors are generally harmless, but if left unchecked, theycould cause unwanted activation of preventative RAS features that can impact systemperformance.This feature will prevent continued triggering of the Dell Predictive Failure Algorithm (PFA) that wouldotherwise activate other RAS features (such as ADDDC) or consume self-healing resources. This results infewer performance impacts and corrective actions required by users. This feature is available on all third GenXeon Scalable Processors. There are no memory or system configuration requirements necessary to takeadvantage of PCLS.1.2 Improvements to Advanced ECC (or Single Device Data Correction)Intel has redesigned and optimized their Advanced Error Correcting Code in 3rd Gen XeonScalable Processors to handle the most common failure patterns known among the majorDRAM suppliers. In doing this, many of the multi-bit error patterns that were uncorrectable byprevious generations of Intel Xeon Scalable Processors are now correctable by 3rd GenXeon SPs. This uplift will result in a significant decrease in uncorrectable memory errors.This enhancement is available on all 3rd Gen Xeon Scalable Processors. There are no memory or systemconfiguration requirements necessary to take advantage of the improved Advanced ECC (or SDDC).For more information about Advanced ECC, see the Memory Errors and Dell EMC PowerEdge YX4X ServerMemory RAS Features technical white paper available on the Dell support site.2 Smarter Self-healingDell Technologies has introduced Memory Health Check and self-healing on full-boots as improvements tomemory self-healing in their PowerEdge YX5X servers. When used together, these features should provide amultiplicative effect on the testing and self-healing coverage for memory while minimally impacting the serverboot time.2.1 Memory Health CheckPrevious Dell EMC PowerEdge implementations of memory self-healing was based ontargeting repairs at a specific memory location where an uncorrectable error or multiplecorrectable errors were detected. For self-healing to take effect, the system must berebooted—thereby, resulting in some downtime. Understanding that downtime can beimpactful and difficult to arrange, it made sense to make use of any downtime because ofself-healing and use it as an opportunity to find other questionable memory locations and repair them. Thiswould prevent the need for future reboot operations and downtime to conduct further self-healing. Thus, DellTechnologies has developed Memory Health Check.Dell Memory Health Check works by first allowing the original self-healing process to execute and completeas described in the original Memory Errors and Dell EMC PowerEdge YX4X Server Memory RAS Featurestechnical white paper. Next, during the same boot cycle as the original self-heal, BIOS executes rigoroustesting throughout the DRAM rank where memory errors have occurred. The rigorous testing used by DellMemory Health Check is the same used by the Dell factory memory screening process. BIOS on thePowerEdge server can quickly and efficiently execute this type of rigorous testing (which can normally takehours) by intelligently testing the affected DRAM rank instead of the entire system memory range.Figure 1 The Memory Health Check process in PowerEdge YX5X serversLastly, if any additional questionable locations are found by this enhanced screening, then those locations are automatically repaired within the same boot cycle, thereby incurring no additional reboot penalty.If BIOS detects that there are more questionable locations than there are self-healing resources available torepair them, then BIOS will automatically map-out that memory region and message to the user that a DIMMreplacement is necessary.The execution of Dell Memory Health Check adds several seconds of boot time and is only executed asneeded based on certain types of memory health events from the previous boot. Normal boots should not see an impact to completion time.This feature is available on all 3rd Gen Xeon Scalable Processors. There are no memory or systemconfiguration requirements necessary to take advantage of Dell Memory Health Check.2.2 Self-healing during Full-Boot Memory testsOn Dell EMC PowerEdge platforms, the default boot behavior is ‘fast boot’, where the DDRbus is trained only partially and memory testing phase is skipped. ‘Full-boot’ is executedafter either a BIOS upgrade, memory configuration change, or after a memory health event(such as uncorrectable error or large number of correctable errors) or the system isconfigured to do so. During a full-boot cycle, BIOS performs memory retraining and a quickmemory test among other activities.On PowerEdge YX5X servers, Dell Technologies has added the ability to perform self-healing based on theresults of this quick memory test which differs from the testing in Memory Health Check in the following ways:Despite the differences in testing, the self-healing and map-out functions of the two features are identical. Ifany questionable locations are detected, then those locations are automatically repaired. If BIOS detects thatthere are more questionable locations than there are self-healing resources available to repair them, thenBIOS will automatically map-out that memory region and indicates to the user (through a message) that aDIMM replacement is necessary.As mentioned earlier, by default, full-boot is executed only in certain conditions. However, full-boot can beenabled to always execute in the BIOS Setup menu. Also, a user can also request for a one-time execution of full-boot. The boot options are accessed as ‘Memory Training’ settings in the System BIOS Settings→Memory Settings. Setting ‘Memory Training’ to ‘Enable’ will cause the system to always perform full-boot, while ‘Retrain at Next Boot’ will perform a one-time full-boot operation during the next reboot cycle.Access to these settings are also available in iDRAC9 by clicking Configuration→BIOS Settings→Memory Settings.Figure 2 Access Boot options on the System Setup pageThis feature is available on all 3rd Gen Xeon Scalable Processors. There are no memory requirements necessary to take advantage of self-healing during full-boot.3 Other Memory Serviceability improvements3.1 Redesigned Memory Event messagingDell Technologies has approached its YX5X PowerEdge servers with the goal of updating memory eventmessaging to improve reliability, usability, and serviceability. Run-time memory event messages in theSystem Event Log and Lifecycle Logs have been redesigned to focus on standardization of memory eventmessaging and prescriptive remediation.3.1.1 Standard Memory Event Message severity levelsAll run-time memory events are now designated by message ID prefixes MEM51xx, MEM61xx, and MEM71xx for alignment according to event severity. Severity is determined based the long-term health outlook of thememory device and whether the event or device correctable by the user.MEM51xx (Informational Severity) Events—Memory device is operational but may benefitfrom additional user action to improve performance or health. The specified user action in theevent message is optional. These event messages will appear only in the Lifecycle Log.MEM61xx (Warning Severity) Events—Memory device has encountered an event that canbe corrected with additional user action. The specified action in the event message is requiredto be performed by the user to prevent potential system outage or resolve an existing outage.If the event persists after performing the specified user action, users should contact thetechnical support teams for further assistance.These events messages will appear in both the Lifecycle Log and System Event Log.MEM71xx (Critical Severity) Events—Memory device has detected an event that hascompromised the reliability or usability of the device. Immediate replacement of the device iseither required or highly recommended. The specified user action in the event message, ifprovided, is intended only for use as a short-term remedy.These event messages will appear in both the Lifecycle Log and System Event Log.3.1.2 Prescriptive RemediationAfter broad review of memory event messaging throughout the server industry, we found it extremelycommon to recommend a general set of user actions to resolve all memory event types. These actions weretypically to reseat the memory device and to update BIOS firmware. Both actions are good practices inmaintaining a healthy memory system. Reseating the memory device can reduce any device-to-socketconnectivity issues, and while updating firmware, ensures that the system is patched for any known issuesand is running the latest memory-training and error-handling algorithms. However, almost all solutions requiresome system downtime and/or physical access to execute. Thus, spending time strictly to complete goodmaintenance practices is an inefficient use of downtime. As mentioned earlier, downtime can be impactful anddifficult to arrange, and as such, every solution prescribed should be carefully considered for the highestlikelihood of resolving the problem.Therefore, in addition to severity, run-time memory event messages on YX5X PowerEdge servers are nowalso categorized based on the recommended resolution. These recommended user actions arerecommended as the best engineering recommendation to address each underlying event. Recommendeduser action aligns with the last two digits of the message ID for all MEM51xx, MEM61xx, and MEM71xxevents.Important—The table above is a simplified version of the listed Recommended Response Actions for MEM51xx, 61xx, and 71xx. Refer to the appropriate product documentation for the fully detailed RecommendedResponse Actions which will include next steps for resolution beyond those listed here.3.2 User-directed Memory Map-outPowerEdge YX5X servers configured with BIOS version 1.3.x or later will enable usersthrough firmware to directly remove a DIMM from the system memory pool withoutphysically handling the device. When a DIMM is no longer part of the system memory pool,its DRAM components will not be used for any read/write operations. Total system memorycapacity is reduced to exclude the mapped out DIMM. Any DRAM faults on this DIMM willno longer produce correctable or uncorrectable errors during run-time operation.Important—DIMMs that are mapped out from the system memory pool are still configured, trained, andtested during the boot operation. Therefore, enabling memory map-out cannot prevent many memory relatedissues that may arise during boot.Mapping out one or more DIMMs can produce an imbalanced memory configuration and significantly impactsystem performance. To retain a supported memory configuration, Dell EMC BIOS may automatically mapout other DIMMs that are not targeted by you. This is necessary to maintain an optimized memoryconfiguration. Therefore, total system memory capacity may be reduced beyond the expected amount. This type of automatic map-out behavior also occurs if the DIMM was physically removed and the system is in an unsupported memory configuration.DIMMs that have been manually mapped out will display the MEM6101 message in system logs to indicate successful map-out. Meanwhile, DIMMs that have been automatically mapped out to maintain a supported memory configuration will display the MEM6101 message and UEFI0339 message in Lifecycle Logs.User directed memory map-out can be enabled and configured in the BIOS Setup menu under System BIOS Settings → Memory Settings → Memory Map Out. Access to these settings are also available in iDRAC9 via Configuration → BIOS Settings → Memory Settings → Memory Map Out.Figure 3 Enable the Memory Map Out feature on the System Setup pageMap-out takes effect during the next system reboot operation and will remain in effect until the you manually reenable the DIMM. Map-out settings will persist through configuration changes and BIOS firmware updates. There are no CPU or memory configuration requirements necessary to take advantage of user directed memory map-out.Warning—This feature should be used with care to ensure that at least one physically installed DIMMs is always enabled. Accidentally disabling all physically installed DIMMs will prevent the system from booting. The system BIOS NVRAM must be cleared to recover from this state. For information about clearing NVRAM, see the respective Installation and Service Manual available on the support site.Legacy RAS features4 Legacy RAS featuresThe RAS features that were supported in the YX4X servers are still supported in YX4X. See the MemoryErrors and Dell EMC PowerEdge YX4X Server Memory RAS Features technical white paper available on thesupport site. However, some updates to those features are listed here as sub-bullets:•Single Error Correction—Double Error Detection (SEC-DED) ECC•Advanced ECCo Improvements to Advanced ECC (or Single Device Data Correction)•Adaptive Double Device Data Correction (ADDDC)•Memory Patrol Scrub•Operating System Memory Page Retirement / Off-lining•Fault Resilient Memory (FRM) / Address Range Mirroring modeo A Non-Uniform Memory Access (NUMA) aware FRM option has been added. When NUMA aware FRM has been enabled, BIOS enforces creation of fault-resilient memory regions whichspan NUMA nodes, and if SNC is enabled, then also spans across sub-NUMA clusters. Earlier,YX4X servers created a fault- resilient region of memory starting from the lowest system memoryaddress without awareness of NUMA. Address range mirroring requested and enabled by theoperating system is and remains NUMA—aware.•Memory Self-Healingo Smarter Self-Healing•Machine Check Architecture RecoveryThe following legacy RAS features have been deprecated in PowerEdge YX5X servers because of insufficient user enablement:•Memory Rank Sparing•Full Memory MirroringAchieving Maximum Memory Uptime5 Achieving Maximum Memory UptimeThe following summarizes actions that users can take to achieve maximum memory uptime on theirPowerEdge YX5X servers:•Configure server using genuine DIMMs from Dell Technologieso Benefit—Memory modules are fully validated and assured by Dell Technologies. Additional self-healing (PPR) resources above and beyond industry standards. Rigorously screened usingproprietary Dell Technologies test patterns and DRAM vendor unique test patterns.•Configure server with x4 DRAM based DIMMso Benefit—Single DRAM Device Correction and ADDDC.•Configure server to operate in Fault Resilient Memory modeo Benefit—Significantly reduced probability of UCEs in critical portions of memory used byoperating systems. Low memory capacity reduction overhead (depending on the systemsettings).•Configure server to run memory patrol scrub in ‘Extended Mode’o Benefit—Patrol-scrub will run after every four hours (instead of 24). Increased scrubbingfrequency will reduce the accumulation of errors in areas of memory with low utilization that arenot being corrected by demand scrub.•Configure server to run ‘Full-Boot’ instead of ‘Fast Boot’o Benefit—Server performs full DDR bus training and testing after every boot operation. Re-training ensuring that the bus is optimally tuned given the current environmental and operating conditions.The quick memory testing will self-heal questionable memory cells that may have developed overtime.•Regularly check for MEM events in Lifecycle Logs and perform the actions as specified—Even for MEM51xx events displayed for information purposes only.o Benefit—Actions specified by MEM61xx and 71xx events are required while those specified by MEM51xx are optional. Even though the MEM51xx events are optional, PowerEdge BIOS hasmade these recommendations based on monitoring memory health and performing the actionsmay improve memory performance or long-term reliability.•Regularly update system firmwareo Benefit—System firmware is patched for any known issues and is running the latest memory training and error handling algorithms.Applicable PowerEdge YX5X servers6 Applicable PowerEdge YX5X serversThe following platforms are considered PowerEdge YX5X servers equipped with 3rd Gen Xeon ScalableProcessors and are therefore covered by this technical white paper:Important—Subsequent to the publication of this document, Dell may continue to add products to its YX5Xserver lineup. If a product is not listed below, please consult with a Dell sales or support representative toconfirm the server generation.PowerEdge leveraged products such as some Precision workstations may also be covered by this document.Please consult with a Dell sales or support representative to confirm.•PowerEdge R750•PowerEdge R750xa•PowerEdge R750xs•PowerEdge R650•PowerEdge R650xs•PowerEdge R550•PowerEdge C6520•PowerEdge MX750c•PowerEdge R450•PowerEdge T550•PowerEdge XR11•PowerEdge XR12。

Cisco光城域网技术白皮书[001]

Cisco 光城域网技术白皮书概述客户局域网速度的提高和复杂性的增加，要求服务供应商在为客户站点提供新型高带宽服务的同时，改善其现有的服务质量。

服务供应商面临的商业挑战是：在寻找一种途径来满足客户的各种要求的同时，能从可实现赢利增长的新投资中获得快速回报。

城域边缘联网解决方案城域边缘网络将用户连接到更广的公共网络中。

这里我们强调的是服务的多样性和客户的反应。

客户局域网速度的提高和复杂性的增加，要求服务供应商在为客户站点提供新型高带宽服务的同时，改善其现有的服务质量。

服务供应商面临的商业挑战是：在寻找一种途径来满足客户的各种要求的同时，能从可实现赢利增长的新投资中获得快速回报。

客户局域网（LAN）通过办公楼、园区和路边的业务接入点来连接。

LAN逐渐超越了其名义上的局域性，因为LAN可以通过公共网络中的高速服务连接很多站点。

对于企业客户来讲，他们最关心的是是否有一种可负担得起的高速服务，能满足他们的全面通信需求。

Long Haul/Extended Long Haul-长距离/超长距离思科光设备产品包括针对于企业接入点的多业务生成平台（MSPP）、城域边缘和局间设备。

思科COMET MSPP支持DS1，DS3，EC1，OC-3/3c，OC-12/12c，OC-48/48c，和OC-192/192c TDM 业务，以及非TDM服务，范围从10/100-Mbps直到吉比特和10吉比特以太网。

可以根据企业的任何要求，满足具有企业系统连接（ESCON）、光连接（FICON）、光通道和其它连接要求的存储区域网络（SAN）的所有需求，同时还可满足话音、视频、LAN数据流量的需要。

服务供应商可将MSPP应用于整个网络中，并可通过添加一块相关接口卡来升级任意服务，来降低库存、维护和培训成本。

城域光联网解决方案用于业务连续性、存储整合和多业务集成的城域光联网城域光网络的应用思科城域网用于业务连续性、存储整合和多业务集成的城域光联网互联网和企业商务应用的迅猛增长对全球企业和服务供应商网络提出了巨大的需求。

超融合技术白皮书之欧阳法创编

深信服超融合架构技术白皮书深信服科技有限公司修订记录第1章、前言错误!未定义书签。

1.1IT时代的变革41.2白皮书总览5第2章、深信服超融合技术架构错误!未定义书签。

1.1超融合架构概述51.1.1超融合架构的定义51.2深信服超融合架构组成模块51.2.1.1系统总体架构61.2.1.2aSV计算虚拟化平台61.2.1.2.1概述61.2.1.2.2aSV技术原理71.2.1.2.2.1aSV的Hypervisor架构81.2.1.2.2.2Hypervisor虚拟化实现101.2.1.2.3aSV的技术特性151.2.1.2.3.1内存NUMA技术151.2.1.2.3.2SR-IOV161.2.1.2.3.3Faik-raid161.2.1.2.3.4虚拟机生命周期管理171.2.1.2.3.5虚拟交换机181.2.1.2.3.6动态资源调度181.2.1.2.4aSV的特色技术191.2.1.2.4.1快虚191.2.1.2.4.2虚拟机热迁移191.2.1.2.4.3虚拟磁盘加密201.2.1.2.4.4虚拟机的HA201.2.1.2.4.5多USB映射211.2.1.3aSAN存储虚拟化221.2.1.3.1存储虚拟化概述221.2.1.3.1.1虚拟后对存储带来的挑战221.2.1.3.1.2分布式存储技术的发展221.2.1.3.1.3深信服aSAN概述221.2.1.3.2aSAN技术原理231.2.1.3.2.1主机管理231.2.1.3.2.2文件副本231.2.1.3.2.3磁盘管理241.2.1.3.2.4SSD读缓存原理261.2.1.3.2.5SSD写缓存原理311.2.1.3.2.6磁盘故障处理机制351.2.1.3.3深信服aSAN功能特性441.2.1.3.3.1存储精简配置441.2.1.3.3.2aSAN私网链路聚合451.2.1.3.3.3数据一致性检查451.2.1.4aNet网络虚拟化461.2.1.4.1网络虚拟化概述461.2.1.4.2aNET网络虚拟化技术原理461.2.1.4.2.1SDN461.2.1.4.2.2NFV471.2.1.4.2.3aNet底层的实现471.2.1.4.3功能特性501.2.1.4.3.1aSW分布式虚拟交换机501.2.1.4.3.2aRouter501.2.1.4.3.3vAF501.2.1.4.3.4vAD511.2.1.4.4深信服aNet的特色技术511.2.1.4.4.1网络探测功能511.2.1.4.4.2全网流量可视521.2.1.4.4.3所画即所得业务逻辑拓扑521.2.2深信服超融合架构产品介绍521.2.2.1产品概述521.2.2.2产品定位52第3章、深信服超融合架构带来的核心价值错误!未定义书签。

ARRIS CAS虚拟化技术白皮书说明书

CAS VirtualizationARRIS Global ServicesARRIS Conditional Access System (CAS) is introducing support for operation of DAC, CASMR, CAST, Advisor and other CAS applications in a Virtual Machine (VM) environment to support hardware/operating system independence, high availability/redundancy, and disaster recovery.ARRIS Global Services is prepared to help you make this transition to Virtualization by offering these services: • Network Discovery and Readiness Planning • Software/ Hardware Procurement • Readiness Plan Implementation • Migration ProcessSERVICE OVERVIEW:WHY VIRTUALIZE?Virtualization enables these essential features for assuring CAS System business continuity and data integrity:• Hardware IndependenceEnables running CAS applications on ARRIS qualified hardware or VMware-supported hardware with comparable performance.• High Availability / RedundancyProvides CAS redundancy/failover, optimizes uptime and data integrity with automated restore and recovery. • Disaster RecoverySupports business-specified recovery time objectives (RTO), recovery point objective (RPO), automated failover testing, and planned migrations. Automated and manual failover is supported.CAS Virtualization ARRIS Global ServicesHORIZONTAL IMAGE ARRIS Global Services ensures your transition to virtualized CAS environment is done with minimal impact to your operations and customer population, allowing you to enjoy the advantages of hardware independence, high availability, hardware consolidation and disaster recovery. Our expert assistance allows you to workthrough the migration to a virtualized environment more quickly while allowing your own staff to focus on day to day operations.Network Discovery and Readiness Plan Creation•Collect critical data points on network •Review data collected•Assess existing architecture, backup/recovery process •Develop Network Plan and architecture •Create BOM, Services List, Action list•Develop proposal for High Availability, Disaster Recovery •Deliver Proposal and Readiness PlanSoftware/ Hardware Procurement•ARRIS procures hardware and software through industry leading 3rd party vendor partners (VMware®, HP®) to support end-to-end virtualization requirements. Virtualization Readiness Implementation •Basic Service•Validate network is ready for virtualization•Install new equipment (server/SAN) and software •Create Virtual Machine for CAS installation •Custom Service•Implement network changes•Implement other changes from the action list •Collect critical data points on network Migration Process•Load DAC software on previously created Virtual Machine •Execute the migration process•Validate the installation and system functionality •Release to productionSERVICE DESCRIPTION©ARRIS Enterprises, Inc. 2014 All rights reserved. No part of this publication may be reproduced in any form or by any means or used to make any derivative work (such as translation, transformation, or adaptation) without written permission from ARRIS Enterprises, Inc. (“ARRIS”). ARRIS reserves the right to revise this publication and to make changes in content from time to time without obligation on the part of ARRIS to provide notification of such revision or change. ARRIS and the ARRIS logo are all trademarks of ARRIS Enterprises, Inc. Other trademarks and trade names may be used in this document to refer to either the entities claiming the marks and the names of their products. ARRIS disclaims proprietary interest in the marks and names of others. VMware, vCenter, and vSphere are registered trademarks of VMware. HP, Store Virtual are registered trademarks of Hewlett-Packard.365-095-27333 x.1 11/2014IMPLEMENTATION EXAMPLEDAC6000 High Availability with Disaster Recovery Phase 1 - Site Preparation•Upgrade CAS hardware to support current software •Replace KLS and other HE components as needed•Upgrade DAC software to current release in native mode •Upgrade software from native to VM with HA •Upgrade to Disaster Recovery•Upgrade CSS/RADD software to current release – nativePhase 2- CAS Application – Native Mode •Upgrade CAS and install SAN servers •Install HP®StoreVirtual® on SAN servers •Upgrade PC to run VMware® vSphere®•Upgrade DAC software to current release – nativePhase 3 - High Availability•Implement N+1 backup DACs per site •Match the current configuration•Install VMware® vCenter® on the CAS hardware •Upgrade DAC apps from native mode to VM mode•Set up VMs so that all fail over to the backup CAS serverPhase 4 – Disaster Recovery•Implement/configure cold backup CAS/KLSs • Uses HP® StoreVirtual® snapshot feature • Uses SANs installed for backup DACs•Confirm remote cold KLS3000 is online/in sync with DAC •Implement failover mechanismORDERING INFORMATIONPart NumberDescription596226-001-00 CAS Virtualization Network Discovery and Readiness Plan Creation 596227-001-00 CAS Virtualization Readiness Implementation476666-043-00CAS Virtualization Migration ProcessNote: Specifications are subject to change without notice.。

青鸟环宇VPN系列产品-技术白皮书v1

青鸟环宇VPN系列产品技术白皮书北大青鸟环宇科技股份XX2002年5月目录第一章企业网络系统信息安全问题11.1企业（政府）网络系统典型架构及其安全现状11.2企业网络面临的威胁及安全需求31.3网络安全基本技术与VPN技术61.4IPS EC网络安全体系81.5用SJW10IPXX机构建VPN系统10第二章青鸟环宇VPN设备SJW10 IPXX机122.1SJW10IPXX机技术说明122.1.1 硬件平台及主要功能122.1.2 软件系统及网络位置132.1.3 功能指标及配置说明152.2SJW10IP安全包技术说明192.2.1 概述192.2.2 系统总体结构202.2.3 功能特点202.3SJW10安全管理中心212.3.1 概述212.3.2 密钥管理方案24第三章典型应用案例253.1某省银行应用SJW10构建安全金融业务网253.2某市银行应用SJW10构建安全银证联网系统273.3某省环保局应用SJW10在I NTERNET上构建安全数字环保网络28第一章企业网络系统信息安全问题1.1企业（政府）网络系统典型架构及其安全现状随着我国通信基础设施建设的飞速发展和互连网络技术的日趋完善，各行各业及政府各部门纷纷借助公共网络基础设施将分散在不同地域的相对封闭的信息系统联成一个整体，以加快信息的流动速度，提高企业的综合竞争力，提升政府办公的工作效率。

就目前大部分网络应用而言，典型的企业（政府）网络结构一般都由一个总部（网络中心）、若干分支机构、数量不等的合作伙伴及移动远程（拨号）用户所组成。

除远程用户外，其余各部分均为规模不等的局域网络系统。

其中总部局域网络是整个网络系统的核心，为企业（政府）各类中心服务器所在地，同时也是网络管理中心。

各部分之间的联接方式多种多样，包括远程拨号、专线、Internet等。

从互联方式来看，则可分为三种模式：通过拨号远程访问企业网络，拨号又可分为通过网络拨入访问服务器和拨入网络服务提供商（如：ISP）两种方式；远程分支机构局域网通过专线或公共网络和总部局域网络连接；合作伙伴（客户、供应商）局域网通过专线或公共网络和总部局域网连接；该典型结构如下页图1.1所示。

OcestorN集群NAS存储系统产品技术白皮书V

保留一切权利。

非经本公司书面许可，任何单位和个人不得擅自摘抄、复制本文档内容的部分或全部，并不得以任何形式传播。

商标声明和其他华为商标均为华为技术有限公司的商标。

本文档提及的其他所有商标或注册商标，由各自的所有人拥有。

除非合同另有约定，华为公司对本文档内容不做任何明示或默示的声明或保证。

由于产品版本升级或其他原因，本文档内容会不定期进行更新。

除非另有约定，本文档仅作为使用指导，本文档中的所有陈述、信息和建议不构成任何明示或暗示的担保。

华为技术有限公司为客户提供全方位的技术支持，用户可与就近的华为办事处联系，也可直接与公司总部联系。

华为技术有限公司地址：深圳市龙岗区坂田华为总部办公楼邮编：518129网址：客户服务电0 18话：客户服务传1真：客户服务邮箱：目录图目录图 1 产品逻辑结构图 (6)图 2 FileStore 软件及应用架构示意图 (7)图 3 N8500集群NAS存储系统关系图 (9)图 4 SAN、NAS一体化典型组网 (10)图 5 使用NBU客户端备份方式组网 (11)图 6 使用NDMP备份组网 (11)图 7 数据通道全互联技术 (14)图 8 通信通道全互联技术 (14)图 9 全局存储空间管理技术 (16)图 10 高效的集群锁技术 (17)图 11 对称式元数据处理技术 (17)图 12 全局名字空间技术 (18)图 13 集群多节点并发访问技术 (18)图 14 主文件系统及其快照文件集 (20)图 15 新生成快照 (20)图 16 主文件系统数据更新 (21)图 17 对快照进行数据更新 (22)图 18 N8500多卷文件系统示意图 (23)图 19 N8500远程复制技术实现步骤 (27)图 20 N8500远程复制技术实现机制 (28)图29 配额管理逻辑架构 (30)图30 配额管理执行方式 (31)1 概述传统NAS系统存在扩展性差、可靠性差等缺陷。

opc ua 白皮书

opc ua 白皮书OPC UA (OLE for Process Control Unified Architecture) 是一种工业自动化领域中常用的通信协议。

它提供了一种标准的、可靠的、高效的和安全的方法，用于连接、监控和控制各种工业设备和系统。

在白皮书中，我们将详细介绍OPC UA 的背景、目的、工作原理、应用场景和优势。

通过深入了解OPC UA，您将能够更好地理解其如何在工业自动化领域中发挥作用，并更好地应用它来提高您的设备和系统的性能和可靠性。

一、背景和目的随着工业自动化技术的不断发展，设备和系统之间的通信和数据交换变得越来越重要。

为了解决不同设备和系统之间的互操作性问题，OPC 基金会制定了一种统一的通信协议，即OPC UA。

OPC UA 的目的是提供一种标准的、可靠的、高效的和安全的方法，用于连接、监控和控制各种工业设备和系统。

它提供了一种通用的通信框架，使得不同的设备和系统能够以一致的方式进行通信，从而简化了工业自动化系统的集成和实施。

二、工作原理OPC UA 是一种基于发布/订阅模型的通信协议。

它使用SOAP 和XML 对数据进行封装和传输，并支持多种传输协议，如TCP/IP、HTTP 和HTTPS。

在OPC UA 系统中，每个设备和系统都有一个OPC UA 服务器，用于发布其数据和状态。

客户端可以通过OPC UA 客户端与服务器进行通信，以获取或设置数据。

通过这种方式，不同的设备和系统可以相互通信，形成一个统一的工业自动化系统。

三、应用场景OPC UA 可以应用于各种工业自动化场景，如：1. 设备集成：将不同厂商的设备和系统集成到一个统一的系统中，实现设备之间的互操作性和数据共享。

2. 数据监控：实时监控设备和系统的状态和数据，实现预警和故障排除。

3. 远程控制：通过OPC UA 客户端远程控制设备和系统，实现远程操控和自动化控制。

4. 数据挖掘：通过OPC UA 获取设备和系统的历史数据，进行数据分析和挖掘，优化设备和系统的性能。

1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。

Ethernet over RPR技术白皮书华为技术有限公司Huawei Technologies Co., Ltd.目录1Ethernet Over RPR的需求 (1)2Ethernet Over RPR技术基本原理介绍 (1)2.1RPR概述 (1)2.2RPR的基本原理 (2)2.3Ethernet Over RPR基本原理 (3)2.4Ethernet Over RPR地址学习 (4)3RPR关键技术 (6)3.1网络保护－故障自愈 (6)3.2拓扑发现 (8)3.3队列技术 (8)3.4公平算法 (10)3.5RPR相交环 (12)4Ethernet Over RPR技术的优势 (16)5典型组网应用 (17)5.1双子卡RPR组网 (17)5.2RPR相交环典型组网 (17)5.3城域业务平台典型组网 (18)6RPR标准化情况 (19)附录A 参考资料 (20)附录B 缩略语 (20)1 Ethernet Over RPR的需求随着电信业务的IP化，NGN、宽带接入、专线等越来越多的业务采用IP网络进行承载，以太网技术作为一种高速、廉价的技术，越来越成为城域网的主要技术。

以太网交换机组网模式主要以星型和环型组网为主，星型组网由于没有保护路径，可靠性差，无法满足电信级业务如NGN业务的要求。

环网技术是一种在电信网络用到比较普遍的技术，能够很好的解决保护倒换的问题，目前应用比较多的以太环网主要以生成树协议(STP)技术为主，但是基于STP的环网技术存在以下问题：1) 收敛速度慢，一般在秒级以上，不满足电信级业务50ms保护倒换的要求；2) 环网技术通常会带来跳数的增加，带来额外的时延和时延抖动，难以满足NGN业务的要求，而基于STP的环网技术没有考虑时延和时延抖动的规避措施；3) 环网作为一个共享带宽的网络，环上节点共享环的带宽资源，必然存在资源的竞争，有可能出现离汇聚节点越远的节点得到的带宽越少，资源的使用不公平，导致环上各个节点上环带宽无法预测，服务质量难以保证；4) 环上任何一个节点的上环业务需要经过环上所有的节点转发，需要环上所有的节点进行相关的配置，比如VLAN信息的配置，否则业务不通，导致在环上添加节点时有可能影响其他节点的业务，升级扩容比较困难；2 Ethernet Over RPR技术基本原理介绍2.1 RPR概述RPR是一种基于分组的环网技术，采用互逆双环拓扑结构，环上的每段光路工作在同一速率上。

RPR的两个环被分别称为0环（Ringlet0）和1环（Ringlet1）。

图1RPR环网构成RPR 0环的数据传送方向为顺时针方向，1环的数据传送方向为逆时针方向。

每个RPR节点(station)都采用了一个以太网中用到的48位MAC地址作为地址标识，从RPR节点设备链路层来看，这两对收发的物理光接口只是一个链路层接口；两个相邻RPR节点之间链路称为段(span)，多个连续的段和其上的节点构成域(domain)。

与传统的分组网络结构相比，RPR技术充分考虑了QoS、可用性、可管理性，使IP网络达到电信级的要求。

2.2 RPR的基本原理节点与环配合，采用分组数据交换，完成各种数据操作。

常用的基本数据操作包括：上环(insert)：节点设备把从其它接口转发过来的帧插入到RPR环的数据流中；下环(copy)：节点设备从RPR环的数据流中接收数据，交给节点上层作相应处理；过环(transit)：将途经本节点的数据流继续转发到下一个节点；剔除(strip)：使途经本节点的数据不再往下转发。

对于过环的数据操作，过环数据流不需要设备上层处理，这样一来，设备处理性能大大提高。

这种数据分组交换体系很容易支撑各种高速链路接口。

对于单播、组播、广播流量，各节点分别采用这些基本数据操作及其组合来予以实现。

下面是单播的实现示意图：图2 单播业务上环/下环/过环在源节点处，采用上环操作，使数据承载到0环或1环中。

目的节点执行数据下环和数据剔除操作。

而中间节点只执行数据过环操作。

对于组播和广播流量，由于有多个目的节点，所以采用的数据传送机制与单播有所不同。

下面是广播业务的一种实现方式：图3 组播业务上环/下环/过环这种方式为0环广播。

其他方式还包括1环广播、双环循环广播等。

2.3 Ethernet Over RPR基本原理Ethernet Over RPR（简称EoRPR）是在RPR链路层上直接承载以太网帧的一种以太网传送技术，华为Ethernet Over RPR符合IEEE 802.17定义的扩展帧规范。

EoRPR的封装格式如下图说明： IP/MPLS Core EoRPR 帧格式示意图RPR PE DSLAMAMGCE LSWDA SA802.1Q DAT ADA SA 802.1Q DA SA802.1Q DA SA TTL 基数基本控制字扩展控制字RPR 扩展头以太网头HEC 图4 EoRPR 帧格式以太网帧封装在RPR 数据帧内，上环时将Ethernet 帧封装在RPR 数据帧内部，在环上进行转发，下环时，剥掉RPR 帧头，用Ethernet 帧进行转发；广播帧在环上绕一圈以后，回到源节点，源节点判断RPR 的SA 为本机的RPR MAC ，将广播帧剥离，不在环上继续转发，有效解决了环网的广播问题。

2.4 Ethernet Over RPR 地址学习如图5所示，转发流程如下所示：图5 EoRPR 转发流程从逻辑上把设备分为两部分：一部分为以太网本地交换部分，简称以太网侧，另一部分为EoRPR处理部分，简称为RPR侧。

从以太网侧收到帧，先根据以太网目的MAC地址查到上哪一个RPR环，在EoRPR部分再根据用户MAC地址查到目的RPR MAC地址和RPR的跳数，在封装RPR帧头以后发送到RPR接口。

从RPR接口接收EoRPR数据帧时，设备会自动学习到以太网源MAC和源RPR节点的RPR MAC地址。

所以，在一个RPR节点中存在两张MAC表，一个是以太网侧的MAC表，用于记录用户MAC地址与以太网出端口的关系，另一个是RPR侧MAC表，用于记录用户MAC与RPR目的节点的关系。

下面详细说明EoRPR在上环、下环、过环节点的MAC地址学习过程：按照图6所示：当用户侧第一个以太网帧从设备3上RPR环1时，在设备3的以太网侧MAC表中学习到用户的源MAC地址A1，如下所示：以太网MAC 用户VLAN 出端口A1 2 Ethernet端口1在设备3的RPR侧的MAC表中不记录本地用户A1的MAC表。

用户第一个帧上环后，穿过设备4到达设备1节点，在设备1的RPR侧的MAC表中学习到用户的源MAC地址A1，如下所示：以太网MAC 用户VLAN 目的RPR节点的MAC 目的RPR节点的跳数A1 2 RPR环1的节点设备3(R1-MAC3)2在设备1的以太网侧的MAC表中学习到用户的源MAC地址A1，如下所示：以太网MAC 用户VLAN 出端口A1 2 RPR环1上环端口在两个相切环的组网情况下，EoRPR环上各节点学习到的MAC地址如下图6所示：图6EoRPR 用户MAC地址学习示意图当一台设备接入多个RPR环时，在以太网侧的MAC中需要学习到用户MAC与哪个RPR 环的上环端口的对应关系。

在EoRPR组网中，过环节点的以太网侧MAC表不学习其它节点之间流量的用户MAC地址，这样减少了过环节点的以太网侧MAC表消耗，消除了过环节点受到MAC地址扫描攻击的风险。

3 RPR关键技术3.1 网络保护－故障自愈RPR采用了SDH的环形结构，同时也继承了一个特大特点，就是故障自愈能力非常强，能够实现50ms时间内的故障保护切换。

下面是链路故障时的保护情况，在故障链路两端的节点内部把0环和1环连接在一起，重新形成一个新的环网。

图7 网络保护－故障自愈对于环上正在传送的数据流量，存在两种保护方式：绕回(wrapping)保护方式和抄近(steering,也有称为源路由的)保护方式。

这里以下图为例加以说明：图8WRAPPING和STEERING保护方式图8左图是故障前的正常数据流，A节点到D节点，走0环，路径为A－B－C－D；图8中图为故障后绕回保护方式，故障发生后，在故障链路两端的节点上通过光路环回，数据路径也在此环回，总的路径为A－B－A－F－E－D－C－D；图8右图为故障后抄近保护方式，从A节点到D节点的数据流量改抄近道，走另外一个环(这里是1环)到达目的节点，路径为A－F－E－D。

绕回方式的优点是故障切换的恢复时间非常短(50ms以内)，只可能丢失极少量的帧，不会造成业务中断的情况，问题是占用带宽较多。

抄近方式避免了带宽的浪费，但是由于需要重新收敛，恢复时间较长，可能会造成一些业务的中断。

部分厂家同时支持绕回和抄近这两种方式，并且结合这两种方式各自的优点，取长补短，配合应用，采取先用绕回方式，后转用抄近方式的保护流程。

当RPR链路故障出现时，立即启动绕回方式进行保护，不中断业务；当各节点的各种状态数据(拓扑信息等)重新收敛并稳定下来后，切换到抄近方式，以便节省带宽。

这样可以到达最佳的保护和利用效果。

3.2 拓扑发现RPR支持拓扑自动发现。

在保护消息或拓扑发现消息帧中包含拓扑信息，这些拓扑信息在环网上广播。

可能的拓扑结构包括全环回结构和链状结构(当部分链路故障时)。

拓扑自动发现对于链路故障时的保护有帮助，同时也给网络扩展提供很好的支持，能够支持节点级别的即插即用，即在环网上新增一个节点或者删除一个节点时，都不需要手工配置数据。

3.3 队列技术RPR在对于过环的流量进行转发处理时，有两种队列转发模式：存储转发模式和直通模式。

存储转发模式实现起来简单通用，而直通模式则能够进一步提高效率。

存储转发模式是基础，必须要要求支持。

即便是采用直通模式的时候，仍然有可能用到存储转发模式，例如直通队列被临时阻塞，就需要转入存储转发模式。

RPR节点具有上环缓存队列和过环缓存队列。

RPR节点具有3个上环缓存队列：A队列、B队列和C队列，分别对应A、B、C三个数据业务等级，提供不同的队列调度优先级。

RPR把要上载的流量分为三个等级：级别A、级别B和级别C。

级别A用于低时延/严格抖动的高优先级流量，提供最低的端到端延时、抖动，具有承诺信息速率定义(CIR)；级别B用于承诺信息速率(CIR)和突发信息速率(EIR)的中等优先级流量，其中对于CIR必须保证一定的带宽，以及端到端的延时和抖动，而对于EIR则不需保证；级别C用于尽力传送的低优先级普通流量，没有带宽定义。

参照下面两图，RPR环路MAC的每个业务通路（对于每个环）可以有一个过环队列，也可以有两个过环队列－PTQ(主过环队列)和STQ(次过环队列)。