OTII服务器白皮书v0.8
USE2.0白皮书

企业统一调度平台
Unified Scheduling for Enterprise (USE)
2.0 版
白皮书
2014 年 6 月
/
USE2.0 白皮书
北京宇信易诚科技有限公司
目录
1. 概述...........................................................................................................................................3 1.1. 编写目的.......................................................................................................................3 1.2. 背景...............................................................................................................................3 1.3. 术语、缩略词说明.......................................................................................................4
超融合技术白皮书

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深信服超融合架构技术白皮书文档密级:内部第1章、前言 (8)1.1IT时代的变革 (8)1.2白皮书总览 (9)第2章、深信服超融合技术架构 (11)1.1超融合架构概述 (11)1.1.1超融合架构的定义 (11)1.2深信服超融合架构组成模块 (11)1.2.1.1系统总体架构 (11)1.2.1.2aSV计算虚拟化平台 (12)1.2.1.2.1概述 (12)1.2.1.2.2aSV技术原理 (13)1.2.1.2.2.1aSV的Hypervisor架构 (14)1.2.1.2.2.2Hypervisor虚拟化实现 (17)1.2.1.2.3aSV的技术特性 (25)1.2.1.2.3.1内存NUMA技术 (25)1.2.1.2.3.2SR-IOV (26)1.2.1.2.3.3Faik-raid (27)1.2.1.2.3.4虚拟机生命周期管理 (28)1.2.1.2.3.5虚拟交换机 (29)1.2.1.2.3.6动态资源调度 (30)1.2.1.2.4aSV的特色技术 (30)1.2.1.2.4.1快虚 (30)1.2.1.2.4.2虚拟机热迁移 (31)1.2.1.2.4.3虚拟磁盘加密 (32)1.2.1.2.4.4虚拟机的HA (33)1.2.1.2.4.5多USB映射 (33)1.2.1.3aSAN存储虚拟化 (35)1.2.1.3.1存储虚拟化概述 (35)1.2.1.3.1.1虚拟后对存储带来的挑战 (35)1.2.1.3.1.2分布式存储技术的发展 (35)1.2.1.3.1.3深信服aSAN概述 (36)1.2.1.3.2aSAN技术原理 (36)1.2.1.3.2.1主机管理 (36)1.2.1.3.2.2文件副本 (37)1.2.1.3.2.3磁盘管理 (38)1.2.1.3.2.4SSD读缓存原理 (39)1.2.1.3.2.5SSD写缓存原理 (45)1.2.1.3.2.6磁盘故障处理机制 (49)1.2.1.3.3深信服aSAN功能特性 (60)1.2.1.3.3.1存储精简配置 (60)1.2.1.3.3.2aSAN私网链路聚合 (61)1.2.1.3.3.3数据一致性检查 (61)1.2.1.4aNet网络虚拟化 (61)1.2.1.4.1网络虚拟化概述 (61)1.2.1.4.2aNET网络虚拟化技术原理 (62)1.2.1.4.2.1SDN (62)1.2.1.4.2.2NFV (63)1.2.1.4.2.3aNet底层的实现 (64)1.2.1.4.3功能特性 (68)1.2.1.4.3.1aSW分布式虚拟交换机 (68)1.2.1.4.3.2aRouter (68)1.2.1.4.3.3vAF (69)1.2.1.4.3.4vAD (69)1.2.1.4.4深信服aNet的特色技术 (69)1.2.1.4.4.1网络探测功能 (69)1.2.1.4.4.2全网流量可视 (70)1.2.1.4.4.3所画即所得业务逻辑拓扑 (70)1.2.2深信服超融合架构产品介绍 (71)1.2.2.1产品概述 (71)1.2.2.2产品定位 (71)第3章、深信服超融合架构带来的核心价值 (73)1.1可靠性: (73)1.2安全性 (73)1.3灵活弹性 (73)1.4易操作性 (73)第4章、超融合架构最佳实践 (74)第1章、前言1.1 IT时代的变革20 世纪90 年代,随着Windows 的广泛使用及Linux 服务器操作系统的出现奠定了x86服务器的行业标准地位,然而x86 服务器部署的增长带来了新的IT 基础架构和运作难题,包括:基础架构利用率低、物理基础架构成本日益攀升、IT 管理成本不断提高以及对关键应用故障和灾难保护不足等问题。
TP-Link R系列企业级路由器主要功能配置实例说明书

声明Copyright © 2021 普联技术有限公司版权所有,保留所有权利未经普联技术有限公司明确书面许可,任何单位或个人不得擅自仿制、复制、誊抄或转译本手册部分或全部内容,且不得以营利为目的进行任何方式(电子、影印、录制等)的传播。
为普联技术有限公司注册商标。
本手册提及的所有商标,由各自所有人拥有。
本手册所提到的产品规格和资讯仅供参考,如有内容更新,恕不另行通知。
除非有特殊约定,本手册仅作为使用指导,所作陈述均不构成任何形式的担保。
目录第1章前言 (1)1.1 目标读者 (1)1.2 本书约定 (1)1.3 适用机型 (1)第2章基础联网设置 (3)2.1 企业路由器基本设置指南 (3)2.1.1 应用介绍 (3)2.1.2 需求介绍 (3)2.1.3 设置方法 (3)2.1.4 注意事项 (7)2.2 企业路由器IPv6上网配置指导 (8)2.2.1 应用介绍 (8)2.2.2 需求介绍 (8)2.2.3 设置方法 (8)2.2.4 疑问解答 (15)第3章设备管理 (17)3.1 如何在外网远程管理(控制)路由器? (17)3.1.1 应用介绍 (17)3.1.2 需求介绍 (17)3.1.3 设置方法 (17)3.1.4 注意事项 (20)3.1.5 疑问解答 (21)3.2 如何设置自动重启? (22)3.2.1 应用介绍 (22)3.2.2 需求介绍 (22)3.2.3 设置方法 (22)3.2.4 注意事项 (23)第4章负载均衡 (24)4.1 多WAN口路由器负载均衡的设置指南 (24)4.1.1 应用介绍 (24)4.1.2 需求介绍 (24)4.1.3 工作原理 (24)4.1.4 设置方法 (25)第5章路由转发模块 (27)5.1 策略路由设置指南 (27)5.1.1 应用介绍 (27)5.1.2 需求介绍 (27)5.1.3 设置方法 (28)5.1.4 疑问解答 (31)5.2 ISP选路设置指南 (33)5.2.1 应用介绍 (33)5.2.2 需求介绍 (33)5.2.3 设置方法 (34)5.3 静态路由设置指南 (36)5.3.1 应用介绍 (36)5.3.2 需求介绍 (36)5.3.3 设置方法 (37)5.4 线路备份设置指南 (38)5.4.1 应用介绍 (38)5.4.2 需求介绍 (38)5.4.3 设置方法 (38)5.4.4 注意事项 (40)5.5 虚拟服务器设置指南 (41)5.5.1 应用介绍 (41)5.5.2 需求介绍 (41)5.5.3 设置方法 (42)5.5.4 疑问解答 (43)5.6 NAT-DMZ功能设置指南 (44)5.6.1 应用介绍 (44)5.6.2 需求介绍 (44)5.6.3 设置方法 (45)第6章AP和易展管理 (47)6.1 AP管理设置指南 (47)6.1.1 应用介绍 (47)6.1.2 需求介绍 (47)6.1.3 设置方法 (47)6.2 易展AP设置指南 (53)6.2.1 应用介绍 (53)6.2.2 需求介绍 (53)6.2.3 设置方法 (54)6.2.4 注意事项 (58)第7章行为管控 (59)7.1 连接数限制设置指南 (59)7.1.1 应用介绍 (59)7.1.2 需求介绍 (59)7.1.3 设置方法 (59)7.1.4 疑问解答 (60)7.2 访问控制设置指南 (61)7.2.1 应用介绍 (61)7.2.2 需求介绍 (61)7.2.3 设置方法 (61)7.2.4 疑问解答 (67)7.3 应用限制设置指南 (68)7.3.1 应用介绍 (68)7.3.2 需求介绍 (68)7.3.3 设置方法 (68)7.4 网址过滤设置指南 (71)7.4.1 应用介绍 (71)7.4.2 需求介绍 (71)7.4.3 设置方法 (71)7.4.4 疑问解答 (75)7.5 网页安全设置指南 (76)7.5.1 应用介绍 (76)7.5.2 需求介绍 (76)7.5.3 设置方法 (76)第8章安全防护 (78)8.1 ARP防护设置指南 (78)8.1.1 应用介绍 (78)8.1.2 需求介绍 (78)8.1.3 设置方法 (78)8.1.4 疑问解答 (84)8.2 MAC地址过滤设置指南 (86)8.2.1 应用介绍 (86)8.2.2 需求介绍 (86)8.2.3 设置方法 (86)第9章VPN模块 (88)9.1 IPSec VPN设置指南 (88)9.1.1 应用介绍 (88)9.1.2 需求介绍 (88)9.1.3 设置方法 (89)9.2 L2TP VPN设置指南 (96)9.2.1 应用介绍 (96)9.2.2 需求介绍 (96)9.2.3 设置方法 (97)9.3 PPTP VPN设置指南 (105)9.3.1 应用介绍 (105)9.3.2 需求介绍 (105)9.3.3 设置方法 (106)9.4 L2TP VPN代理上网设置指南 (115)9.4.1 应用介绍 (115)9.4.2 需求介绍 (115)9.4.3 设置方法 (115)9.5 PPTP VPN代理上网设置指南 (120)9.5.1 应用介绍 (120)9.5.2 需求介绍 (120)9.5.3 设置方法 (120)第10章认证管理 (125)10.1 一键上网设置指南 (125)10.1.1 应用介绍 (125)10.1.2 需求介绍 (125)10.1.3 设置方法 (126)10.2 短信认证设置指南 (130)10.2.1 应用介绍 (130)10.2.2 需求介绍 (130)10.2.3 设置方法 (131)10.3 Portal认证设置指南—使用内置WEB服务器和内置认证服务器 (136)10.3.1 应用介绍 (136)10.3.2 需求介绍 (136)10.3.3 设置方法 (137)10.4 Portal认证设置指南—使用内置WEB服务器和外部认证服务器 (141)10.4.1 应用介绍 (141)10.4.2 需求介绍 (141)10.4.3 设置方法 (142)10.5 Portal认证设置指南—使用外置WEB服务器和内置认证服务器 (146)10.5.1 应用介绍 (146)10.5.2 需求介绍 (146)10.5.3 设置方法 (147)10.6 Portal认证设置指南—使用外置WEB服务器和外置认证服务器 (150)10.6.1 应用介绍 (150)10.6.2 需求介绍 (150)10.6.3 设置方法 (151)10.7 免认证策略的使用方法 (154)10.7.1 应用介绍 (154)10.7.2 需求介绍 (154)10.7.3 设置方法 (155)10.8 Portal认证中,外部WEB服务器建立规范 (158)10.8.1 应用介绍 (158)10.8.2 流程规范 (159)第11章工业级特性 (163)11.1 如何使用工业级路由器? (163)11.1.1 产品介绍 (163)11.1.2 需求介绍 (163)11.1.3 设置方法 (164)第12章其它功能 (168)12.1 地址组的设置与管理 (168)12.1.1 应用介绍 (168)12.1.2 需求介绍 (168)12.1.3 设置方法 (168)12.1.4 疑问解答 (170)12.2 带宽控制设置指南 (172)12.2.1 应用介绍 (172)12.2.2 需求介绍 (172)12.2.3 设置方法 (172)12.2.4 疑问解答 (175)12.3 PPPOE服务器应用设置指南 (177)12.3.1 应用介绍 (177)12.3.2 需求介绍 (177)12.3.3 设置方法 (178)12.3.4 疑问解答 (181)12.4 网络唤醒功能使用指南 (183)12.4.1 应用介绍 (183)12.4.2 需求介绍 (183)12.4.3 设置方法 (183)12.5 诊断工具使用指南 (186)12.5.1 应用介绍 (186)12.5.2 需求介绍 (186)12.5.3 设置方法 (187)第1章前言本手册旨在帮助您正确使用R系列企业级路由器。
SPARC T8 系列服务器安全指南说明书

如果将本软件或相关文档交付给美国政府,或者交付给以美国政府名义获得许可证的任何机构,则适用以下注意事项:
U.S. GOVERNMENT END USERS: Oracle programs, including any operating system, integrated software, any programs installed on the hardware, and/or documentation, delivered to U.S. Government end users are "commercial computer software" pursuant to the applicable Federal Acquisition Regulation and agency-specific supplemental regulations. As such, use, duplication, disclosure, modification, and adaptation of the programs, including any operating system, integrated software, any programs installed on the hardware, and/or documentation, shall be subject to license terms and license restrictions applicable to the programs. No other rights are granted to the U.S. Government.
2023年OTII 2U服务器技术规范

目录版权声明 (I)编制说明 ................................................................. I I 前言 . (I11)一、Mii (1)二、术语、缩略语 (1)三、总体要求 (2)四、配置要求 (2)(一)单路服务器规格 (2)(二)双路服务器规格 (4)五、设计要求 (5)(一)机框 (5)1. 规格尺寸 (5)2. 前面板布局 (6)3. 后面板布局 (7)4. 按钮和指示灯 (8)5. 上架安装 (9)(二)部件 (9)1. 逻辑框图 (9)2. CPU (10)3. ...................................................H6. 加速硬件 (12)7. 时钟和同步(可选) (13)8. IO模块(可选) (14)(三)电源 (15)1. ∙'J (15)2. 转换效率 (15)3. 冗余和热插拔 (15)4∙全和保护.・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・155. 电源管理 (15)(四)风扇 (16)1. 规格尺寸 (16)2. 冗余和热插拔 (16)3. 风扇调速 (16)六、环境要求 (16)七、劄里要求 (17)(一)带外管理功能 (17)1. 资产管理 (17)2. 部件信息管理 (17)3. 传感器监控 (18)4. 电源和风扇管理 (18)5. 故障告警 (19)7.其他功能 (20)(二)带内管理功能 (20)八、软件兼容性 (21)OTII(OpenTeIecomITInfraStr11Ct11re)即电信开放IT基础设施项目,隶属于中国最大的开源硬件组织一开放数据中心委员会(ODCC)。
2023年11月由中国移动联合中国电信、中国联通、中国信通院、英特尔等公司共同发起,是业界首个由多家运营商联合发起的服务器开发合作项目,首要目标是形成面向5G及边缘计算的深度定制、开放标准、统一规范的服务器技术方案及产品。
TongGTP v6.0技术白皮书(参考仅供)

T ongT ech®TongGTP6.0技术白皮书东方通科技公司技术白皮书(14)目录1.前言 (3)2.TONGGTP V6.0主要新特性 (3)3.基本术语 (5)4.体系结构 (6)4.1. 产品组成 (6)4.2. GTP节点 (6)4.3. GTP客户端 (7)4.4. 监控管理中心 (7)5.主要功能 (8)5.1. 文件收发 (9)5.1.1.多种文件类型 (9)5.1.2.任务的灵活调度 (9)5.1.3.文件筛选及过滤 (9)5.1.4.压缩加密 (9)5.1.5.断点续传 (10)5.1.6.流量控制 (10)5.1.7.文件组播 (10)5.1.8.文件发送后归档 (10)5.1.9.文件重发 (10)5.2. 权限管理 (10)5.3. 系统管理 (11)5.3.1.管理方式 (11)5.3.2.分域管理 (11)5.3.3.传输进度监控 (11)5.3.4.预警及故障告警 (11)5.3.5.集中任务管理 (11)5.3.6.任务配置 (12)5.3.7.动态配置 (12)5.3.8.监控管理 (12)5.3.9.查询统计 (12)5.3.10.操作审计 (12)5.4. 日志管理 (12)5.5. 链路管理 (13)5.6. 易用工具 (13)5.6.1.批量导入导出 (13)5.6.2.本地日志查询统计 (13)5.6.3.本地运行监控 (13)5.6.4.链路检测工具 (14)5.7. 传输扩展 (14)5.7.1.任务出口 (14)5.7.2.灵活的加密算法 (14)6.系统特点 (14)7.应用案例 (15)1.前言随着大型企业或机构实现业务集中处理后,各个数据中心产生的大量数据信息,以及各种图像文件等需要及时快速地从省、市一级的服务器或终端传输到全国中心,或者从全国中心往省、市、县级中心下发,多数企业还在使用传统的FTP解决传输问题。
主要因为它是一个免费的非侵入式的软件,一个技术人员稍加培训即可在基于IP的网络上进行各种类型文件的传输操作。
Oracle Exadata X8数据库机器白皮书说明书

Wherever Oracle Databases Live and However They Are Consumed, Exadata X8 Delivers Performance in a League of Its OwnBy Mark Peters, Principal Analyst & Practice DirectorJune 2019This ESG White Paper was commissioned by Oracleand is distributed under license from ESG.™Oracle Databases Achieve Unprecedented Speed Running on the New Exadata Database Machine X8 White PaperContentsIntroduction and Executive Summary (3)Databases: Their Market and Their Varied, Voracious Data Demands (3)The Essence of Exadata, X8, and Cloud-enhancement as Desired (5)The Bigger Truth (7)Introduction and Executive SummaryMany things in IT are talked about a lot just because they have to be, while other things are talked about very little because they cannot be. As examples: There are lengthy written guides and web communities based on myriad approaches to system integration because (let’s face it) a lot of IT over many decades has been about trying to make operational round pegs fit application square holes. On the flip side, “unique” is a word that is eschewed by pretty much everyone except for vendor marketing departments because (let’s face this, too) there’s a lot of similarity across many capabilities from much of the vendor community. The journey may vary a tad, but the destination is often similar.Oracle Exadata Database Machine has now stood for a full decade in stark contrast to both these “norms.” As an Oracle Engineered System, the need to talk about integration is essentially precluded; the same engineers who build Oracle Database also ensure the Exadata is a seamless place to run the software. Control of all the IP means Oracle has a controlled and predictable environment in which to run its Database on Exadata, which in turn means it can deliver features that bring additional business value to its customers and enable them to obtain additional operational benefits and financial payback from their Oracle Database investment.And that’s it in a nutshell: Oracle’s justifiable claim to be providing, with Exadata, the best platform for Oracle Database is derived from it having the same DNA across both elements. That commonality helps make it economically attractive, scalable, and secure. It also generates the contrast that Oracle fairly enjoys making when comparing Oracle Database running on Exadata to Oracle Database running on any other vendor’s equipment; the others are constrained to essentially treat Oracle Database like any other application sitting on a VM or running as a process on generic hardware. Exadata delivers incremental business value when combined with Oracle Database, whether in the cloud or on-premises. The arrival of Exadata X8 provides yet another positive turn of the specification—and hence, value—screw, while Exadata turning ten years old is a timely reminder to review the power of the platform in general. It’s a power borne by suitability; and to appreciate that suitability, a quick reminder of the database market and its data needs is warranted. Databases: Their Market and Their Varied, Voracious Data DemandsThe Database MarketDatabases have long served as the lifeline of crucial business applications, and thus are often a foundation of businesses. Contemporary IT—and the world as we understand it—could not function without databases. But exponential data growth, the need to support more database workloads, and the resulting database sprawl have created significant challenges for IT organizations…challenges that are now complicated and extended by the arrival of multiple iterations of the cloud. With most organizations having mandates to adopt the cloud to some extent—database workloads are no exception—they should look to a partner that can not only meet their on-premises requirements, but also provide a path to a cloud that best fits their needs. And for many, getting an identical user and operational experience would be perfect. Moreover, as much as the IT environment has changed, so have the databases themselves…such as Oracle’s latest 19c iteration. These are not the databases of past generations, and so you wouldn’t want to use obsolete approaches of the past to serve them their lifeblood (i.e., data).Database Data: Proliferation, Demands…and the CloudManaging data growth and database size are far and away the greatest challenges organizations face in their current database environments, cited by 48% of respondents in ESG research (see Figure 1).1 Other key aspects of contemporary database environments include:1 Source: ESG Survey, Enterprise Database Trends, January 2017. All ESG research references and charts in this white paper have been taken from this survey, unless otherwise noted.• Additional top challenges are database performance, supporting databases in cloud environments, and the drive for ever-better security and compliance. This is not a pick-and-mix list; users need solutions to all of these problems from their chosen vendor.• If organizations had only one database, life would certainly be easier; but ESG research shows 55% of organizations have more than 25 databases in production, and having hundreds is not at all rare (see Figure 2). These tools support everything from transactional and analytical workloads to the latest machine learning applications, and they can be running everywhere from on-premises through “cloud-adjacent” environments to the cloud.• Often databases are siloed (with their own infrastructure), which compounds the management complexity and leads to inefficient utilization of hardware and software investments.• Maintaining multiple independent environments results in complex and growing security controls/policies, access needs/rules, and patching requirements, and this only adds to the management headaches IT organizations areforced to either suffer through or try to address.Figure 1. Challenges with Database Environment and Supporting InfrastructureSource: Enterprise Strategy Group48%We do not have any challenges Lack of skilled staffCreating test/development environmentsProvisioning infrastructure on-premisesPatch, maintenance and version update processesSupporting new applications and/or usersMeeting security/compliance requirementsSupporting databases in cloud environmentsMeeting database performance requirementsManaging data growth and database sizeIn general, which of the following challenges does your organization have with its current database environment and supporting infrastructure? (Percent of respondents, N=354, three responses accepted)Figure 2. Number of Production Databases DeployedSource: Enterprise Strategy GroupThe sum of the above—complexity and database iteration combined with data growth and the need to share it across and between organizations—is driving a need for consolidation. When ESG last investigated this, 50% of organizations were already consolidating databases, with another 25% planning on it. That consolidation in turn means DBAs must merge multiple workloads onto a given infrastructure, invariably creating more performance issues to be addressed.And this is all before adding in the operational curve ball of the cloud! Database provisioning and consolidation exercises increasingly must acknowledge that the right platform can change over time. Not surprisingly, ESG’s research has found that when evaluating and selecting databases, the most important attributes are that the database be capable of being cloud-based and include the ability to easily scale up and/or down as workloads demand.2The Essence of Exadata, X8, and Cloud-enhancement as DesiredGood news comes in the shape of the “Oracle on Oracle and for Oracle” solution: Exadata. Its value proposition is a piece of cake to convey: It is an optimized platform on which to run Oracle Database. While Oracle Database can run on myriad other systems, the optimal experience is running it on Exadata. It is an Oracle Engineered System that is tuned to be best for Oracle Database and integrates unique value-add features that are unavailable from other vendors. As such, it is not general purpose, nor simply fit-for-purpose: It is purpose-built. Over its decade in the market, Exadata has grown as the—identical everywhere—foundation for on-premises, Oracle Cloud, and Oracle Autonomous Database operations. The company states that Exadata is run by 77% of the Fortune Global 100 (25% of which have also adopted Exadata Cloud Service).This paper is not designed to provide extensive technical details, but suffice it to say Exadata is an ideal database platform that combines scale-out, database-optimized compute, networking, and storage hardware (to deliver both performance and economy) with smart software in the form of both specialized algorithms that can vastly improve all aspects of 2 Source: ESG Brief, Database Purchase Criteria , June 2017.17%Less than 5 5 to 1011 to 2526 to 5051 to 7576 to 100101 to 150151 to 200More than 200Don’t know How many production databases does your organization currently have deployed? (Percent of respondents, N=354)database processing, as well as automated management to optimally handle configuration, updates, and performance.3 Better yet, all the above is true across all workloads, whether they are petabyte-scale data warehouses or business-crucial OLTP applications, and whether they are traditional business applications or next-generation analytics and machine learning. Organizations can confidently consolidate their Oracle Databases and associated workloads onto a single platform born on-premises and then extend to some, or full, public cloud consumption options if desired.With Exadata, Oracle is allowing its users to deliver against the common (but also not commonly well-defined or stable) mandate to leverage the cloud by providing paths to deploy cloud operating models ranging from optimizing on-premises infrastructure before moving to the public cloud,4 through utilizing an on-premises cloud service (Oracle Cloud at Customer) that provides a public cloud experience but with the security and control of an on-premises environment, to a Cloud Adjacent offering,5 and all the way to a public cloud deployment with Exadata Cloud Service or Autonomous Database.Exadata X8As with Exadata in general, this paper is not the place to conduct a lengthy exposition of the new Exadata X8. A few headlines serve to convey the arrival of not only dramatic specification improvements, but also some key functionality:•Speed—Scale-out servers use the improved clock-speed of the latest 24-core Intel processors, as well as NVMe flash, to help Exadata X8 deliver up to 560GB/sec I/O throughput and 6.57M OLTP read IOPS (with 3.5M IOPS at under 250 microseconds). For obvious reasons, Oracle encourages all its prospects to compare its throughput and speed with typical all-flash storage arrays.•Capacity—There are 40% higher capacity 14TB Helium disk drives (the highest capacity at which Oracle’s prerequisite availability targets can be achieved). Additionally, a new Storage Server X8-2 XT expansion unit delivers dramatically less expensive storage for less-accessed, older, or regulatory data, but with all the benefits of Oracle Database storage(e.g., Hybrid Columnar Compression,6 identical operations, and security/encryption). Think of this as seamlesslyconverging the data[base] in addition to converging the infrastructure.•Function—60% more cores in Exadata X8-2 Storage Servers help offload more Oracle Database processing and also improve security by mitigating threats such as “Spectre” and “Meltdown” in silicon, eliminating software overhead.Also, automated, cloud-scale, full-stack performance management and monitoring combines AI and years of real-world performance “triage experience” with best practices to automatically detect performance issues, proactively determine root causes, and address problems without human intervention.Exadata + Oracle Database: ‘Hand in Glove’ and Extended ValueWhile it is impressive and valuable to know about Exadata’s specifications or extensive HA features, what really sets it apart are over 60 Exadata-only features for Oracle Database, such as Fast Node and Cell Death Detection, Automatic ASM Mirror3 In patching alone, for example, organizations can use Exadata to drastically reduce the number of patches. Oracle’s website details a $200B retailer that has reduced its patches 95% p.a. This enabled its IT group to improve productivity, while allowing its DBAs to more easily ensure their infrastructure was up to date and reduce risk.4 The cloud alliance announced in June 2019 between Oracle and Microsoft further demonstrates Oracle’s determination to be a key element in the public cloud world, while acknowledging that it will play to its strengths (such as high-performance Oracle Database) and allow customers to utilize the value of other mega-clouds as appropriate.5 This offering is outlined in the ESG White Paper, The Business Advantages of Cloud Adjacent Oracle Databases on Exadata, published March 2019.6 Hybrid Columnar Compression is an approach to database data compression that utilizes both the database and storage capabilities of Exadata to significantly reduce Oracle Database capacity requirements (10-15X is the norm). Less data handled also improves performance.Reads on IO Error Corruption, Automatic Statistics, and many more. It’s a veritable smorgasbord of delights for IT operations and DBAs to use, which delivers unsurpassed—and genuine—value to Oracle Database users.Extensions to this unique hand-in-glove functionality have continued with the latest Exadata X8, most notably the machine learning that enables Automatic Indexing with Oracle Database 19c. A fully-automated capability based on technology from Oracle Autonomous Database (what Oracle refers to as a “self-driving database”), this improves database performance and eliminates manual index-tuning tasks for critical applications.7 This means that DBAs can simultaneously stop the manual task of creating indexes for applications running on Oracle Database, achieve faster performance, and gain time to focus on more strategic business objectives.Users can also use in-database machine learning without having to move any data. All the algorithms run inside Oracle Database—close to the data—which ensures no production impact. This delivers dramatic performance gains for actions such as the faster scoring and validating of models against the freshest production data.While Oracle can at times display a love for extreme detail, Exadata’s powerful core value makes that largely unnecessary. Simply, it is the most integrated, optimized, flexible, and capable place to run Oracle Databases, with a set of attributes that allow it to cost less, scale better, and be more secure than alternatives. As such, it has become a foundation for Oracle Autonomous Database and Oracle SaaS applications, and the architecturally equivalent stepping stone to Oracle Cloud. The Bigger TruthWhile the term “no brainer” is a tad simplistic and can overlook relevant adjacent factors, it is clear that unique capabilities and user-value are available with Oracle Exadata when it’s used in concert with Oracle Database. That means any Oracle Database user not utilizing Exadata is missing out on some or all of those capabilities and value.As Exadata X8 heralds the second decade for this offering, it has increased its deployment choices alongside Oracle Database to mirror the increasingly hybrid cloud world, and thus offers a congruence of architecture, features, and security wherever and however it is deployed—on-premises or in the cloud. Oracle also continues to add impressive functionality that can translate to not only IT improvements, but also business value; a prime example being the machine learning in Oracle Database 19C to enable auto indexing with Exadata X8.Oracle is not famed for being shy in its own marketing claims; regarding the use of Exadata to run Oracle Database, it likes to say that the combination is “so smart that there’s no second place.” It would be easy to dismiss that as marketing bluster, but it actually deserves consideration; “second place” would suggest vendors competing at the same level…but really Oracle has put Exadata in a league of its own.7 Oracle reports a recent test in which 15 years of manual engineering were compared with what could be done in less than 24 hours with modern machine learning when the Automatic Indexing successfully created two-thirds the number of NetSuite application indexes and also increased database and application performance.All trademark names are property of their respective companies. Information contained in this publication has been obtained by sources TheEnterprise Strategy Group (ESG) considers to be reliable but is not warranted by ESG. This publication may contain opinions of ESG, which are subject to change from time to time. This publication is copyrighted by The Enterprise Strategy Group, Inc. Any reproduction or redistribution of thispublication, in whole or in part, whether in hard-copy format, electronically, or otherwise to persons not authorized to receive it, without the express consent of The Enterprise Strategy Group, Inc., is in violation of U.S. copyright law and will be subject to an action for civil damages and, if applicable, criminal prosecution. Should you have any questions, please contact ESG Client Relations at 508.482.0188. ********************** P. 508.482.0188 Enterprise Strategy Group is an IT analyst, research, validation, and strategy firm that provides actionable insight and intelligence to the global IT community.© 2019 by The Enterprise Strategy Group, Inc. All Rights Reserved.。
浪潮英信服务器 NE3160M5 技术白皮书说明书

浪潮英信服务器NE3160M5技术白皮书文档版本V1.3发布日期2022-07-18版权所有© 2022浪潮电子信息产业股份有限公司。
保留一切权利。
未经本公司事先书面许可,任何单位和个人不得以任何形式复制、传播本手册的部分或全部内容。
环境保护请将我方产品的包装物交废品收购站回收利用,以利于污染预防,共同营造绿色家园。
商标说明Inspur浪潮、Inspur、浪潮、英信是浪潮集团有限公司的注册商标。
本手册中提及的其他所有商标或注册商标,由各自的所有人拥有。
安全声明服务器产品安全一直是浪潮关注的焦点,保障产品安全是浪潮的关键战略之一。
为使您更清晰地了解服务器产品,请注意如下安全风险声明。
●在调整用途或淘汰服务器时,为了保护数据隐私,建议从BIOS、BMC中恢复固件出厂设置、删除信息、清除日志。
同时,建议采用第三方安全擦除工具对硬盘数据进行全面安全擦除。
●您购买的产品、服务或特性在业务运营或故障定位的过程中将可能获取或使用用户的某些个人数据(如告警邮件接收地址、IP地址),故您有义务根据所适用国家的法律制定必要的用户隐私政策并采取足够的措施,以确保用户的个人数据受到充分的保护。
●服务器开源软件声明的获取,请直接联系浪潮客户服务人员咨询。
●部分用于生产、装备、返厂检测维修的接口、命令,定位故障的高级命令,如使用不当,将可能导致设备异常或者业务中断,故不在本资料中说明。
如需要,请向浪潮申请。
●浪潮服务器的外部接口未使用私有协议进行通信。
●浪潮建立了产品安全漏洞应急和处理机制,保证第一时间及时处理产品安全问题。
若您在浪潮产品中发现任何安全问题,或者寻求有关产品安全漏洞的必要支持,可以直接联系浪潮客户服务人员。
浪潮将一如既往的严密关注产品与解决方案的安全性,为客户提供更满意的服务。
内容声明您购买的产品、服务或特性等应受浪潮集团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。
对象标示符(OID)白皮书_201507

对象标识符(OID)白皮书(2015)中国电子技术标准化研究院2015年7月——————————————————————版权声明——————————————————————本白皮书1版权属于中国电子技术标准化研究院(工业和信息化部电子工业标准化研究院),凡转载或引用本文的观点、数据,请注明来源。
中国电子技术标准化研究院OID注册中心拥有本报告内容的永久的、非专有的、免费的全球性的版权许可,未经OID注册中心的书面授权,不得以任何方式复制、抄袭、影印、翻译本报告的任何部分。
1本白皮书由我国OID白皮书以及国际OID白皮书组成,版权声明及前言仅适用于我国OID白皮书。
前言进入二十一世纪以来,随着信息技术的迅速发展,我国信息化建设的步伐也在不断加快,物联网、云计算等新兴技术层出不穷,使得信息处理的对象从传统的计算机等终端设备扩展至日常生活中的各种事物,延伸了网络通信的外延和内涵,把任何物体与互联网相连接,进行信息交换和通信。
为了更好地实现不同应用、不同系统间的互连互通,我们需要建立兼容性强、技术成熟的标识体系,对我国各类对象进行标识管理。
国务院发布的《推进物联网有序健康发展的指导意见》中主要任务部分指出“加快物联网编码标识等基础共性标准研究制定”;在国家发展改革委、工业和信息化部、科技部等部门联合印发的《物联网发展专项行动计划》中,在标准制定、技术研发、政府扶持措施、法律法规等多个部分中明确指出要加强物联网标识标准、标识解析平台、标识解析和管理技术、标识等资源立法等各方面工作。
建立我国标识管理体系,首先是未来信息化建设发展的技术需求,传统的TCP/IP协议地址分配已经无法满足数量庞大、形态各异的信息终端互连的需求;其次是各应用领域实现信息交互与共享的需求,不同对象间的通信、各应用领域的互联互通等需要各应用领域在兼容原有机制的同时,将其基础建设纳入到国家体系下;再次是安全保障的需求,我国信息资源实现科学有序的标识管理,既是信息化产业发展的客观需求,也是保障我国信息安全、信息资源自我管理和利用、维护国家主权的重要手段。
TopADS3000产品-技术白皮书

天融信异常流量管理与抗拒绝服务系统(TopADS系列)技术白皮书天融信TOPSEC®北京市海淀区上地东路1号华控大厦100085电话:(86)10-82776666传真:(86)10-82776677服务热线:400-610-5119800-810-5119Http: //目录1拒绝服务攻击背景 (1)1.1拒绝服务攻击定义 (1)1.2分布式拒绝服务攻击现状 (2)1.3僵尸网络现状 (3)1.4典型DD O S攻击原理 (4)2产品概述 (8)3产品技术架构 (9)4产品主要功能 (10)5产品优势与特点 (11)5.1先进的新一代S MART AMP并行处理技术架构 (12)5.2全面的拒绝服务攻击防御能力 (13)5.3全64位的原生IPV6支持 (13)5.4完善的应用层攻击防御功能 (13)5.5灵活多样的部署方式 (14)5.6针对运营需求的大客户两级保护对象策略 (14)5.7高可靠的业务保障能力 (15)5.8可视化的实时报表功能 (15)6产品典型部署方案 (15)6.1在线串接部署方式 (16)6.2旁路部署方式 (16)7产品型号和规格 (17)8产品资质 (18)1拒绝服务攻击背景1.1 拒绝服务攻击定义拒绝服务攻击(DOS)定义。
DoS是Denial of Service的简称,即拒绝服务,造成DoS 的攻击行为被称为DoS攻击,其目的是使计算机或网络无法提供正常的服务。
最常见的DoS 攻击有计算机网络带宽攻击和连通性攻击。
带宽攻击指以极大的通信量冲击网络,使得所有可用网络资源都被消耗殆尽,最后导致合法的用户请求就无法通过。
连通性攻击指用大量的连接请求冲击计算机,使得所有可用的操作系统资源都被消耗殆尽,最终计算机无法再处理合法用户的请求。
分布式拒绝服务(DDoS: Distributed Denial of Service)攻击指借助于客户/服务器技术,将多个计算机联合起来作为攻击平台,对一个或多个目标发动DDoS攻击,从而成倍地提高拒绝服务攻击的威力。
产品白皮书

Ibot8.0产品白皮书第 1 页共47 页目录1.前言 (6)1.1.目的范围 (6)1.2.产品定位 (6)1.3.名词术语 (6)2.公司简介 (8)2.1.小I简介 (8)2.2.小I发展历程 (8)3.产品概述 (11)3.1.智能机器人 (11)3.2.产品架构 (11)3.1.1.机器人前端平台 (12)3.1.2.智能服务引擎平台 (12)3.1.3.机器人统一管理平台 (14)3.1.4.知识库组织说明 (14)3.3.产品演进路径 (17)3.4.技术特点 (17)3.5.技术指标 (18)3.6.扩展接口 (19)3.7.优势特性 (19)4.产品主要功能 (22)4.1.基础智能问答 (22)4.2.前端用户功能 (23)4.2.1 Web机器人功能 (23)4.2.2微信机器人功能 (28)4.2.3IM机器人功能 (31)4.2.4短信机器人功能 (33)4.3.后台主要管理功能 (35)4.3.1知识管理功能 (35)4.3.2服务管理功能 (36)4.3.3渠道管理功能 (36)4.3.4素材管理功能 (37)4.3.5语音管理功能 (38)4.3.6运维管理功能 (38)4.3.7系统管理功能 (39)5产品实施 (39)5.1.实施流程 (39)5.2.知识建设 (39)5.2.1 语言知识库构建 (40)5.2.2 业务知识库构建 (41)5.3.二次开发 (42)5.4.系统部署 (44)5.4.1 常规部署 (45)5.4.2 扩展部署 (45)5.4.3 集群部署 (47)版权声明版权所有© 2013-2015上海智臻网络科技有限公司。
本文档的内容,所有文本全部或部分均受版权保护,上海智臻网络科技有限公司是本文档所有版权作品的拥有者。
除非预先得到本公司的书面授权,否则严禁对本文档进行复制、改编、翻译、发布等等。
本文信息如有变动,恕不另行通知。
白皮书:PERC Point-to-Point Resistance(P2P resistance)

WHITE PAPER Introduction PERC Point-to-point resistance (P2P resistance) functionality is a crucial EDA technology to enable complex P2P effective resistance measurement along ESD paths in automation forfoundry qualified ESD/Latch-up checker or in-house custom checker. This technology is appliedto the entire chip, block, and IP designs on cell or transistor level layout database. Since theESD path count could grow to thousands or even ten thousand, it is vital that the ESD path-oriented R extraction and distributed matrix solving capability has outstanding performance.This technology does not estimate P2P resistance using shortest or longest path schemes,but instead uses an accurate simulation focused on the critical layout polygons of ESDpaths, including the P/G network. The P2P resistance measurement yields result in effectiveresistance (ohms) for each path measured. That result is a lumped resistance value reflecting allinterconnect polygon layers between Source (current injection) and Sink (current Sink). Althoughthe P2P resistance value gives users immediate information on how effective the ESD pathbehaves in discharging an ESD surge, a violated P2P resistance value alone is difficult for layoutengineer to act on for any layout fix. When these checks fail, the resistance is reported for thefailing source/sink pair, but a single resistance value does not guide how to fix it. A physical layerchange will need to occur to fix it, but that single value provides no guidance on where to lookand what to do, and that is especially problematic on complex paths that can span dozens ofphysical layers. Ultimately this can lead to design delays and even failing silicon. IC Validator PERCIC Validator™ PERC is part of the more prominent IC Validator physical verification solution. ICValidator provides industry-leading solutions for DRC, LVS, FILL, pattern matching, and manyother applications.IC Validator PERC leverages StarRC™ for R-extraction and Python for a rich programmingenvironment, and together those are the technology backbone for the flow. That flow can thenbe used for netlist checks, netlist driving layout DRC checks, P2P , and current density.AuthorsFrank FengDir, Business Development,Synopsys Jonathan White Dir, Applications Engineering, SynopsysDebugging Point-to-Point Resistance Using Contribution by Layer in IC Validator PERCIC ValidatorFigure 1: IC Validator Physical Verification SolutionIC Validator PERC is qualified for significant foundry ESD/LUP checking of P2P resistance measurement, even at the full-chip level. IC Validator PERC P2P flow employs StarRC for R extraction, which is the industry gold standard. To provide better P2P resistance analysis for the layout engineer to act when there is a P2P resistance violation, IC Validator PERC offers a distinctive featureto analyze P2P resistance result contribution by layer. This feature enables beneficial information for the user to decide which interconnect layers are high contributors to be the candidates for layout fix.Enabling IC Validator PERC Collecting Database for P2P Resistance Contribution by Layer To use this debug capability, R reduction will need to be turned off in the P2P resistance flow. The reason is that reduction greatly simplifies the R network, and information about the actual fractured layout polygons is lost. This control is in the StarRC tech file, and the IC Validator PERC flow has a mechanism to define rules to PERC that then get passed into StarRC. The “readme” for your foundry runset for P2P resistance would provide the information needed to do this.Performance Impact of Producing Data for P2P Resistance Contribution by LayerIf the user is only running PERC P2P on IO nets and other non-P/G nets, then enabling P2P resistance contribution by layer will have minimal impact on performance. However, if the user is running PERC P2P, including P/G nets, then the user should expecta performance impact, depending on the size of the design. This performance impact is insignificant if the design is small (chip size is less than a few mm^2). If the chip size is larger than 10 mm^2, the performance cost will be more significant due to StarRC not reducing the R network of the P/G nets. The chip size estimation for performance cost is a rough guide to keep in mind andnot a hard rule.Using IC Validator VUE to Access P2P Resistance Contribution by Layer in Conjunction with Layout HighlightingThe user can launch IC Validator PERC job as usual. Upon the successful completion of IC Validator PERC P2P job, the user will be able to analyze P2P resistance contribution by layer in IC Validator VUE together with a layout viewer supported by IC Validator VUE. In the below section, the debug process in conjunction with P2P resistance contribution by layer is described. The user starts a layout viewer such as IC Validator WorkBench, Virtuoso, or Custom Compiler™. With the layout database open, the user invokes IC Validator VUE and loads topcell.vue file. Select “PERC Errors” tab to point to IC Validator PERC P2P run results; the violation P2P paths associated with each check name are listed on the Violation Browser page (on the left panel of VUE main window). Select one of the paths, and more details are shown as Violation Details/Description (on the right panel of VUE main window). Select the top line for each path in the Violation Details panel, right-click to drop-down list of can-do function, select (left click) on “PERC Path Heatmap,” and then a “Highlight Path” dialog window pops out. In the top portion of “Highlight Path” dialog window, there are list of symbols. Select the rightmost symbol (looks like a table), the P2P resistance contribution by layer table named “Contributionto Path Resistance” is shown. Each of these steps in the VUE debug procedures to access P2P resistance contribution by layer is displayed in figure 2.Figure 2: A flow chart describes accessing P2P resistance contribution by layer in VUE.The P2P resistance contribution by layer data (named as Contribution to Path Resistance table) provides how the lumped total resistance of the selected P2P path is summed up from various layers. Since each layer has its sheet resistance, layout polygons alone can’t tell the user what to do. The top contributors of resistance combined with layer polygon highlight capability in the “Highlight Path” dialog window give the layout engineer a much better idea of what to do for a layout fix. Figure 3 shows one P2P path measured from ESD diodes one physical power Pad of a power net.©2021 Synopsys, Inc. All rights reserved. Synopsys is a trademark of Synopsys, Inc. in the United States and other countries. A list of Synopsys trademarks is availableat /copyright .html . All other names mentioned herein are trademarks or registered trademarks of their respective owners.Figure 3: Pictures show how P2P resistance contribution by layer table provides valuable information for a user to focus debug and layout fix priorityInterpreting Contribution by Layer Results to Fix Design IssuesFixing P2P resistance issues in a design can be a complex problem. Proper fixing depends most heavily on the designer’s knowledge of their design and what changes they can make to resolve it. Additional information like a contribution by layer is intended to help understand the results more effectively so that the designer can apply their design knowledge with more confidence and greater speed. So, for example, knowing that the top thick metal represents 70% of the P2P resistance contribution does show what to change with that metal routing. But it does indicate quickly to a design whether the results are as expected or whether something unusual has occurred. And it gives confidence for the designer to make changes to the top think metal knowing that there will not be unintended consequences for that changes.SummaryPoint-to-point resistance checking is an essential component of robust ESD design verification. However, debugging the reported errors can be a real challenge and frustration to ESD engineers. IC Validator PERC provides the “contribution by layer” feature in its P2P Heatmap interface to IC Validator VUE, which offers tremendous insight into fixing these critical design errors. This saves time in the design cycle and gives higher confidence going into silicon ESD testing.。
华为FusionServer 5288 V3技术白皮书

华为 FusionServer 5288 V3技术白皮书文档版本06发布日期2016-10-31版权所有 © 华为技术有限公司 2016。
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:技术白皮书目录目录1 产品概述 (1)2 产品特点 (3)3 逻辑结构 (6)4 硬件描述 (7)4.1 外观 (7)4.2 接口 (12)4.3 指示灯和按钮 (12)4.4 物理结构 (15)5 产品规格 (20)6 部件兼容性 (25)6.1 CPU (25)6.2 内存 (28)6.3 存储 (34)6.4 IO扩展 (39)6.5 电源 (49)6.6 支持的操作系统和软件 (50)7 系统管理 (56)8 维保 (58)9 物理环境规格 (59)10 通过的认证 (61)1产品概述FusionServer 5288 V3服务器(以下简称5288 V3)是华为全新一代4U双路或单路存储型机架服务器,凭借高效设计,在确保卓越计算性能的同时,提供灵活、超大容量的本地存储扩展能力,适用于冷数据存储、视频监控、云存储、大数据等应用场景,在媒资、金融、公安等领域将广泛应用。
5288 V3有以下四种不同的配置。
l5288 V3(24块硬盘配置,支持配置单CPU或双CPU)支持24块前置的3.5英寸SAS/SATA/SSD硬盘,可选配4块内置的3.5寸或4块2.5寸SAS/SATA/SSD硬盘,需要配置1块SAS卡或SAS RAID控制卡。
P6 Professional Cloud Connect 白皮书 - 版本8.4说明书

P6 Professional Cloud Connect White Paper - Release 8.4BENCHMARK DISCLAIMERBenchmark results are highly dependent upon a number of factors, including workload, specific application requirements, and system design and implementation. Results under other conditions may vary significantly. The platform-specific hardware and software requirements included in this document were current when this document was published. However, because new platforms and operating system software versions might be certified after this document is published, review the certification matrix on My Oracle Support. Oracle makes no warranty on system performance, price nor price/performance and you should not rely upon this performance report, nor on any of the information contained herein, when making capacity planning or product purchase decisions. Some jurisdictions do not allow the limitation or exclusion of liability. Accordingly, some of the above limitations may not apply to you.Table of ContentsBusiness Problem (4)Overview (5)Batch SQL (7)Brief Configuration Details (7)Business ProblemThe Oracle Primavera P6 Professional client has traditionally been used in a LAN environment. P6 Professional Cloud Connect Server enables P6 Professional for the Cloud or Enterprise WAN by:•Ensuring all communication between client and server are over standard Hypertext Transfer Protocol (HTTP or HTTPS) request/response calls.•Ensuring security over the Internet.•Compressing data to improve performance.The P6 Professional application loads most of its required data up-front to significantly improve the performance of ensuing transactional requests. The price of loading data is paid mostly during login and when opening projects. This approach has a number of advantages:• A responsive GUI for most purposes by eliminating or minimizing the need for repeated server requests for information.•The ability to perform complex calculations on the client machine, including scheduling, leveling, and applying actuals.The aggressive caching and potentially complex computations on the client side require the client machine to have a modern CPU (for example, an Intel Core i5 class or better) and enough memory (8 GB or more).A higher bandwidth and lower latency will naturally help with faster data load times. For example, for a 1 MB compressed response, the bandwidth required between the server and the end-user is simply 1 MB divided by the time that the end-user is willing to wait over and above the time it took the server to perform the computation. For example, a 240 Kbps DSL line would add about 30 seconds to the overall request and a 1 Mbps connection would add about 8 seconds. The latency, which is the time between the request dispatch and the arrival of the first byte of a response, adds an additional delay to each request.To reduce the network time spent on data-hungry requests, P6 Professional Cloud Connect Server compresses data. P6 Professional Cloud Connect Server can be used for P6 in the Oracle Cloud or for an on-premises installation with your EPPM server.P6 Professional Cloud Connect Server vs. CitrixCitrix is one solution for slow WANs currently proposed for Project Management client users, and it has been shown to be a viable alternative for many customers. Citrix provides the benefits of reduced network bandwidth requirements, however it can require additional resources for configuration and maintenance.P6 Professional Cloud Connect Server vs. Compression ServerThe following is a comparison between P6 Professional Cloud Connect Server and Compression Server:•P6 Professional Cloud Connect Server offers performance on a par with or better than Compression Server.•As an installable component of P6 EPPM, P6 Professional Cloud Connect Server runs on all platforms supported by P6 EPPM. Compression Server runs only on a32-bit Windows server platform.•P6 Professional Cloud Connect Server supports secure communication over HTTPS. Compression Server supports HTTP only.•Compression Server is approaching end of life and will be deprecated in 2016. OverviewFigure 1: P6 Professional Classic Client-Server ArchitectureFigure 2 illustrates the P6 Professional architecture client extended to connect to the P6 Professional Cloud Connect Server.Figure 2 : P6 Professional Client-Server Architecture with Cloud Connect ServerA typical scenario for the flow of data can be described as follows:1.The user logs into the application or opens a project.2. A sequence of HTTP requests are made by the client to the server. Individualrequests correspond to Client/Server SQL statements. A token correspondingto one of a known set of statements is embedded in the HTTP request alongwith any required parameters. Each request is individually encrypted.3.P6 Professional Cloud Connect Server receives these HTTP requests, de-cryptsand de-tokenizes the request, and then runs the SQL statement on behalf ofthe client.4.P6 Professional Cloud Connect Server receives the result set from thedatabase, encrypts and compresses the data and forms it into an HTTPresponse, and then sends the response across the WAN back to the client.5.The client decodes and decompresses the data as required.Note that P6 Professional Cloud Connect Server does not wait until the entire result set is obtained from the database. Rather, the data is compressed into blocks of a preset size and sent to the client, even as P6 Professional Cloud Connect Server is fetching additional rows of the same result set. This keeps the P6 Professional Cloud Connect Server memory footprint to a minimum, because it does not have to compile the entire result set into a huge block of compressed data. It also prevents the client from starving while P6 Professional Cloud Connect Server compiles a large result set.Batch SQLIn order to reduce network traffic, the communication protocol between the client and P6 Professional Cloud Connect Server supports packing of multiple SQL requests and dataset responses. Using this Batch SQL feature, the client can further reduce the adverse impact of network latency and achieve improved throughput.Brief Configuration DetailsYou can install P6 Professional Cloud Connect Server as a product component of P6 EPPM hosted in the same WebLogic server instance as the server for P6 EPPM, but as a different managed server. The P6 Professional Cloud Connect Server URL can be configured to use the same hostname and port as P6 Web or it can use a different hostname and port. P6 Professional Cloud Connect Server has a separate application context from the server for P6 EPPM. P6 Professional Cloud Connect Server is configured by the P6 EPPM Configuration Wizard.Similarly, you can install the API Apps (Update Baseline, Schedule Compare, and XML Import) as product components of P6 EPPM configured by the P6 EPPM Configuration Wizard. Refer to Installing and Configuring P6 EPPM for installation instructions. After installation, click Use Remote Mode for P6 Professional Update Baseline, Schedule Compare, and XML Import and set the Integration API Server URL on the Application Settings page of the P6 EPPM server. Also, refer to the P6 EPPM Post Installation Administrator's Guide for other administration options.P6 Professional users configure their connection to the P6 database using the P6 Professional Cloud Connect driver as shown in the Figure 3:Figure 3: Configuring P6 Professional Cloud Connect ServerDocument References。
华为 CloudEngine 系列 TRILL技术白皮书

虚拟机任意迁移
作为云计算的核心技术之一,服务器虚拟化已经得到越来越广泛的应用。为了更大 幅度地增大数据中心内业务可靠性、降低 IT 成本、提高业务部署灵活性、降低运 维成本高,需要虚拟机能够在整个数据中心范围内进行动态迁移,而不是局限在一 个汇聚或接入交换机范围内进行迁移。
文档版本 01
华为专有和保密信息
1
版权所有 © 华为技术有限公司
TRILL 技术白皮书
1 TRILL
规模受限于 VLAN 数量限制,最多只支持 4096 的规模。随着云计算的发展,未来 数据中心网络架构租户规模要求能够突破 VLANID 的限制。
网络规模大、具有可扩展性
对于云计算时代下的大型数据中心来说,支持的服务器要能够达到十万甚至百万级 别,为了实现无阻塞转发,网络规模要能够达到几百台甚至上千台交换机,在这种 大规模组网情况下,组网协议要能够有效避免环路。网络内部的节点和链路故障, 要能够触发整网快速收敛,业务迅速恢复。网络维护简单,方便用户业务部署。另 外,为了适应数据中心高速发展的需要,数据中心网络需要具有良好的可扩展性。
TRILL 技术白皮书
文档版本 发布日期
01 2013-04-09
华为技术有限公司
版权所有 © 华为技术有限公司 2013。 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。
传统二层网络由于以太报文头部没有 TTL 字段,xSTP 协议收敛机制设计的比较保 守,在网络拓扑变化情况下,收敛速度比较慢,有的情况下甚至需要几十秒时间才 能收敛,不能满足数据中心业务高可靠性要求。TRILL 采用路由协议生成转发表项, 并且 TRILL 头部有 Hop-Count 字段能够允许短暂的临时环路,在网络出现节点和 链路故障情况下收敛时间能达到亚秒级。
sysfonts 0.8.8 软件说明说明书

Package‘sysfonts’October14,2022Type PackageTitle Loading Fonts into RVersion0.8.8Date2022-03-13Author Yixuan Qiu and authors/contributors of theincluded fonts.Seefile AUTHORS for details.Maintainer Yixuan Qiu<*******************>Description Loading system fonts and Google Fonts<https:///>into R,in order tosupport other packages such as'R2SWF'and'showtext'.Suggests curl,jsonliteCopyright seefile COPYRIGHTSSystemRequirements zlib,libpng,FreeTypeURL https:///yixuan/sysfontsBugReports https:///yixuan/sysfonts/issuesLicense GPL-2RoxygenNote7.1.2NeedsCompilation yesRepository CRANDate/Publication2022-03-1313:40:02UTCR topics documented:font_add (2)font_add_google (4)font_families (6)font_families_google (7)font_files (8)font_info_google (9)font_paths (10)Index111font_add Add New Font Families to’sysfonts’DescriptionThe two versions of this function are equivalent,but the"underscore"naming is preferred.This function registers new font families that can be used by package showtext and the SWF device in package R2SWF.Currently supported formats include but not limited to TrueType fonts(*.ttf, *.ttc)and OpenType fonts(*.otf).Usagefont_add(family,regular,bold=NULL,italic=NULL,bolditalic=NULL,symbol=NULL)font.add(family,regular,bold=NULL,italic=NULL,bolditalic=NULL,symbol=NULL)Argumentsfamily a character string of maximum200-byte size,indicating the family name of the font.See"Details"for further explanation.regular path of the fontfile for"regular"font face.This argument must be specified asa character string and cannot be missing.bold path of the fontfile for"bold"font face.If it is NULL,the function will use the value of argument regular.italic,bolditalic,symboldittoDetailsIn R graphics device,there are two parameters combined together to select a font to show text.par("family")is a character string giving a name to a series of font faces.Here series implies that there may be different fonts with the same family name,and actually they are distinguished bythe parameter par("font"),indicating whether it is regular,bold,or italic,etc.In R,par("font") is an integer from1to5representing regular,bold,italic,bold italic,and symbol,respectively.In sysfonts package,there are three default font families,sans,serif,and mono,each withfive font faces as mentioned above.If one wants to use other font families,the function font_add() needs to be called to register new fonts.Note that the family argument in this function can be an arbitrary string that does not need to be the real font name.The specified family name will be used in functions like par(family="myfont")and text("Some text",family="myfont").The Examples section shows a complete demonstration of the usage.Tofind the fontfile of argument regular(and the same for other font faces),this function will first check the existence of the specified path.If not found,file will be searched in the directories returned by font_paths()in turn.If thefile cannot be found in any of the locations,an error will be issued.ValueA character vector(invisible)of currently available font family names.Author(s)Yixuan Qiu<https://statr.me/>See AlsoSee par()for explanation of the parameters family and font.Examples##Not run:##Example:download the font file of WenQuanYi Micro Hei,##add it to SWF device,and use it to draw text in swf().##WenQuanYi Micro Hei is an open source and high quality##Chinese(and CJKV)font.wd=setwd(tempdir())ft.url="/projects/wqy/files/wqy-microhei"ft.url=paste(ft.url,"0.2.0-beta/wqy-microhei-0.2.0-beta.tar.gz",sep="/")download.file(ft.url,basename(ft.url))##Extract and add the directory to search pathuntar(basename(ft.url),compressed="gzip")font_paths("wqy-microhei")##Register this font file and assign the family name"wqy"##Other font faces will be the same with regular by defaultfont_add("wqy",regular="wqy-microhei.ttc")##A more concise way to add font is to give the path directly,##without calling font_paths()#font_add("wqy","wqy-microhei/wqy-microhei.ttc")##List available font familiesfont_families()if(require(R2SWF)){##Now it shows that we can use the family"wqy"in swf()swf("testfont.swf")##Select font family globallyop=par(family="serif",b=2)##Inline selecting fontplot(1,type="n")text(1,1,intToUtf8(c(20013,25991)),family="wqy",font=1,cex=2)dev.off()swf2html("testfont.swf")}setwd(wd)##End(Not run)font_add_google Load Google Fonts into’sysfonts’DescriptionThe two versions of this function are equivalent,but the"underscore"naming is preferred.This function will search the Google Fonts repository(https:///)for a spec-ified family name,download the proper fontfiles,and then add them to sysfonts.This function requires the jsonlite and curl packages.Usagefont_add_google(name,family=name,regular.wt=400,bold.wt=700,repo="/",db_cache=TRUE,handle=curl::new_handle())font.add.google(name,family=name,regular.wt=400,bold.wt=700,repo="/",handle=curl::new_handle())Argumentsname name of the font that will be searched in Google Fontsfamily specifies the family name of this font in R.This can be any string,not neces-sarily the same as name.The value of this parameter will be used in R plottingfunctions.See the example code below.regular.wt font weight for the regular font face,usually400bold.wt font weight for the bold font face,usually700repo the site that hosts the fontfiles.Default is the official repository / provided by Google Fonts.db_cache whether to obtain font metadata from a cache ing cache is typicallyfaster,but not as update-to-date as using the official API.If db_cache is set toFALSE,then metadata are retrieved from the Google Fonts API.handle a curl handle object passed to curl::curl_download().DetailsThere are thousands of open source fonts in the Google Fonts repository(https://fonts.google.com/).This function will try to search the font family specified by the name argument,and thenautomatically download the fontfiles for all possible font faces("regular","bold","italic"and"bolditalic",but no"symbol").If fonts are found and downloaded successfully,they will be also added tosysfonts with the given family name.Author(s)Yixuan Qiu<https://statr.me/>See Alsofont_families_google()Examples##Not run:font_add_google("Alegreya Sans","aleg")if(require(showtext)){wd=setwd(tempdir())pdf("google-fonts-ex.pdf")showtext_begin()par(family="aleg")6font_families plot(0:5,0:5,type="n")text(1:4,1:4,"Alegreya Sans",font=1:4,cex=2)showtext_end()dev.off()setwd(wd)}##End(Not run)font_families List Font Families Loaded by’sysfonts’DescriptionThe two versions of this function are equivalent,but the"underscore"naming is preferred.This function lists font families currently available that can be used by R2SWF and showtext packages.Usagefont_families()font.families()DetailsBy default there are three font families loaded automatically,i.e.,"sans","serif"and"mono".If one wants to use other fonts,font_add()needs to be called to register new fonts by specifying a family name and corresponding fontfiles.See font_add()for details about the meaning of"family name"in this context,as well as a complete example of registering and using a new font.ValueA character vector of available font family names.Author(s)Yixuan Qiu<https://statr.me/>See Alsofont_add()Examplesfont_families()font_families_google7 font_families_google List Font Families Available in Google FontsDescriptionThe two versions of this function are equivalent,but the"underscore"naming is preferred.This function lists family names of the fonts that are currently available in Google Fonts.Whenrunning this function for thefirst time,it may take a few seconds to fetch the font informationdatabase.This function requires the jsonlite and curl packages.Usagefont_families_google(db_cache=TRUE,handle=curl::new_handle())font.families.google()Argumentsdb_cache whether to obtain font metadata from a cache ing cache is typicallyfaster,but not as update-to-date as using the official API.If db_cache is set toFALSE,then metadata are retrieved from the Google Fonts API.handle a curl handle object passed to curl::curl_download()and curl::curl_fetch_memory().ValueA character vector of available font family names in Google Fonts.Author(s)Yixuan Qiu<https://statr.me/>See Alsofont_add_google()Examples##Not run:font_families_google()##End(Not run)8font_files font_files List Font Files Available in the Search PathsDescriptionThe two versions of this function are equivalent,but the"underscore"naming is preferred.This function lists fontfiles in the search path that can be loaded by font_add().Currently sup-ported formats include TrueType fonts(*.ttf,*.ttc)and OpenType fonts(*.otf).Usagefont_files()font.files()ValueA data frame containing the following information of the fontfiles:path The directory that the fontfile is located in.file File name of the font.family Family name.face Font face.version Version of the font.ps_name PostScript font name.Author(s)Yixuan Qiu<https://statr.me/>See Alsofont_paths(),font_add()Examples##Not run:font_files()##End(Not run)font_info_google9 font_info_google Display Information of Available Google FontsDescriptionThis function returns a data frame that contains the metadata of font families available in GoogleFonts,for example the family name,available font face variants,the version number,etc.Whenrunning this function for thefirst time,it may take a few seconds to fetch the database.Thisfunction requires the jsonlite and curl packages.Usagefont_info_google(db_cache=TRUE,handle=curl::new_handle())Argumentsdb_cache whether to obtain font metadata from a cache ing cache is typicallyfaster,but not as update-to-date as using the official API.If db_cache is set toFALSE,then metadata are retrieved from the Google Fonts API.handle a curl handle object passed to curl::curl_download()and curl::curl_fetch_memory().ValueA data frame containing metadata of Google Fonts.Author(s)Yixuan Qiu<https://statr.me/>See Alsofont_families_google()Examples##Not run:font_info_google()##End(Not run)10font_paths font_paths Get/Set Font Search PathsDescriptionThe two versions of this function are equivalent,but the"underscore"naming is preferred.This function gets/sets the search paths for fontfiles.See font_add()for details about how sys-fonts looks for fontfiles.There is also a complete example showing the usage of these functions in the help page of font_add().Usagefont_paths(new)font.paths(new)Argumentsnew a character vector indicating the search paths to be prepended.If the argument is missing,the function will return the current search paths.DetailsDefault search paths will be assigned when package is loaded:•For Windows,it is%windir%\Fonts,usually expanded into C:\Windows\Fonts•For Mac OS,default paths are/Library/Fonts and~/Library/Fonts and their subdirecto-ries•For Linux and other Unix-like OS,/usr/share/fonts,/usr/local/share/fonts,~/.fonts, ~/.local/share/fonts,and their subdirectoriesValueThe updated search paths.Author(s)Yixuan Qiu<https://statr.me/>Indexfont.add(font_add),2font.add.google(font_add_google),4font.families(font_families),6font.families.google(font_families_google),7font.files(font_files),8font.paths(font_paths),10font_add,2,6,8,10font_add_google,4,7font_families,6font_families_google,5,7,9font_files,8font_info_google,9font_paths,3,8,10par,311。
NF8260M5 产品技术白皮书说明书

NF8260M5产品技术白皮书文档版本2.0发布日期2019-07尊敬的用户:版权© 浪潮2018. 版权所有未经事先书面同意,本文档的任何部分不得复制或以任何形式或任何方式修改、外传注:您购买的产品、服务或特性等应受浪潮集团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,浪潮集团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。
Inspur和“浪潮”是浪潮集团的注册商标。
Windows是微软公司的注册商标。
Intel、Xeon是Intel公司的注册商标。
其他商标分别属于其相应的注册公司。
技术服务电话:4008600011地址:中国济南市浪潮路1036号浪潮电子信息产业股份有限公司邮编:250101版本控制目录目录 ............................................................................................................................................... i v1产品概述 (8)2产品特点 (9)3逻辑架构 (11)4产品规范 (12)4.1前视图 (12)4.1.1前面板正视图 (12)4.1.2前控板及LED指示灯 (12)4.1.3硬盘托架指示灯 (14)4.2后视图 (15)4.2.1PCIE Riser配置后视图 (15)4.2.2GPU配置后视图 (16)4.3内部俯视图 (17)4.4PCIe转接卡 (18)4.5主板布局图 (21)5系统规格 (23)6兼容性列表 (25)6.1处理器 (25)6.2内存 (25)6.3存储 (26)6.4RAID/SAS 卡 (29)6.5I/O 扩展 (30)6.5.1网卡 (30)6.5.2FC HBA 卡 (32)6.5.3HCA 卡 (32)6.5.4PCIE SSD (32)6.6电源 (33)6.7操作系统 (33)7配置注意选项 (36)8系统管理 (37)9物理规格 (40)10认证 (41)11支持与服务 (42)12新技术点描述 (43)12.1Intel 可扩展架构 (43)12.2Intel VROC 技术 (43)13相关文档 (44)14商标 (45)图表1 NF8260M5视图 (8)图表2 NF8260M5逻辑框图 (11)图表3 NF8260M5前视图 (12)图表4 NF8260M5前面板组件说明 (12)图表5 NF8260M5前控板组件 (13)图表6 NF8260M5前控板组件说明 (14)图表7 硬盘托架视图 (14)图表8 硬盘托架指示灯说明 (14)图表9 PCIE Riser配置后视图 (15)图表10 PCIE Riser后视图说明 (16)图表11 GPU配置后视图 (16)图表12 GPU配置后视图说明 (16)图表13 PCIE Riser配置内部俯视图 (17)图表14 PCIE Riser配置内部俯视图 (17)图表15 GPU配置内部俯视图 (18)图表16 GPU配置内部俯视图说明 (18)图表17 PCIEX8*3板卡视图 (19)图表18 PCIEX16*2+PCIEX8板卡视图 (19)图表19 PCIEX16+PCIEX8板卡视图 (19)图表20 PCIEX16+PCIEX8板卡视图 (20)图表21 PCIEX16*1板卡视图 (20)图表22 主板布局图 (21)图表23 主板布局说明 (22)图表24 产品规格 (24)图表25 内存规格支持 (25)图表26 内存兼容性列表 (26)图表27 HDD兼容性列表 (27)图表28 SSD兼容性列表 (27)图表29 NVME SSD兼容性列表 (28)图表30 M.2 SSD兼容性列表 (28)图表31 RAID卡兼容性列表 (29)图表32 Expander卡兼容性列表 (29)图表33 NVME RAIDkey兼容性列表 (29)图表34 PCIE Riser模组视图 (30)图表35 PCIE Riser规格组成 (30)图表36 OCP及PHY卡兼容性列表 (31)图表37 PCIE外插网卡兼容性列表 (31)图表38 HBA卡兼容性列表 (32)图表39 HCA卡兼容性列表 (32)图表40 电源兼容性列表 (33)图表41 OS兼容性列表 (33)图表42 BMC智能管理列表 (39)图表43 物理规格 (40)1产品概述浪潮英信服务器NF8260M5产品保持了浪潮服务器一贯的高品质、高可靠的表现,在计算性能、可扩展性、智能管理等方面,实现极致创新与突破,特别适用于大规模兴起的对空间密度有较高要求的企业、互联网等大规模数据中心市场的网络负载均衡以及虚拟化场景的应用。
IBM WAS V8.0新特性介绍

IBM WebSphere Application Server V8.0 中的新特性引言WebSphere Application Server 产品从V6.1 开始,以功能部件包(Feature Pack) 的形式向客户提供新的功能,而下一个大版本都会融合上一个版本Feature Pack 提供的功能,并对其进行进一步改进。
举例来说,WebSphere Application Server V6.1 有三个功能部件包,其中的EJB 3.0 和Web Services 功能部件包被集成到WebSphere Application Server V7.0 中,并且其功能进一步得到增强。
WebSphere Application Server V7.0 有六个Feature Pack,其中CEA, XML, SCA, OSGi 应用和JPA 2.0 和Modern Batch 这五个功能部件包的功能被集成到了V8.0 版本中,并对其进行了改进。
而Web 2.0 & Mobile 和Dynamic Scripting 这两个编程模型以WebSphere Application Server V8 的功能部件包的形式提供给用户。
与IBM Rational Application Developer 开发工具结合使用,WebSphere Application Server V8.0 支持Java EE 6、OSGi 应用、Mobile、SCA、SIP/CEA、XML、Batch 等应用的开发,成为名副其实的,为客户提供了最广泛的编程模型的应用服务器(参看图1)。
图 1. WebSphere Application Server 支持的编程模型IBM 提供了一款免费的应用迁移工具(Application Migration Toolkit) [2],支持将您的Java EE 应用快速地迁移到WebSphere Application Server V8.0 上来。
- 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)。
OTII定制服务器技术方案及行动计划书
目录
1概述 (3)
2技术需求及面临的挑战 (3)
2.1业务需求 (4)
2.2边缘机房环境挑战 (4)
2.3运维管理需求 (5)
3解决方案 (5)
3.1配置规格及关键组件 (5)
3.2物理形态、供电及环境适应性 (6)
3.3BIOS、BMC及硬件管理 (6)
4阶段性成果 (6)
5未来行动计划 (7)
1概述
网络转型是运营商发展的必由之路。
从内生原因来看,随着运营商网络流量激增、业务多样化加快,传统网络设施架构越来越无法满足业务发展的需要,业务需求的多样化与个性化要求网络更加灵活和弹性;从外部原因来看,长期以来互联网成功探索了低成本、高效率技术路线,例如开源软件、开放硬件,以及软件定义和云化架构等,这些正在冲击着运营商的网络价值和利润空间。
为应对这一挑战,中国移动提出NovoNet下一代网络发展愿景,以降低网络成本、实现业务敏捷高效。
随着运营商网络转型与重构的逐步推进,打造一个适应网络转型、满足未来电信应用需求的硬件架构平台尤为重要,从而以满足运营商愈加突显的核心网与接入网网络云化、以及新型MEC(Multi-Access Edge Computing,多接入边缘计算)的业务需求。
为此,2017年11月,中国移动联合中国电信、中国联通、中国信通院、英特尔等公司,在ODCC(Open Data Center Committee,开放数据中心委员会)共同发起了面向电信应用的开放IT基础设施项目——OTII(Open Telecom IT Infrastructure),首要目标就是形成运营商行业面向电信应用的深度定制、开放标准、统一规范的服务器技术方案及原型产品。
OTII项目得到了产业界的广泛关注,迄今为止已经得到传统电信设备、服务器、部件、固件和管理系统等领域的超过20家主流供应商的积极支持。
中国移动联合业界合作伙伴,开启了针对下一代网络业务的定制化服务器技术方案与产业生态的探索。
2技术需求及面临的挑战
运营商的机房分集团级、省级、地市、区县、接入(包括汇聚机房、综合接入机房、基站等)等众多层次,每个层次承载不同的5G及NFV网元业务。
例如核心网用户面网元多部署在地市机房;无线接入网RAN-CU功能虚拟化优先选
择在区县边缘机房;MEC业务覆盖区县、接入等各类机房。
2.1业务需求
不同的上层业务对底层硬件平台提出不同的技术需求,具体包括:
•无差别兼容上层多个VIM平台和VNF业务的需求。
硬件平台应尽量采用统一的设计和部件选型。
否则型号众多的服务器将带来大量的适配工作;•服务器性能需求。
不同网元的性能关注点有所差异,例如转发面网元对于网络带宽、转发时延和性能稳定性要求极高;
•时钟精度与同步要求。
对于部分涉及计费功能的网元应用,服务器需要具有较高的时钟精度,对于无线接入网元应用,服务器还需同时具有较高的时间同步精度;
•异构计算要求。
大量网元的虚拟化部署,如核心网和RAN CU虚拟化等,需要通过配置基于FPGA、ARM等的网卡或其它硬件加速方案卸载部分CPU功能,以节约CPU资源并提高处理效率。
2.2边缘机房环境挑战
边缘机房与核心机房相比条件较为恶劣,很多方面无法满足通用服务器的运行要求:
•机架空间有限。
传输及接入机房机架多为600mm深,部分能够达到800mm,通用服务器无法部署;
•部分机房环境温度较高。
由于制冷系统的稳定性无法有效保证,机房最高温度有时会达到45℃;
•机房空气质量欠佳;
•承重有限。
众多边缘机房并未按照数据中心标准建设;
•其它还包括抗震、供电等诸多限制。
边缘机房条件难以与大型数据中心等同,且数量庞大,所以单纯的改造机房并不现实,其中既有改造难度大造价高的原因,也有机房作为“战略资源“,很难自由扩展空间的因素,因此对服务器进行重新设计就是必经之路。
2.3运维管理需求
OTII边缘服务器承载大量电信级业务,并部署在较为恶劣的边缘机房,所以需要有强大管理运维能力保障:
•统一管理接口。
服务器需要有统一完善的管理接口要求,统一是因为多样化的管理接口将给VIM/PIM对接带来大量适配工作,完善是为了更加有效的管理服务器;
•运维高效。
边缘服务器应尽量降低对运维人员水平的要求,提高运维效率。
因此,服务器需要支持前维护,建议采用统一面板/指示灯,支持风扇热插拔等。
3解决方案
3.1配置规格及关键组件
在服务器的平台选型、主板设计、部件规格等方面,OTII项目结合运营商业务需求和面临的挑战,联合行业合作伙伴进行了一系列前期调研分析,初步技术方案如下:
在配置规格方面,核心网控制面网元对CPU、内存需求较高,同时机房环境相对较好,宜采用主流两路服务器;对于用户面网元、RAN侧网元、MEC等下沉到边缘的应用,负载以网络流量转发为主,从功耗、空间和性能需求等多方面考虑,倾向于单路低功耗方案。
例如采用英特尔至强D等SoC方案。
同时,考虑到电信业务多网络平面、网络加速和边缘异构计算等要求,也需要预留一定的扩展插槽。
在主板设计方面,对于两路CPU的配置,将采用NUMA Balance设计,以满足多PCIe设备应用环境下的性能及稳定性。
在部件规格方面,一是对网卡的性能、兼容性等有较高要求,可能需要推动
25G、100G网卡的应用以及生态的不断完善,同时加强对部件的选型要求或者形成比较严格的认证部件列表;二是对于网卡加速功能要求比较迫切,需将部分功能卸载至网卡,以提高网络处理速度并降低CPU负载,具体功能包括网络转发、IPSec、DPI和HQoS等。
3.2物理形态、供电及环境适应性
OTII边缘服务器不但需要适应边缘机房的环境,还需要满足各类边缘业务在边缘机房的交付、部署与本地运维需求。
具体包括:
•为适应边缘数据中心空间限制和机架深度,服务器深度推荐不超过470mm,最多不超过500mm;
•开关、指示灯、硬盘、线缆等采用前维护,以提高维护效率,减少对机架后方空间的要求;
•风扇能够支持热插拔,保证在线清理或更换;
•部分边缘应用场景,可能需要支持在更宽的温度范围(例如-5度~45度)内运行,并可能需要满足B级EMC、抗震等需求;
•边缘数据中心功率和承重能力有限,对服务器密度要求不高,一般计算型服务器2U高度即可,存储型服务器可进一步放宽;但考虑边缘业务未来交付方便,可能会考虑“机框+多节点”的整体设计形态。
3.3BIOS、BMC及硬件管理
OTII项目将与服务器、BMC及FW厂商合作,开发统一的针对NFV场景的服务器硬件监控、远程管理功能,使上层管理平台能够无差别的与不同供应商、不同配置规格的服务器对接。
4阶段性成果
OTII项目成立以来,三家运营商持续在内部收集、分析业务需求,合作伙伴也提供了大量技术方案建议。
经过多次的沟通讨论,已经在配置规格、核心部件选型、主板设计和硬件形态等方面达成了许多共识,多项技术研究、开发和测试
验证工作正按计划逐步推进。
在这些前期工作的基础上,现已推出首款OTII深度定制服务器参考设计原型机。
此原型机总体上反映了OTII在配置规格、硬件设计和管理维护等方面的技术要求:470mm (D) x 434mm (W) x 87mm (H)的机箱尺寸,适应边缘机房的空间、供电条件,并满足高达45摄氏度的恶劣边缘环境要求。
在配置方面也具备较强的可扩展性,以单路设计为例,原型机最多可支持18核的处理器,512G内存,具备8个2.5寸盘位,并预留3个PCIe插槽。
原型机目前已经初步具备运行能力,后续将在此基础上,进一步完善和验证软硬件兼容能力、硬件管理能力等,并将结合实际5G、MEC、ORAN等业务测试,进一步明确后续的配置规格和部件选型。
5未来行动计划
5G是运营商网络转型和新业务发展的重要契机,核心网、接入网的云化和边缘计算的兴起,都对以IT化为标志的新一代网络基础设施提出了新的要求,也创造了新的市场空间。
中国移动在全球5G引领中发挥着重要作用,从标准、技术、产业等方面多点发力,全面突破。
目前,中国移动在现网多省开展5G、MEC和NFV等应用试点,取得了不少成果与进步,当然在诸多问题上还有待进行攻关突破。
为此,OTII项目联合产业界制定了未来行动计划:
•2018年6月,推出OTII服务器原型设计方案,推动结合实际NFV、MEC等业务进一步开展测试验证、优化现有方案;
•2018年10月份,在ODCC峰会向产业界发布OTII服务器阶段性技术研究和测试成果;
•2019年,基于英特尔新一代平台和前期配置规格、部件选型和硬件管理等研究成果,正式启动OTII主板和硬件设计;
•2020年,实现OTII服务器规模应用,支撑5G业务发展。
OTII项目组希望与产业链一起,在面向电信应用的IT基础设施方面形成普遍共识与开放标准,打造能够更好满足未来电信应用需求、高效率低成本的通用硬件产品,为中国乃至全球运营商的网络转型提供范例、做出贡献。