Работа с базами данных
<<  BIG DATA Новый вызов Урок по тысяча 2 класс петерсон  >>
Data Center
Data Center
Картинки из презентации «Resource Management in Virtualization-based Data Centers» к уроку информатики на тему «Работа с базами данных»

Автор: Bhuvan Urgaonkar. Чтобы познакомиться с картинкой полного размера, нажмите на её эскиз. Чтобы можно было использовать все картинки для урока информатики, скачайте бесплатно презентацию «Resource Management in Virtualization-based Data Centers.ppt» со всеми картинками в zip-архиве размером 637 КБ.

Resource Management in Virtualization-based Data Centers

содержание презентации «Resource Management in Virtualization-based Data Centers.ppt»
Сл Текст Сл Текст
1Resource Management in 18data, it may be likely to do that again
Virtualization-based Data Centers. Bhuvan E.g., queries may be issued in bursts
Urgaonkar Computer Systems Laboratory Trade-off between domain context switch
Pennsylvania State University. and how much extra time you let a sender
2Data Center. Cluster of compute and DomU continue.
storage servers connected by high-speed 19Multi-processor Scheduling. Idea: Dom0
network Rent out resources in return for should be scheduled together with a DomU
revenue Internet applications, Scientific doing I/O Utilize the multiple CPUs to
applications, … Revenue scheme expressed “co-schedule” a communicating DomU with
using SLAs. Dom0 Ensure domains that communicate a lot
3Resource Management in Data Centers. do not starve others Relaxed fairness: 50%
Goal: Meet application SLAs Easy solution: CPU over intervals > 1 second Approach:
Over-provision resources Over-provisioning Decay the CPU priority of communicating
can be very wasteful Energy, management, DomUs to ensure relaxed fairness is not
failures, … Data center would like to violated.
maximize revenue! Dynamic capacity 20Outline. Introduction and Motivation
provisioning: match resource allocations Resource Management in a Xen-based Data
to varying workloads Challenges: Center Resource Accounting Resource
Determining changing resource needs of Allocation and Scheduling Performance
applications Effective sharing of Optimizations for Xen Other Research
resources among applications E.g., server Concluding Remarks.
consolidation can reduce cost Automating 21Performance Optimizations for Xen.
resource management. Switching between native & virtual
4Resource Management in Data Centers. hosting Dynamic merging and splitting of
Goal: Meet application SLAs Easy solution: domains Overbooking of memory Improved
Over-provision resources Over-provisioning migration techniques Coalesce network
can be very wasteful Energy, management, packets directed to the same physical
failures, … Data center would like to server.
maximize revenue! Dynamic capacity 22Performance Optimizations for Xen.
provisioning: match resource allocations Switching between native & virtual
to varying workloads Challenges: hosting Dynamic merging and splitting of
Determining changing resource needs of domains Overbooking of memory Improved
applications Effective sharing of migration techniques Coalesce network
resources among applications E.g., server packets directed to the same physical
consolidation can reduce cost Automating server.
resource management. 23Optimizing Network Communication.
5Motivation for Virtualized Hosting in Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
Data Centers. Key idea: Design data center database. Apache. Quake 1. Mysql. Quake 2.
using virtualization Virtual machine Windows’. Linux’. Windows’. Linux’. Xen
monitor (VMM) and virtual machine (VM) A hypervisor. Xen hypervisor. Hardware.
software layer that runs on a server and Hardware.
allows multiple OS/applications to 24Optimizing Network Communication.
co-exist Each OS/application is given the Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
illusion of its own “virtual” machine that database. Apache. Quake 1. Mysql. Quake 2.
it has to itself Why is this good? Windows’. Linux’. Windows’. Linux’. Xen
Consolidation of diverse OS/apps possible hypervisor. Xen hypervisor. Hardware.
Migration made easier Small code of VMM Hardware.
=> improved security Not a new idea, 25Optimizing Network Communication.
but existing solutions are inadequate Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
Goal: Devise efficient resource management database. Apache. Quake 1. Mysql. Quake 2.
solutions for a virtualization-based data Windows’. Linux’. Windows’. Linux’. Xen
center. hypervisor. Xen hypervisor. Hardware.
6The Xen Virtual Machine Monitor. VMM = Hardware.
hypervisor VM = domain Para-virtualization 26Optimizing Network Communication.
Special domain called Dom0. Dom0. Dom1. Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
Dom2. Apache Web server. Mysql database. database. Apache. Quake 1. Mysql. Quake 2.
Windows’. Linux’. Xen hypervisor. Windows’. Linux’. Windows’. Linux’. Xen
Hardware. hypervisor. Xen hypervisor. Hardware.
7Outline. Introduction and Motivation Hardware.
Resource Management in a Xen-based Data 27Optimizing Network Communication.
Center Resource Accounting Resource Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
Allocation and Scheduling Performance database. Apache. Quake 1. Mysql. Quake 2.
Optimizations for Xen Other Research Windows’. Linux’. Windows’. Linux’. Xen
Concluding Remarks. hypervisor. Xen hypervisor. Hardware.
8Xen-based Data Center. Each Hardware.
application component runs within a Xen 28Optimizing Network Communication.
domain. Online book-store. Online game Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
server. Dom0. Dom0. Dom1. Dom2. Dom1. database. Apache. Quake 1. Mysql. Quake 2.
Dom2. Mysql database. Apache. Quake 1. Windows’. Linux’. Windows’. Linux’. Xen
Mysql. Quake 2. Windows’. Linux’. hypervisor. Xen hypervisor. Hardware.
Windows’. Linux’. Xen hypervisor. Xen Hardware.
hypervisor. Hardware. Hardware. Physical 29Optimizing Network Communication.
machine # 1. Physical machine # 2. Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
9Resource Usage Accounting. Need for database. Apache. Quake 1. Mysql. Quake 2.
accurate resource accounting Estimate Windows’. Linux’. Windows’. Linux’. Xen
future needs Relate performance and hypervisor. Xen hypervisor. Hardware.
resource consumption Charge applications Hardware.
for resource usage Accounting in Xen-based 30Optimizing Network Communication.
hosting Statistics for each DomU can be Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
gathered by hypervisor E.g., number of database. Apache. Quake 1. Mysql. Quake 2.
bytes sent by a DomU Hidden activity: CPU Windows’. Linux’. Windows’. Linux’. Xen
activity performed by Dom0 Similar to hypervisor. Xen hypervisor. Hardware.
activity done by a kernel for a process Hardware.
Techniques to de-multiplex Dom0’s activity 31Optimizing Network Communication.
across DomUs How much work does Dom0 have Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
to do for each DomU? database. Apache. Quake 1. Mysql. Quake 2.
10Resource Allocation. Multi-time scale Windows’. Linux’. Windows’. Linux’. Xen
resource allocation Server assignment: hypervisor. Xen hypervisor. Hardware.
course time-scale Scheduling: fine Hardware.
time-scale Placement Like a knapsack 32Optimizing Network Communication.
problem What time-scale? Migration versus Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
replication. database. Apache. Quake 1. Mysql. Quake 2.
11Intelligent Scheduling of Distributed Windows’. Linux’. Windows’. Linux’. Xen
Applications. Motivation: Co-scheduling of hypervisor. Xen hypervisor. Hardware.
parallel applications Schedule distributed Hardware.
communicating components together. 33Optimizing Network Communication.
Physical machine # 1. Physical machine # Dom0. Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
2. database. Apache. Quake 1. Mysql. Quake 2.
12Intelligent Scheduling of Distributed Windows’. Linux’. Windows’. Linux’. Xen
Applications. Motivation: Co-scheduling of hypervisor. Xen hypervisor. Hardware.
parallel applications Schedule distributed Hardware.
communicating components together. 34Optimizing Network Communication. (-)
Physical machine # 1. Physical machine # Increased CPU processing for coalescing
2. and splitting packets (+) Reduced
13Intelligent Scheduling of Distributed interrupt processing at receiver. Dom0.
Applications. Motivation: Co-scheduling of Dom0. Dom1. Dom2. Dom1. Dom2. Mysql
parallel applications Schedule distributed database. Apache. Quake 1. Mysql. Quake 2.
communicating components together. Windows’. Linux’. Windows’. Linux’. Xen
Physical machine # 1. Physical machine # hypervisor. Xen hypervisor. Hardware.
2. Hardware.
14Intelligent Scheduling of Distributed 35Optimizing Network Communication. What
Applications. Motivation: Co-scheduling of kinds of packets can be coalesced? TCP
parallel applications Schedule distributed ACKs? Other packets? Would it make sense
communicating components together. to do anticipatory packet scheduling at
Physical machine # 1. Physical machine # the sender? Dom0. Dom0. Dom1. Dom2. Dom1.
2. Message waits till yellow app gets the Dom2. Mysql database. Apache. Quake 1.
CPU. Mysql. Quake 2. Windows’. Linux’.
15Intelligent Scheduling of Distributed Windows’. Linux’. Xen hypervisor. Xen
Applications. Motivation: Co-scheduling of hypervisor. Hardware. Hardware.
parallel applications Schedule distributed 36Outline. Introduction and Motivation
communicating components together. Resource Management in a Xen-based Data
Physical machine # 1. Physical machine # Center Resource Accounting Resource
2. Message can be received Immediately if Allocation and Scheduling Performance
the yellow app gets the CPU. Optimizations for Xen Other Research
16Intelligent Scheduling of Distributed Concluding Remarks.
Applications. Motivation: Co-scheduling of 37Provisioning a Directional
parallel applications Schedule distributed Antenna-based Network. Directional
communicating components together. antennas Longer reach Less interference
Physical machine # 1. Physical machine # => Increased capacity.
2. 38Provisioning a Directional
17Co-ordinated Scheduling of Antenna-based Network. Theoretical results
Communicating Domains. Idea #1: User-centric version Fair bandwidth
Preferentially schedule a DomU when it allocation Optimal algorithm based on
receives data Modify Xen CPU scheduler to dynamic programming Provider-centric
give higher preference to receiving DomU version Maximize revenue NP-hard,
Important: Also need to ensure that Dom0 2-approximation algorithm Ongoing work
gets to run to take care of I/O Scheduler Heuristics to incorporate mobility
should partition the CPU allocation for a Evaluation through simulation
DomU into those for Dom0 and DomU Implementation … may be.
appropriately. 39Concluding Remarks. Resource mgmt. in
18Co-ordinated Scheduling of virtualized environments Provisioning
Communicating Domains. Idea #2: Try to wireless networks Energy optimization in
schedule a sender DomU when it is expected sensor networks Distributed systems,
to receive the response An application Operating systems Combination of analysis,
knows best, but mods undesirable Let the algorithm design and experimentation with
hypervisor learn from past behavior E.g., prototypes Acknowledgements: Faculty:
query responses might be returning in 1-2 Anand, Piotr, Wang-Chien Students:
seconds Idea #3: Anticipatory CPU Amitayu, Arjun, Ross, Shiva, Sriram.
scheduling If a domain has sent/received
Resource Management in Virtualization-based Data Centers.ppt
http://900igr.net/kartinka/informatika/resource-management-in-virtualization-based-data-centers-229012.html
cсылка на страницу

Resource Management in Virtualization-based Data Centers

другие презентации на тему «Resource Management in Virtualization-based Data Centers»

«Data Mining» - Data Mining не может заменить аналитика! Введение в Data Mining. Методы Data Mining. Статистические методы. Алгоритмы. Целью поиска является не гарантированно верное решение, а лучшее из возможных. Процесс конструирования. Технологические методы. Стадии Data Mining. Проблемы и вопросы. Анализ связей (корреляционный и регрессионный анализ, факторный анализ, дисперсионный анализ).

«Триггеры баз данных» - Создание замещающих триггеров. Обновим группу. Типы триггеров. Триггер замещения. Модифицируемые и немодифицируемые представления. Привилегии для создания триггера. Замещающий триггер. Представления, которые содержат соединения. Имя отдела. Создание замещающих триггеров баз данных. Пример замещающего триггера.

«Практические работы по базам данных» - Сформировать условие отбора. Создание отчета в базе данных. Практическая работа №3 Создание связей между таблицами. Информационные системы и базы данных. Запрос-выборка. Открыть закладку «запросы»; выполнить команду Создать, выбрать «Конструктор». Практические работы. Типы данных. Построение модели данных. 5. Создание базы данных в среде MS Access.

«Работа с базами данных» - Правила восстановления. Дальнешее обсуждение. Правила восстановления: протокол-возврата. Когда протокол (или его часть) становится ненужным. База данных должна отражать реальный мир. Что делать при восстановлении. «Не останавливающие» контрольные точки. Восстановление при сбоях. Согласованность данных.

«Хранимые процедуры» - Системные хранимые процедуры. Реализация триггеров. Хранимые процедуры. Примеры использования. Типы триггеров. Понятие хранимых процедур. Сервер. Триггеры. Создание, изменение и удаление хранимых процедур. Триггер.

«Проектирование баз данных» - Работа с сохраненной базой данных. Задание структуры базы данных. Нормализация. Плохо нормализованная таблица. Таблица может быть: Хорошо нормализованной Плохо нормализованной. Этапы создания базы данных. Создание структуры базы данных и заполнение. Организация информации в табличную форму. Проектирование баз данных.

Работа с базами данных

11 презентаций о работе с базами данных
Урок

Информатика

130 тем
Картинки