Title slide with picture · 2017. 10. 30. · Новый класс инфраструктуры...

HPE Synergy Передовая платформа для инфраструктуры ЦОД

Новый класс инфраструктуры для Экономики Идей

3

Опт

имиз

ация

при

лож

ений

Оптимизация процессов (скорость внедрения)

Повышение продуктивности • Частичное упрощение

процессов развертывания и сопровождения

• Предконфигурированные решения

• Специализированные нагрузки

Быстрый старт • Простые процессы

развертывания и сопровождения

• Программно определяемые системы хранения данных

• Виртуализация

Инфраструктура как услуга • Единая архитектура

оборудования и ПО • Полностью определяется

программно • Различные рабочие нагрузки:

физические, виртуализация, контейнеры

Разрозненная инфраструктура • Сложные процессы • Различные рабочие

нагрузки: физические, виртуализация, контейнеры

Традиционная

Конвергентная

Гиперконвергентная

Компонуемая

Компонуемость

Определение компонуемой инфраструктуры

Программно-определяемая логика

Унифицированные API

Компонуемая Инфраструктура

Унифицированные API • Единый код для абстрагирования любого

элемента инфраструктуры • Интерфейс для программирования

Инфраструктуры как Услуги

Программно-определяемая логика • Управление на основе шаблонов • Простое сопровождение инфраструктуры на протяжении всего жизненного цикла

Пластичные пулы ресурсов • Единая инфраструктура ресурсов вычислителей,

хранения данных и сетевых • Любые рабочие нагрузки: физические, виртуализация и

контейнеры

Пластичные пулы ресурсов

Выступающий

Заметки для презентации

The way we’re going to deliver this is through an architecture that has three key elements to it. Composable infrastructure starts with fluid resource pools, which is really a single structure that boots up ready for any workload with fluid pools of compute, storage, and fabric that can be instantly turned on and flexed. Second, it’s software defined intelligence, which means embedding intelligence into the infrastructure and using workload templates to tell it how to compose and re-compose and update on the fly in a very repeatable, frictionless manner. Third thing is taking of all these capabilities and exposing them through a unified API that allows infrastructure to be programmed like code so it can become infrastructure as a service. It can be served up to people who just want to write existing apps or code new apps and be able to derive infrastructure directly as code.

HPE BladeSystem

• Общая, модульная инфраструктура – простые дополнения/изменения

• Энергоэффективность – общее питание и охлаждение

• Унифицированное управление серверами, СХД и сетями

HPE BladeSystem Сети

СХД

Управление

Серверы



Blades share power, network, and storage infrastructure at the BladeSystem enclosure level. Since equipment is not needed for each server, there is a drastic reduction in power distribution units, power cables, LAN and SAN switches, connectors, adapters, and cables. You can bring in the newest generation technologies by simply changing the components that need to be changed instead of the “rip and replace” approach used for traditional infrastructures and that have led to the present sprawl. In addition, compared a traditional environment, the time needed for routine server administrative tasks with HP BladeSystem can be reduced by up to 15 percent. With HP Insight Control, you can provision servers quickly, manage health proactively, control servers from anywhere, and manage power confidently. With over 5 million software licenses shipped, Insight Control is the most broadly used systems management platform in the industry. Customer Segments suited for HP BladeSystem: Financial Telecom Web 2.0 HPC Hosting Enterprise SMB

Инфраструктура FlexFabric

Flex-10 Fibre Channel

Передовое решение по подключению блейд-серверов к сетям HPE Virtual Connect

– Модуль VC FlexFabric – 10Gb Ethernet, 8Gb Fibre Channel – Single/dual-hop FCoE

– Модуль VC Flex-10/10D

– 10Gb Ethernet – Single/dual-hop FCoE – Isolated or converged

traffic

– Модули VC Fibre Channel

– 16Gb FC – 8/4Gb FC – Native Fibre Channel

SAN – Isolated traffic

– Модуль VC FlexFabric-20/40 F8

– 10&40Gb Ethernet, 8Gb Fibre Channel – 20Gb server downlinks – Converged FC gateway and FlatSAN



KEY TAKEAWAYS: The Virtual Connect portfolio delivers solutions for many types of business models and networks including Ethernet, FC and converged FCoE and iSCSI. The family is now growing with new solutions to unlock new performance heights for BladeSystem customers. TRANSITION: The Virtual Connect family continues to expand and innovate to deliver new solutions which help customers transform their architecture and be agile to support the demands of the business. As HP moves forward to serve customer needs, HP is bringing new innovative solutions today to help customers be more agile and drive more performance for ther needs.

Программно-определяемое Управление серверами, СХД, сетями

Интеграция с партнерскими платформами

Основа для облачной модели

Быстрая доставка ИТ сервисов на основе SLA для ускорения перехода к гибридному облаку.

OneView Automation

Hub

Быстрое и надежное развертывание на основе инфраструктурных шаблонов.

65+ плагинов

Конвергентное управление – HPE OneView

Admin control

User control

Application control

Common Building Blocks

Software-defined management

Один подход для всей инфраструктуры: серверы, сети, СХД



Для того чтобы справиться с актуальными проблемами управления ЦОД мы предлагем: Конвергентную платформу управления для всех элементов ЦОД: серверов, СХД, сетей Программно-определяемые шаблоны, позволяющие быстро и надежно развертывать инфраструктуру для ИТ-сервисов Автоматизированное взаимодействие со средствами развертывания облачных сред Интеграцию с партнерскими платформами управления: VMware, Microsoft, Red Hat и др.

Сквозное управление и взаимодействие всех элементов инфраструктуры

Собранная из блоков Конвергентные блоки с атрибутами компонуемой инфраструктуры

Компонуемая инфраструктура HPE

Новая категория инфраструктуры для традиционного ИТ и Экономики Идей

Защита инвестиций в существующую инфраструктуру и ПО

Полностью компонуемая

ConvergedSystem ProLiant DL

BladeSystem

HPE OneView

3PAR

Hyper Converged 380

3PAR

StoreVirtual VSA

Synergy

HPE BladeSystem c-Class - производим и продаем до 2020 г. !

9

HPE Synergy

Компонуемые фабрики масштабируются на уровне стоек

Компонуемые вычислители различной производительности и конфигурации

Компоновшик (Composer) Программная платформа для управления инфраструктурой

Компонуемые системы хранения внешние, встроенные и программно-определяемые

Компонуемые шасси (Frame) Шасси для всех компонентов инфраструктуры Автоматическая интеграция на уровне стоек

HPE Synergy: новейшая инфраструктура для Экономики Идей

11



HPE Synergy, the first platform built from the ground up for Composable Infrastructure, offers an experience that empowers IT to create and deliver new value instantly and continuously. It is a single infrastructure that reduces operational complexity for traditional workloads and increases operational velocity for the new breed of applications and services. Through a single interface, HPE Synergy composes physical and virtual compute, storage, and fabric pools into any configuration for any application. As an extensible platform, it easily enables a broad range of applications and operational models such as virtualization, hybrid cloud, and DevOps. With HPE Synergy, IT can become not just the internal service provider but the business partner to rapidly launch new applications that become the business.

Вычислители HPE Synergy

HPE Synergy 480 Gen9

• Приложения для совместной работы

• Обработка контента • Бизнес-приложения • Инфраструктура • Высокопроизводительные

вычисления

HPE Synergy 660 Gen9

• Консолидация • Виртуализация • Базы данных • Бизнес процессинг • Системы принятия решений

HPE Synergy 620/680 Gen9

• Бизнес процессинг • Системы принятия решений • Базы данных • Высокопроизводительные приложения,

требующие максимального объема памяти

Intel® Xeon® E7-4800 v4 series 5 choices / 8 to 16 cores

Intel® Xeon® E7-8800 v4 series 6 choices / 4 to 24 cores

Intel® Xeon® E5-2600 v4 series 19 choices 4 to 22 cores

Intel® Xeon® E5-4600 v4 series 7 choices 10 to 22 cores



Driving a truly agile and efficient IT infrastructure requires the transformation of traditionally rigid physical systems into flexible physical or virtual resource pools. HPE Synergy Composable Compute creates a pool of flexible compute capacity that can be configured almost instantly to rapidly provision infrastructure for a broad range of applications. HPE Synergy supports both two-socket and four-socket compute modules which provide the performance, scalability, density optimization, storage simplicity, and configuration flexibility to power a variety of workloads, including business processing, IT infrastructure, web infrastructure, collaborative, and high-performance computing. Run anything: Optimize all apps and service levels A single pool of flexible resources: Disaggregated compute resources with the capacity to handle a broad range of applications The HPE Synergy 480 Gen9 Compute Module delivers superior capacity, efficiency, and flexibility to power more demanding workloads and increase virtual machine density by providing a full range of processor choices, right-sized storage options, and a simplified I/O architecture. It is designed to optimize general-purpose enterprise workload performance including business processing, IT infrastructure, web infrastructure, collaborative, and high performance computing in physical and virtualized environments while lowering costs within a Composable Infrastructure. The HPE Synergy 480 Gen9 Compute Module has increased memory capacity (up to 1.5 TB), twenty-four DIMM slots, and supports the entire Intel E5-2600 v3 processor family without any DIMM slot restrictions. Greater consolidation and efficiency are achieved through an increase in virtual machine density per compute module. The HPE Synergy 660 Compute Module handles data-intensive workloads with uncompromised performance and exceptional value. The HPE Synergy 660 Gen9 Compute Module is a full-height, high-performance with high-density Intel Xeon E5-4600 V3 processors, 48 DIMM slots providing up to three TB of available memory, flexible I/O fabric connectivity, and right-sized storage options. The HPE Synergy 660 offers a large memory footprint and powerful processors that required to support more demanding workloads like in-memory and structured databases. Its HPE DDR4 SmartMemory offers up to a 30%i performance increase over the previous generation. The HPE Synergy 620 and HPE Synergy 680 Gen9 Compute Modules are designed to meet the needs of almost any enterprise IT tier and workload. These two-socket and four-socket x86 compute modules are ideal for financial, insurance, healthcare, manufacturing, and retail enterprises that require mission-critical levels of availability, extended versatility, and real-time performance. HPE Synergy Composable Compute’s flexible design optimizes internal storage choices to match workload requirements. The compute module has a flexible front bezel design with two HPE SmartDrive bays, each of which supports up to two drives per bay or four micro form factor drives with built-in RAID with a choice of SAS, SATA, HDD, and flash storage. If internal storage is not required, such as environments using the HPE Synergy Image Streamer, there is an option to order compute modules with no drives for a stateless configuration. For workload acceleration, PCIe NVMe SSDs from HPE combine the high throughput and low latency of a storage workload accelerator with the flexibility and convenience of a standard form factor drive, ideal for web scaling, cloud, OLTP, Big Data, or business intelligence. HPE Synergy also supports software-defined data services and external attached storage options. The Smart Array controllers deliver superior reliability and increased compute module uptime with additional fault-tolerance (RAID) levels, online spare capability to facilitate automatic rebuilds after a drive failure, and fault prevention with predictive spare activation before a failure occurs. The compute modules converge traffic over high-speed 10/20 GB connections using HPE Converged Network Adapters. When connected to HPE Virtual Connect SE 40Gb F8 module, each module provides up to eight adjustable connections (seven data and one storage, or all data) to each 20 GB compute module port. Move faster: Accelerate app and service delivery Composable compute resources: Pooled compute resources provide the flexibility and performance needed accommodate a wide range of workloads With its infrastructure management simplicity and agility, HPE Synergy Composable Compute modules become a composable resource that is self-discovered, quickly provisioned, easily managed, and seamlessly redeployed. Freed of stranded compute resources, enterprise data centers can now deliver the right compute capacity for changing workload needs. Work efficiently: Reduce operational effort and cost Non-disruptive updates: Infrastructure changes are implemented automatically through template-based operations to reduce downtime, manual operations, and errors. HPE Synergy Composable Compute’s superior, enterprise-grade availability allows IT to quickly and confidently implement infrastructure changes. Change operations such as updating firmware are automatically implemented. Updates can be applied instantly for initial set up or staged to automatically take effect later. Staged updates copy the new firmware to compute modules where they are stored until a convenient maintenance window, at which time it is applied. The deployment of the firmware is completed with one-tool, one-step template-based operations and the changes can be timed with application needs, significantly reducing manual interactions and errors, and ensuring real-time compliance.

Графический сервер HPE Synergy ProLiant SY 480 Широкий выбор профессиональных графических карт

Высокопроизводительная 3D графика – HPE Multi MXM Expansion Module

– До 6 адаптеров NVIDIA Tesla M6, Quadro M3000SE и AMD FirePro S7100X

Ультра-производительная 3D графика – NVIDIA Quadro M5000

40 SSDs / на модуль (50,000 IOPs каждый)

Модуль хранения Synergy D3940 – Зонированный DAS

*4KB Random Read Workload using SSD. 2M IOPs is for a single storage module connected to multiple compute modules.

Два домена сети 12Gb SAS (24 порта, 4 линии на порт)

10 контроллеров P542D (200,000 IOPS+ на контроллер)

16



Detailed Explanation in backup slide 2M IOPS explained 1) Test was conducted using 4k random reads 2) Tested Config: Single D3940 storage module with 40 SSD drives and dual IO Adapters, dual 12Gb SAS Connection modules, 10 Synergy 480 Gen 9 compute modules each with a Smart Array P542D Controller. 3) The math explained: Each SSD can deliver up to 50,000 IOPS, thus we have a need to push 2 million IOPS from the storage module. This is done across a SAS connection module that offers 12 ports, each with 4 lanes of 12Gb SAS Each P542D controller has dual ports routed to the SAS fabric, each port also has 4 lanes. Each controller can handle ~1 million IOPS at maximum In the test conducted each compute module showed it was handling 200,000+ IOPS which proved that we were in fact fully evacuating all 2M IOPS from the storage module HPE Synergy with D3940 as DAS = 2M IOPs Dell PS-M4110 = 37,000 IOPs http://www.storagereview.com/dell_equallogic_ps6110xs_review http://www.storagereview.com/dell_equallogic_family_handson IBM V7000 = 120K IOPs http://www.storageperformance.org/benchmark_results_files/SPC-1/IBM/A00116_IBM_Storwize-V7000-SSDs/a00116_IBM_V7000-SSDs_SPC-1-executive-summary.pdf D2220sb = 200K IOPs using DAS

Оптимальное выделение дискового пространства

VM, VDI, Exchange Базы данных Hadoop, Аналитика Максимальная емкость

12

Общий пул ресурсов хранения для всех вычислителей в шасси

40 SFF Drives 120 SFF Drives 160 SFF Drives

На сервер: 500K IOPs и 10 SFF Всего: 4M IOPs и 307TB

80 SFF Drives

На сервер: 200K IOPs и 4SFF Всего: 2M IOPs и 153TB

На сервер: 1M IOPs и 20 SFF Всего: 6M IOPs и 460TB

На сервер: 1M IOPs и 40 SFF Всего: 4M IOPs и 612TB



HPE Synergy Composable Storage offers high-density storage options that span from fully integrated internal storage modules and software-defined storage to directly connected and fully orchestrated flash arrays. No matter the data type, connectivity protocol, or service level requirement, with HPE Synergy Composable Storage, you can respond effortlessly to unpredictable demands. As a key building block for many applications, HPE Synergy Composable Storage is designed to store and share anything — file, block and object data — with enterprise-class reliability. It delivers a new level of simplicity, density, and flexibility by eliminating the complexity and limitations of siloed resources and administration. HPE Synergy Composable Storage is composed along with compute and fabric resources to meet the demands of a wide range of data workloads, such as OS boot, database, collaboration (email/communication), file, web, and media streaming. And it is flexible enough to support virtualization, network file sharing, clustering, and object storage. Fully integrated composable storage configured with Synergy Composer Highest SFF drive count & density with 40 SFF bays per Storage Module High performance for SSD storage with 2M IOPs per Storage Module External SAN: Leading performance and scale with 3PAR StoreServ flash arrays

Программно определяемая система хранения HPE StoreVirtual VSA и HPE Synergy D3940

Общая СХД на несколько шасси

СХД на основе серверных узлов. ПО HPE StoreVirtual VSA объединяет узлы в единый кластер и создает на его основе сеть iSCSI;

Сервер приложений видит единый IP-адрес кластера, в то время как данные распределяются по всем узлам хранения;

Надежность пять девяток, дисковые и сетевые RAID

Линейное масштабирование до 240 ТБ и 224k IOPs, свободное перемещение данных между шасси, добавление новых узлов «на лету», адаптивная оптимизация



All-inclusive, Enterprise-class Rich data services Synchronous replication; Linear, federated scale-out; Peer Motion mobility; Sub-volume auto-tiering; Thin provisioning; Metro stretch cluster; space reclamation Established enterprise-class storage functionality without the need for a dedicated storage array Proven five 9’s high availability Ability to set HA levels per volume, change on-the-fly Continuous data availability – disk, controller, server node, power, network, appliance, rack, room, or site failure Automated failover and failback – SRM, StoreVirtual Remote Copy Storage federation (Peer Motion) Ability to move data across the infrastructure to other SV-based systems Online migration of data across systems, locations, generation changes Scale-out shared storage All available capacity and performance is aggregated and accessible to every volume in the cluster Ability to share storage out to other servers (virtual or physical)

Масштабирование для интенсивной работы с данными HPE Synergy с СХД HPE 3PAR StoreServ Flash

Быстрое, эффективное развертывание СХД FC Flash storage с HPE Synergy Composer

Доступ для любого вычислителя Synergy Масштабирование до 60PB Многофункциональная корпоративная система хранения уровня business critical

Простое подключение к HPE Synergy Через модули Virtual Connect или коммутаторы Brocade SAN

Сетевые фабрики HPE Synergy HPE Synergy поддерживает три резервируемые фабрики

Interconnect Bay 1

Interconnect Bay 2

Interconnect Bay 3

Interconnect Bay 4

Interconnect Bay 5

Interconnect Bay 6

A-Side: 1,2,3 B-Side: 4,5,6

Fabric 3

Fabric 2

Fabric 1

1 2

3

4

5

6

Сеть управления физически отделена от сети данных

Сеть управления

Сеть данных

Сеть управления • Frame Link Module (FLM) • Link port to cascade FLMs • Connects to Composer and Streamer

Сеть данных • Interconnects including Master and

Satellite modules • Uplinks to network • Downlinks to Compute Modules • Leveraged by streamer • Management port for CLI

Высокая производительность плоской масштабируемой фабрики

• Обеспечивает коммутацию • 8 сетевых портов (QSFP) • 4 порта для модулей «Спутник»

• 1-2 порта для подключения к «Мастер» • Простой повторитель сигнала

Кабели для подключения «Спутник» • Кабели 12 x 10Gb

AOC (3, 5, 10, 15м) DAC (1.1, 1.6, 2.1 м)

Модуль «Мастер» (Master)

Модуль «Спутник» (Satellite)



Virtual Connect offers rich set of options for uplinks. The six QSFP+ uplinks ports support following options: QSFP+ Transceivers SR4, LR4, SR4 300M for 1x40Gb and 4x10Gb Ethernet Unified SR4 for 1x40, 4x10 and 4x8 (FC) BiDi over LC is planned that can interoperate with BiDI from Cisco and other vendors AOC cables Various length offering up to 15M both 1x40Gb and splitter (4x10Gb) DAC cables Various length offering up to 3M both 1x40Gb and splitter (4x10Gb) QSFP+toSFP+ adapter Supports sfp+ transceivers and DAC cables Supports sfp+ FC transceiver 10Gbase-T transceiver QSFP+To dual 10Gbase-T transceiver The ICM Cluster ports can be connected via AOC or DAC QSFP+ cables.

Повышение надежности, уменьшение сложности, сокращение затрат

Master

Master

Satellite Satellite

Satellite

Satellite

Нет переподписки между вычислителями, т.е. выше пропускная способность, ниже задержки

Масштабирование до 5 шасси, до 60 вычислителей на фабрику

Дублированные подключения для обеспечения высокой доступности

Up to all 16 x40Gb or all 48 x10Gb

Подключение к внешним сетям

ICM-A ICM-B

Network

LACP

iRF/vPC

ICM cluster links

N+2 Failover scenario

Возможность двум независимым коммутаторам выглядеть одним логическим коммутатором (M-LAG)

Разрешено агрегирование каналов (LACP) между модулями, создание виртуальных портов

Подключения внутри Synergy



ICMP (англ. Internet Control Message Protocol — протокол межсетевых управляющих сообщений) — сетевой протокол, входящий в стек протоколов TCP/IP. LACP (англ. link aggregation control protocol) — открытый стандартный протокол агрегирования каналов, описанный в документах IEEE 802.3ad[⇨] и IEEE 802.1aq. IRF from HPE and Cisco's VPC basically the physical hardware becomes one big logical device. Synergy fabrics will offer that resilience and performance with standards based MLAG. Технология МLAG (Multi-Switch Link Aggregation ) дает возможность двум независимым коммутаторам выглядеть одним логическим коммутатором для других устройств без объединения в стек. top-of-rack (TOR - агрегирующий). Multihoming is the practice of connecting a host or a computer network to more than one network HPE Synergy Composable Fabric’s disaggregated, rack-scale design uses a master/satellite architecture to consolidate data center network connections, reduce hardware and management complexity, and scale network bandwidth across multiple frames. The master module contains intelligent networking capabilities that extend connectivity to satellite frames through the HPE Synergy 10/20Gb Interconnect Link Module, which eliminates the need for a top of rack switch and substantially reduces cost. The reduction in components also simplifies fabric management at scale while consuming fewer ports at the data center aggregation layer. HPE Synergy Composable Fabric modules support up to 3 redundant fabrics per frame, and the QSFP unified ports can be configured for either Ethernet or Fibre Channel. HPE Synergy Composable Fabric employs a flat, east/west architecture to maximize data throughput and minimize latency with only a single hop within large domains of virtual machines or compute modules. In contrast, legacy hierarchical architectures use a north/south design that creates oversubscription bottlenecks and adds latency caused by multiple hops, both of which negatively impact performance. With HPE Synergy Composable Fabric, scaling is fast and simple. When adding a new frame using HPE Synergy’s Fabric Interconnect Link Module, the new frame is becomes an extension of the existing fabric, and the east/west design scales so the performance of the existing workload is not negatively impacted.

Подключение мастер модулей к внешним сетям Коннекторы QSFP+ к сетям 40/10 Gb Ethernet и 8 Gb FC

QSFP+ AOC/DAC cables: – 40 G

QSFP+ Transceivers – MPO/MPO: 40 G – MPO/LC breakout cables: 4x10 G/4x8 Gb

QSFP/SFP+ Adapter – 1x10 G

QSFP/SFP+ AOC/DAC splitter cable – 4x10 G



Virtual Connect offers rich set of options for uplinks. The six QSFP+ uplinks ports support following options: QSFP+ Transceivers SR4, LR4, SR4 300M for 1x40Gb and 4x10Gb Ethernet Unified SR4 for 1x40, 4x10 and 4x8 (FC) BiDi over LC is planned that can interoperate with BiDI from Cisco and other vendors AOC cables Various length offering up to 15M both 1x40Gb and splitter (4x10Gb) DAC cables Various length offering up to 3M both 1x40Gb and splitter (4x10Gb) QSFP+toSFP+ adapter Supports sfp+ transceivers and DAC cables Supports sfp+ FC transceiver 10Gbase-T transceiver QSFP+To dual 10Gbase-T transceiver The ICM Cluster ports can be connected via AOC or DAC QSFP+ cables.

Портфолио сетевых модулей HPE Synergy

Расширение технологии HPE Virtual Connect до масштаба стойки Flex-10/20, L2 и NPIV подключения к существующим LAN и SAN сетям Ethernet, FCoE и традиционный Fibre Channel

Традиционный L2 функционал коммутации и управления Полный контроль через командную строку и мониторинг через HPE Intelligent Management Center (IMC) Традиционный SAN коммутатор Brocade

HPE Synergy 40Gb F8 Switch Module and Interconnect Link

Modules

Brocade 16Gb FC Switch for HPE Synergy

Компонуемая фабрика Традиционная фабрика

HPE Virtual Connect SE 40Gb F8 Module and Interconnect Link

Modules

HPE VC SE 16Gb Module

25



NPIV or N_Port ID Virtualization is a Fibre Channel feature whereby multiple Fibre Channel node port (N_Port) IDs can share a single physical N_Port. This allows multiple Fibre Channel initiators to occupy a single physical port, easing hardware requirements in Storage Area Network (SAN) design, especially where virtual SANs are called for. This allows each virtual server to see its own storage and no other virtual server's storage.[1] NPIV is defined by the Technical Committee T11 in the Fibre Channel - Link Services (FC-LS) specification. The Fabric portfolio consists of HPE Virtual Connect modules (Composable Fabric) and traditional fabric switches. Composable Fabric modules (HPE Virtual Connect modules) The HPE Virtual Connect Modules are interconnects with a wire-once, change-ready templates that allows workloads to be moved without modifying the network. It provides simple, seamless connectivity between virtualized compute modules and data center fabrics. Its highly available, disaggregated architecture provides cost-efficient, easy-to-manage, and resilient fabric for traditional and next-generation workloads. For customers who required native FC, FCoE, and Flat SAN with converged fabric to the compute modules with the Virtual Connect experience, the HPE Virtual Connect SE 40 GB F8 Module is ideal. For customer with higher demands for connectivity at 16Gb and require a separate SAN fabric with the Virtual Connect experience, the HPE Virtual Connect FC 16GB Module is ideal. Traditional Fabric: For customers who want traditional switch functionality at the edge, HPE offers a broad portfolio of switches for Ethernet and Fibre Channel networking, connecting the frame and compute modules. HPE Synergy offers a complete end-to-end solution for customers who require traditional network managed switches with Fibre channel or Fibre Channel over Ethernet support. The HPE Synergy 40 GB F8 Switch is a Layer 2/3 Fiber Channel switch that gives full manual control to network administrators who want to manage their HPE Synergy interconnects from a Command Line Interface (CLI). This fabric solution allows the network administrator to manage the switch within the frame with the same tools, and features, security. The HPE Synergy 40 GB F8 Switch design uses the same master/satellite architecture as HPE Virtual Connect SE 40 Gb F8 module to consolidate data center network connections, reduce hardware complexity, and scale network bandwidth across multiple frames. The master module contains intelligent networking capabilities that extend connectivity to satellite frames, which eliminates the need for a top of rack switch and substantially reduces cost. In addition to the composable fabric, HPE offers other interconnects such as a pass-through module or a SAN switch. For customers who want to maintain their existing network, HPE offers a pass-through module that provides full compute module connectivity to an existing network switch. The HPE Synergy 10Gb/40Gb pass through module allows for a 1 to 1 connection between a compute model and a top of rack switch. It is an alternative to manage the switch outside of the frame with the operating system of your choice. The Brocade 16 GB Fibre Channel SAN Switch Module for HPE Synergy provides high performance, low latency networking with cut-through mode FC SAN capabilities. This switch is ideal for financial services, trading applications, medical imaging, and rendering.

Раздатчик образов HPE Synergy

(Image Streamer)

Потоковый раздатчик образов для бездисковых

вычислителей

Управление HPE Synergy

Управляющий сервер с встроенным

HPE OneView

Компоновщик HPE Synergy (Composer)

Соединение нескольких

шасси по сети 10GBASE-T

Кольцо сети управления

Сеть управления



The Synergy management architecture has 3 major components 2 appliances – 1 core mgmt and one special purpose Glue that brings together a multi-rack, multi-frame CI, the 10Gb mgmt ring and embedded frame link modules

Управляющие компоненты HPE Synergy Компоновщик HPE Synergy (Composer)

Synergy Composer (Компоновщик)

Отсек для установки

Полное управление Компонуемой Инфраструктурой без лицензирования HPE OneView

Выделенная сеть (out-of-band) управления

Высоконадежная кластерная конфигурация компоновщиков (composer) с установкой в любых фреймах

До 21 фрэйма управляются одной парой компоновщиков

Управляющий интегрированный сервер управления с встроенным HPE OneView



The Synergy management architecture has 3 major components 2 appliances – 1 core mgmt and one special purpose Glue that brings together a multi-rack, multi-frame CI, the 10Gb mgmt ring and embedded frame link modules

Много шасси HPE Synergy– одно кольцо управления

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame FLM2 M L

FLM1 M L

Management Network

Frame

FLM1 M L Composer

1

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame FLM2 M L

FLM1 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame FLM2 M L

FLM1 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame FLM2 M L

FLM1 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame FLM2 M L

FLM1 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Frame FLM2 M L

FLM1 M L

Frame

FLM1 M L

FLM2 M L

Frame

FLM1 M L

FLM2 M L

Composer 2



3 Frames are required in the first Logical enclosure in a management ring. 2 Frames are required in other Logical enclosures in the management ring

Федеративное управление с HPE OneView Global Dashboard

– Единое управление Synergy, c-Class, DL серверами – Масштабирование до 10 платформ Synergy/OneView

(210 фреймов или 400 полок c7000)

– Быстрый поиск/идентификация устройств

– Контекстный запуск консолей OneView / iLO

– Сбор и хранение инвентарных данных

Composer 1 Composer 2 Composer 3 OneView 1 OneView 2



Federate Appliances with HPE OneView+ This is a new capability that will be delivered in roughly the same timeframe as middle column (Composable Infrastructure) on the HPE OneView roadmap HPOV + will provide the ability to federate multiple HPE OneView appliances into a single unified view. This approach provides a Initial phase will provide a unified view of 10 appliances (virtual and physical) along with dashboard for at a glance health status with single sign-on for V2.0 appliance users A pair of composers for HA – counts as a single appliance in the 10 count Plan is to scale above 10 appliances in the future, this will be what is tested in the initial release Summary Updates device status via API’s and state change message bus or polling Searchable to enable end users to quickly locate device(s) of interest. Supports in context from dashboard into specific HPE OneView managed device Based on Hewlett Packard Enterprise Grommet interface Plan is to also include a FW report with CSV export – so the customer can export and sort the report as needed Follow-on phase will include the ability to synchronize templates, networks, enclosure groups across multiple appliance – see next slide

Альтернатива PXE и загрузке по SAN

Бесшовная интеграция с компоновщиком

Встроенное устройство с библиотекой образов

Управляющие компоненты HPE Synergy

HPE Synergy Image Streamer (Потоковый раздатчик образов)

Высоконадежная кластерная конфигурация

Одна пара раздатчиков для 5 шасси

Автоматическая синхронизация образов

Раздатчик образов HPE Synergy (Image Streamer)



Preboot Execution Environment, произносится пикси) — среда для загрузки компьютеров с помощью сетевой карты без использования жёстких дисков, компакт-дисков и других устройств, применяемых при загрузке операционной системы. Для организации загрузки системы в PXE используются протоколы IP, UDP, DHCP и TFTP. Fast, reliable deployment and updates Efficient operations powered by server profiles Easily migrate workloads rapidly with streamed images Lower costs by reducing local compute and SAN storage In the hyper-connected Idea Economy, fluid alignment between infrastructure and workloads, and the fast deployment of end-to-end resources, is paramount. HPE Synergy provisions compute modules with bootable images created from your golden images. This capability lets HPE Synergy deploy and update infrastructure with unmatched speed and agility. Within the HPE Synergy Frame, the HPE Synergy Image Streamer maintains a physical appliance repository for your golden images. These golden images can be rapidly cloned to create unique bootable images for compute nodes. It enables HPE Synergy to quickly deploy a new compute module or update an existing one. This is far faster than the traditional, sequential process of building servers — physical provisioning followed by operating system (OS), hypervisor installation, I/O drivers, application stacks, et cetera. Administrators using HPE Image Streamer can design bootable images for compute nodes, with the OS and application stacks included, for ready-to-run environments. HPE Synergy Image Streamer ensures high availability by providing redundant repositories of bootable images which are used in a secure manner. The separation of management and data networks provides additional security and dedicated bandwidth. In addition, security vulnerabilities due to PXE boot are eliminated, and no external network connectivity or functions are required for operation. Run anything: Optimize all apps and service levels Automate and control: Automate deployments and updates of compute modules using software-defined intelligence and easy-to-use stateless operation; programmatically control processes via the Unified API. HPE Image Streamer uses similar processes for initial deployments and image updates. This is a new paradigm for how to manage images at the compute modules. This paradigm improves administrator control over multiple nodes and multiple virtual machines while improving operating efficiencies and dramatically increasing agility. Deployments are performed from your golden images, with operating environments and personalities, at extreme speed. The image payload consists of your compute module’s profile (from HPE Composer), and your golden image (or operating environment with a bootable OS and application, and the I/O driver version) and your personality (or operating system and application configuration, which includes the hostname, IP config, etc.) Updates can be performed by capturing revised images and re-deploying them at extreme speed. HPE Image Streamer provides a simple user experience to accomplish compute boot/run provisioning and operating system deployment. Unified API access to HPE Image Streamer enables programmatic control of its functions. Provisioning can be controlled from the GUI or it can be seamlessly integrated into existing scripting processes. This allows partners, developers, and users to integrate, automate, and customize HPE Image Streamer use for scaling across large infrastructure blocks. Move faster: Accelerate app and service delivery Streamline deployment and update processes across many nodes at extreme speed: Compute module profiles drive out-of-box deployments and updates quickly and consistently utilizing true stateless operation. Deploying a typical OS or hypervisor can be time consuming because it requires customizing and/or copying each image for each compute module. HPE Synergy Image Streamer accelerates these processes by tight integration with HPE Synergy Composer. Together, they consistently deploy bootable images across stateless compute modules using a simple, consistent process. This deep integration reduces complexity and provides consistency across infrastructure. HPE Image Streamer enables true ‘stateless’ operation by integrating your profiles with your golden images and your personalities for rapid implementation onto available hardware, which might be at a later point in time. True ‘stateless’ operation allows IP addresses to be assigned to software like operating systems (similar to the way we have assigned IP addresses to hardware). This allows environment planning apart from hardware and fast implementation when hardware is available. With HPE Synergy Image Streamer, images can be created and then instantly cloned, personalized with a host name and IP addresses, and then booted as if they were from a SAN environment. Unlike booting from SAN environments, no additional setup or configuration is needed. HPE Synergy Image Streamer can be set up much faster and more efficiently than a boot-from-SAN environment, and it can deploy new images quicker and lower cost than traditional on-server boot disks. Work efficiently: Reduce operational effort and cost Simplify system updates and enforce compliance for infrastructure stability: Manage deployment plans and construct bootable images from capturing, cloning or customizing golden images; enforce compliance using templates; provision, update, or roll-back image quickly to minimize maintenance windows. HPE Image Streamer supports a variety of operations for your flexibility in handling golden images. You can capture golden images for your use, import images from another location, or to edit some of your ‘known good’ images for re-use. Easy manipulation of images helps you to construct golden images for your use. HPE Image Streamer automatically sets up configurations and integrations with HPE Composer for profile and template use. By leveraging profiles and templates from HPE Composer, it can also enforce image compliance by using the ‘monitor, flag, and remediate’ capabilities of templates. Profiles created from the template can be monitored for configuration compliance and out-of-compliance images are flagged. Users then can control the remediation process to bring individual modules or multiple nodes back into compliance to assure consistent operations. HPE Synergy Image Streamer uses profiles to capture all workload configuration information. By keeping customer-defined workload requirements with the profile, the image of any physical compute module can be replaced or updated to achieve the desired state from the image repository. This stateless operation, which separates the image profile from the hardware, makes updates easy and enables quick hardware replacements to minimize downtime. To upgrade an image, IT administrators just revise the profile or template to associate with a new boot image. When the compute module is rebooted, the profile automatically updates to connect the module to the new image. Rollbacks are similarly accomplished by simply changing the profile back to the original boot image, followed by a reboot. This update process avoids complex and time-consuming patching processes. HPE Synergy Image Streamer automates the seamless configuration, provisioning, and updating of composable infrastructure. This unlocks value delivery for faster delivery of on-demand resources. Logical server provisioning with Operating Environment (OE) deployment Enables ‘stateless’ servers – adds OE deployment to HP OneView server profiles Deploy and personalize OE boot volumes from Golden Images Staged provisioning of complete ‘logical servers’ onto ‘physical servers’ Performance – OE deployment at unmatched speed Instant OE deployment for initial out-of-the-box experience Minimize maintenance windows with fast OE updates and changes Simplicity-Consistency - Integrated with HPE Synergy Composer OE deployment driven by HP OneView server profile operations i3S managed via the HP OneView GUI and REST API HP OneView automates i3S discovery and set-up Achieves OE consistency across stateless compute nodes driven by Golden Images Easily manage bundles of Golden Images and their deployment plans Reliability-Security Highly-available OE boot volumes with redundant compute node connectivity Does not use PXE or other remote deployment technology Rapid recovery from compute node outages using saved Server profiles Foundational technology for future OE control functions

Управление бездисковыми вычислителями с помощью профилей

Параметры аппаратной конфигурации и развертываемое ПО хранятся и управляются

отдельно от физических серверов

Профиль сервера

Нет необходимости сохранять состояние

вычислителей

Программные параметры

Образ для загрузки

Конфигурация ОС и приложений

Бездисковый вычислитель

Компоновщик HPE Synergy

Раздатчик образов HPE Synergy

Физические параметры

Сетевые настройки Подключение к СХД внутренним и внешним Версии микрокодов Настройки BIOS



This is a key slide for understanding the operation of HPE Image Streamer. Starting from the left-hand side, you see both Synergy management appliances: HPE Composer is shown in the upper left-hand corner and is required for HPE Synergy operation. It provides the ‘management brains’ of Synergy systems. HPE Image Streamer is shown in the lower left-hand corner and it complements HPE Composer operation. The center green box shows the Server Profile, which provides ‘software defined intelligence’. HPE Synergy Composer defines the ‘Physical (HW) State’ of the compute node in the Server Profile. This should look familiar, because this is part of HPE OneView. (Remember that HPE Composer has HPE OneView embedded in it.) HPE Synergy Image Streamer complements HPE Composer by capturing the ‘Deployed SW State’ into the Server Profile. This enhanced server profile is stored on the Image Streamer appliance. Notice what has happened: We have captured both the HW State and the Deployed SW State into software. This is ‘Infrastructure as Code’. (At this point, we don’t even need physical compute modules to be present.) Furthermore, note that the compute modules can be ‘stateless’ – simplified to run without a local disk drive or RAID controller. (This can save money on the initial hardware purchase.) With the HPE Synergy Image Streamer, Compute modules can remain stateless, only configured for the essentials such as network config, current firmware, etc. and remains available as a resource for any template which may be deployed through the Image Streamer. Any compute module at any time may be used for any of the templates deployed for workloads as long as it is available. This allows compute modules to be a part of the fluid pool of resources, ready to be run for any workload, and remaining stateless when not in use. HPE Image Streamer enables true ‘stateless’ operation by integrating your profiles with your golden images and your personalities for rapid implementation onto available hardware, which might be at a later point in time. True ‘stateless’ operation allows IP addresses to be assigned to software like operating systems (similar to the way we have assigned IP addresses to hardware). This allows environment planning apart from hardware and fast implementation when hardware is available. Now let us describe three operational scenarios – how Image Streamer allows us to ‘manage physical servers like virtual machines (VMs)’ It is important that the creation of OS Boot Volumes in Image Streamer is fast – it only takes seconds to complete. Initial Deployments from bare metal: When a compute node is deployed it can boot from bare metal directly into a running OS – off the OS boot image which is stored on Image Streamer. We have just completed a deployment from bare metal. Updates are likewise fast… What is an update? …Let’s say you have identified a security vulnerability… You need to make an update to your image to fix a security hole. What do you do? You capture an image in use off a currently running node and edit it to create a new Golden Image. (You can use your tools or ours, we don’t care.) Image Streamer creates new OS Boot Volumes – it only takes seconds – and you re-boot the compute nodes off the new OS Boot volumes. You have just completed an update. Rollbacks are likewise simple… How do you perform a rollback? …You boot the server off an older (previously used) set of OS Boot Volumes. Another important point: We have been describing Image Streamer as handling software in the form of an OS… But in reality, Image Streamer can handle any software that the user captures into his Golden Image. To the extent that the user captures an application stack into his Golden Image, then that application can likewise be handled by Image Streamer! For example, in the lab, we have implemented a database application in this manner. This is a new paradigm in how to operate. It enables HPE Synergy to ‘manage physical servers like virtual machines (VMs)’.

Позиционирование HPE Synergy

1. Synergy не сложнее блейдовых систем. Внутри те же ProLiant-ы с привычным контроллером управления iLO. Только в Synergy серверы мощнее по процессорам и, особенно, по объему памяти.

2. Synergy содержит встроенные средства программного управления HPE OneView – такие же как и для BladeSystem c-Class. Общая консоль управления OneView Global Dashboard позволяет управлять с одной точки и Synergy, и c-Class.

3. Synergy очень удобно подключается к сетям ЦОД. Единая сетевая фабрика может влючать до 5 шасси Synergy, что значительно удешевляет сетевые подключения.

32

Позиционирование HPE Synergy

4. Synergy может подключаться как к внешним дисковым системам, так и к внутренним дисковым полкам (до 100 дисков на один сервер). Общая СХД при этом обеспечивается StoreVirtual или другими программными средствами.

5. Synergy готова к новым технологиям на 12-15 лет вперед (100Gb Ethernet и выше); предусмотрена уже сейчас возможность подключения оптических карт и коммутаторов.

6. Особенно ярко проявятся преимущества Synergy при выходе следующего поколения серверов. Будут поддерживаться 200 Вт процессоры и максимально возможный объем памяти.

33

HPE Synergy - ресурсы

Основной раздел на HPE.com

Документация

HPE Synergy For Dummies

На русском языке

– Synergy за 2 минуты (видео на английском языке с русскими субтитрами)

– Брошюра Компонуемая инфраструктура HPE

– Брошюра Пять шагов к построению компонуемой инфраструктуры на базе HPE Synergy

– Брошюра Преимущества компонуемой инфраструктуры - привлекательная локальная альтернатива публичному облаку

34

http://www.hpe.com/info/synergy

http://h17007.www1.hpe.com/us/en/enterprise/integrated-systems/info-library/

https://www.hpe.com/us/en/resources/integrated-systems/synergy-for-dummies.html?parentPage=/us/en/products/integrated-systems/synergy

http://h22228.www2.hpe.com/video-gallery/us/en/products/00096A67-84F9-4464-9F6E-B3FEDF5EC0D5/r/video/

http://h22228.www2.hpe.com/video-gallery/us/en/products/00096A67-84F9-4464-9F6E-B3FEDF5EC0D5/r/video/

https://www.hpe.com/h20195/V2/GetDocument.aspx?docname=4AA5-8813RUE

https://www.hpe.com/h20195/v2/GetDocument.aspx?docname=4AA6-3322RUE



Спасибо!

35

Title slide with picture · 2017. 10. 30. · Новый класс инфраструктуры...

Documents

Transcript of Title slide with picture · 2017. 10. 30. · Новый класс инфраструктуры...