13. www.huawei.com :32-35pt : R153 G0 B0 : FrutigerNext LT Medium : Arial :30-32pt : R153 G0 B0 : :20-22pt (2-5 ) :18pt : FrutigerNext LT Regular : Arial. - презентация

1 13. :32-35pt : R153 G0 B0 : FrutigerNext LT Medium : Arial :30-32pt : R153 G0 B0 : :20-22pt (2-5 ) :18pt : FrutigerNext LT Regular : Arial :18-20pt (2-5 ):18pt : Серверы Huawei- производительность и надежность

2 2 Huawei Cloud Congress 2013 Содержание История развития Обзор серверной линейки Основные преимущества

3 3 Huawei Cloud Congress Начало разработки 1-го поколения блэйд серверов Начало продаж блэйд серверов 1го поколения Т8000 Платформа ATCA (telecom) 1. Массовые внедрения 1G T Начало разработки 2G Запуск 2го поколения T Начало разработки rack серверов. 1. Запуск серверов RH1280 and RH Начало производства сервеных компонентов таких как сетевые карты, SSD и комплекты памяти 2.Массовые инсталляции 2G T Запуск сервера RH Начало разработки собственных серверных компонентов 1. Запуск E6000 RH1285/2285 2G 2. Новое поколение серверных компонентов и начало предоставления заказчикам высоко производительных датацентр решений Постоянные инновации, предложение более конкурентоспособной продукции Запуск Х6000 серверного решения для облачных вычислений 2. Запуск X8000 Более 1000 сотрудников занимающаяся разработкой и производством серверов Более 100 патентов и участие во многих организациях по стандартизации Богатый исследовательский и технический опыт в телеком,датаком сфере. Широкое сотрудничество с производителями оборудования и ПО, такими как Intel, AMD, SUSE и Redhat в технических областях, в том числе проектирования систем,разработки чипов и дизайна операционной системы 11 лет назад... История развития серверного направления в компании Huawei

4 4 Huawei Cloud Congress 2013 Мир изменяют данные 40 Total amount of Data of 40ZB in % X86 servers with virtualization supported 80 % Increase of unstructured data 2%2% Total power consumption of data centers The huge amount of data created is the largest challenge to data center network, storage, and servers. Data Generation Data Transfer Data Compute and Storage

5 5 Huawei Cloud Congress 2013 Проблемы серверного оборудования Low Utilization High Power Consumption Rapid Expansion Low Reliability Traditional Servers Source Michael Bell, research vice president at Stamford

6 6 Huawei Cloud Congress 2013 Input Output CPU Memory Hard Drive Magnetic Tape Compute Centric Depends on compute power Data stored on memory hierarchy Compute is far away from data, result in long latency and small bandwidth Just Enough Computing Just enough Compute power for each application CPU Multiple Storage Media Tiered Storage Converged Switching Technology Low Latency IO, supports virtualization acceleration Data Distributed compute nodes for computation Compute is closer to Data, which reduces latency Compute capability depends on location of data, types of data storage, and appliation CPU Lower TCO Higher Performance Data Centric От Compute Centric к Data Centric

7 7 Huawei Cloud Congress 2013 CPUCPU Super Computer CISC Mini Computer RISC System zBS2000 Power7SPARC T4/T5 X86 Servers Intel Xeon AMD Opteron Micro Servers Intel Atom ARM Cortex Itanium Memory Ethernet PCIe Shared IO QPI FC iSCSI, FCoE Increase BW Reduce Latency Cluster Interconnect Latency BW Gbps 40Gbps 00Gbps 128Gbps 25ns 500ns 2us 20us InfiniBand QDR/FDR/EDR 100GbE IO Interconnect Racetrack MRAM PCM RRAM Storage Процессор более не основной компонент

8 8 Huawei Cloud Congress 2013 Processors are moving toward specialization Performance per Watt is becoming more important Generic CPU + Application specific CPU = Heterogeneous CPU Heterogeneous CPUs can be more flexible, higher cost performance, and high power performance First used by storage systems Internet server begin small scale deployment from Enterprise server application still lag behind 3~5 years Просто добавь вычислений

9 9 Huawei Cloud Congress 2013 Рост твердотельных накопителей Another level in the memory hierarchy Hard Disk (Magnetic) Solid-State Disk (PCIE) Main Memory (Electrical) Shared L3 Cache L2 Cache D&IBuffer Solid-State Disk (PCIE) Solid-State Disk (SATA) Tape (Magnetic) CD DVD Blu-ray ROM (Optical) Tape (Magnetic) Hard Disk (Magnetic) CD DVD Blu-ray ROM (Optical) 100S 100mS 10mS 100uS CPU Clock 1.5T 20M 256K 64K Random Access Memory 5T 125G 4T 2.5T Serial Access Memory

10 10 Huawei Cloud Congress 2013 Ethernet supports DCB Ethernet supports RDMA Convergence of FC and Ethernet to FCoE Converged Switching of Infiniband, Ethernet, and FC: SwitchX Infiniband bandwidth still leads Ethernet, but the amount of lead is reducing, while FC is lagging far behind. Ethernet is absorbing many new features from other IO technology, and a Ethernet based converged network is forming. Конвергенция –будущее сетей

11 11 Huawei Cloud Congress 2013 Немного о виртуализации 65% WW workloads Virtualized Tier 1 Apps being virtualized NFV being promoted by Telcos More Memory per Server Higher Network BW per Server

12 12 Huawei Cloud Congress 2013 Автоматизация –путь к снижению TCO VM Sprawling, IT Management 65+% of IT Spending IT Automation critical to lower TCO Support for IT Automation: Stateless Computing – Physical Server Agentless Server Status Monitoring Bare metal Server Deployment Automation

13 13 Huawei Cloud Congress 2013 История развития Обзор серверной линейки Основные преимущества Content

14 14 Huawei Cloud Congress 2013 Стратегические направления Huawei Server Application acceleration card 2S RH22XX 4S RH5885 8S RH S ~ 64S RH8XXX E9000 X6000 X8000 HPC Private cloud Appliance Virtualization Mission-critical enterprise applications, virtualization, and midrange computer migration Massive data processing Web 2.0 application Public cloud Scale up Scale out Converged Cloud server iNIC Data compression card SSD card

15 15 Huawei Cloud Congress 2013 Полный набор технологий в одном вендоре The only global server vendor who develops chips, boards, cards, and builds up servers by oneself ASICs I/O Accelerator Architect Inter- connect Node Controller, BMC, etc. High performance interconnect technology for server products Huawei Server I/O Acceleration cards, help improve full system performance Indigenous R&D System level architecture design to improve performance and reliability

16 16 Huawei Cloud Congress 2013 Сервера высокой плотности 2S FW node XH320 V2 2S HW node1S HW node XH311 V2 800GB 1.2TB 2.4TB PCIe SSD card «Решения» Exchange ApplianceN2000 NAS RH1288 V2 X6000 2U 4S RH2488 V2RH2485 V2 4U 4S RH5885 V2 (4S) XH621 V2 RH5885 V2 (8S) XH310 V2 RH2285 V2 RH2288 V2 Стоечные сервера Блейд сервера BH620 V2 2*CPU(E5-2400) 12*DIMMs 4*2.5HDDs BH621 V2 2*CPU(E5-2400) 12*DIMMs 2*2.5HDDs 1*PCIe x8 BH622 V2 2*CPU(E5-2600) 24*DIMMs 2*2.5HDDs E6000 8U, 10 лезвий E9000 2S HW blade2S FW blade CH121 12U, 8/16 лезвий CH221CH220 2*CPU(E5-2600) 24*DIMMs 2*2.5 HDDs 4*PCIe x8 FHHL CH222 2*CPU(E5-2600) 24*DIMMs 15*2.5 HDDs 1*PCIe x8 FHHL NXseries 1 GE, 10GE, FC BH640 V2 4*CPU(E5-4600) 24*DIMMs 2*2.5HDDs CH240 4*CPU(E5-4600) 48*DIMMs 8*2.5 HDDs or 4*2.5 HDDs + 1*PCIe x16 FH3/4L 4S FW blade CX seres GE, 10GE, FCoE, FC, Infiniband QDR/FDR Switch module 2*CPU(E5-2600) 24*DIMMs 2*2.5 HDDs 1*PCIe x8 FHHL 2*CPU(E5-2600) 24*DIMMs 2*2.5 HDDs 2*PCIe x16 FHFL 2S blade4S bladeSwitch series 2U, 2/4 ноды 1*CPU(E V2) 4*DIMMs 1*3.5HDD 1*CPU(E V2) 4*DIMMs 4*2.5HDDs 2*CPU(E5-2400) 12*DIMMs 4*2.5HDDs 2*CPU(E5-2600) 24*DIMMs 8*2.5HDDs 1U 2S2U 2S8U 8S 2*CPU(E5-2600) 24*DIMMs 4*3.5HDDs or 8*2.5HDDs 2*CPU(E5-2400) 12*DIMMs 12*3.5+2*2.5 HDDs or 8*2.5 HDDs or 26*2.5 HDDs 2*CPU(E5-2600) 24*DIMMs 12*3.5+2*2.5 HDDs or 8*2.5HDDs or 26*2.5HDDs 4*CPU(E5-4600) 48*DIMMs 8*2.5HDDs 4*CPU(E7-4800) 32*DIMMs 6*3.5HDDs 4*CPU(E7-4800) 64*DIMMs 10*2.5HDDs 8*CPU(E7-8800) 128*DIMMs 16*2.5HDDs 12*3.5+2*2.5HDDs Supports CIFS, NFS, iSCSI, etc. Windows Storage Server users(3GB/user) 1200users(3GB/user) 3000users(5GB/user) Серверное портфолио

17 17 Huawei Cloud Congress TB2.4TB Flash granule Form factor Bus interface MLC Full-height, half-length PCIe 2.0 x8 MLC Full-height, half-length PCIe 2.0 x8 MLC Full-height, half-length PCIe 2.0 x8 Maximum read bandwidth Read IOPS (Stable value, 4 KB, 100% random) Read IOPS (Maximum value, 4 KB, 100% random) Read latency Maximum write bandwidth Write IOPS (Stable value, 4 KB, 100% random) Write IOPS (Maximum value, 4 KB, 100% random) Write latency 2.2 GB/s 570, , µs 1.2 GB/s 120, ,000 8 µs 3.2 GB/s 760, , µs 1.8 GB/s 180, ,000 8 µs 3.2 GB/s 760, , µs 2.8 GB/s 240, ,000 8 µs Mixed IOPS (4 KB, R/W: 7/3) Power consumption Weight Power-down protection Granules failure protection Supported trim 260,000 (R/W: 180,000/80,000) 20 W to 35 W 300g Yes 400,000 (R/W: 280,000/120,000) 25 W to 50 W 350g Yes 430,000 (R/W: 300,000/130,000) 25 W to 60 W 350g Yes HUAWEI Tecal ES3000 Высокопроизводительная карта SSD PCIe Карта SSD PCIe HUAWEI Tecal ES устраняет ограничения системы ввода/вывода и позволяет раскрыть потенциал любых серверов стандартной архитектуры. ES3000 поможет значительно сократить эксплуатационные затраты и TCO, вместе с тем ускорив работу ваших приложений. Емкость800 ГБ1,2 ТБ2,4 ТБ Тип памятиMLC Интерфейс шиныPCIe 2.0 x8 Максимальная проп. способность при чтении 2,2 ГБ/с3,2 ГБ/с IOPS при чтении (стабильное значение, 4 КБ, 100% случайно) 570,000760,000 IOPS при чтении (максимальное значение, 4 КБ, 100% случайно) 600,000770,000 Задержки при чтении 49 us IOPS при записи (стабильное значение, 4 КБ, 100% случайно) 120,000180,000240,000 IOPS при записи (максимальное значение, 4 КБ, 100% случайно) 330,000480,000630,000 Задержки при записи 8 us

18 18 Huawei Cloud Congress История развития Обзор серверной линейки Основные преимущества Content

19 19 Huawei Cloud Congress 2013 Полно управляемое решение Device Management platform Server Network Storage IT Infrastructure eSight 3 rd party software Centralized device management Enterprise management platform Hardware monitor Asset inventory Fault prediction Performance monitor Installation deployment Fault alarm Incident response Firmware Upgrade Fault response Service Life Cycle Operation Management Fault Management Remote control Provide 3rd party management software plug-in, to support a full management solution. Server Device Manager 1 Server Performance Reporter 2 Server resource Management; Remote maintenance, batch configuration deployment; Alarm system Server performance monitor; Analyze performance bottleneck, system optimization

20 20 Huawei Cloud Congress 2013 Huawei Tecal V2 series servers ranked first in 31 Standard Performance Evaluation Corporation (SPEC) test items. The RH2288 V2 rack server ranked first based on the E5-2620, E5-2630, E5-2630L, E5-2640, and E CPUs. The BH622 V2 server blade ranked first based on the E5-2665, E5-2660, and E CPUs. By now, Huawei servers have refreshed 55 world records in the SPEC test, proving excellent performance of Huawei servers…Huawei server Vision: High computing performance: 74 SPEC records in SPEC integer and floating-point computing performance Power-efficient: 9 records in the SPECpower_sj2008 benchmark Records increasing Huawei servers have refreshed 83 world records in the SPEC benchmark test by now. Производительность – 83 SPEC Benchmark Records

21 21 Huawei Cloud Congress 2013 Качество Supplier certification, component selection, and quality control EMC laboratory, vibration test laboratory, high temperature test laboratory, and safety regulation laboratory Huawei's fault-tolerant architecture provides comparable reliability as RISC midrange computers. Huawei has the design capability for architecture with % reliability for high-end products. Board design, server design, and the test process meet Integrated Product Development (IPD) standards. 35 Reliability, Availability, and Serviceability (RAS) features are supported. 150 patents on reliability designs are obtained. 200,000 carrier-class server blades have been delivered to telecom users. 300,000 rack servers, blade servers, and high-density servers have been delivered to enterprise users. Precise design Strict test Precise design Strict test 500,000 servers in large-scale use 500,000 servers in large-scale use Laboratories with sufficient equipment Outstanding reliability in both components and systems Laboratories with sufficient equipment Outstanding reliability in both components and systems High reliability Architecture design capability High reliability Architecture design capability

22 22 Huawei Cloud Congress 2013 Huawei servers utility strict telecommunication grade component derating during design phase, to make sure component will not fail under stress(electrical and thermal), also to delay component wear out, and extend component useful life. Высококачесвенные компоненты Component TypeDe-rating requirements Capacitor Operating voltage derating

23 23 Huawei Cloud Congress 2013 Архитектура повышенной надежности Standard server operating temperature: 10 ~35 Huawei server operating temperature: 5 ~40 Huawei Server can operate continuously at 5 ~40 Product design with consideration for high altitude, high temperature, high humidity, dusty, corrosive environments, Huawei is designing high reliability servers.

24 24 Huawei Cloud Congress 2013 Energy Efficiency DEMT* High efficiency thermal design High efficiency power supply design Power capping technology Hibernation & wakeup Note DEMT Dynamic Energy Management Technology Huawei proprietary technology, reduce power consumption when system is under utilized Энергоэффективный дизайн OS energy management & performance optimization

25 Copyright©2012 Huawei Technologies Co., Ltd. All Rights Reserved. The information in this document may contain predictive statements including, without limitation, statements regarding the future financial and operating results, future product portfolio, new technology, etc. There are a number of factors that could cause actual results and developments to differ materially from those expressed or implied in the predictive statements. Therefore, such information is provided for reference purpose only and constitutes neither an offer nor an acceptance. Huawei may change the information at any time without notice. Thank you