SlideShare ist ein Scribd-Unternehmen logo
1 von 60
Downloaden Sie, um offline zu lesen
{
“名前” : “真壁 徹(まかべ とおる)”,
“所属” : “日本マイクロソフト株式会社”,
“役割” : “クラウド ソリューションアーキテクト”,
“経歴” : “大和総研  HP Enterprise”,
“特技” : “クラウド & オープンソース”
}
http://perspectives.mvdirona.com/2017/04/how-many-data-centers-needed-
world-wide/
http://fortune.com/2017/04/12/mark-hurd-oracle-data-centers/
"If I have two-times faster computers, I don't need as many data centers. If I can speed up the
database, maybe I need one fourth as may data centers. I can go on and on about how tech
drives this.“
[Microsoft Global Datacenters and
Network Infrastructure]
https://www.youtube.com/watch?v=bqZrejosqWU
38announced regions
34available regions
リージョンペア
リージョン リージョン
リージョン
AWS re:Invent 2014 “AWS Innovation at Scale with
James Hamilton” より
12.88ミリ秒18.92ミリ秒
9.00ミリ秒
測ってみました
"There is some absolute data center size where the facility becomes “too big to fail.” This line is gray and open
to debate but the limiting factor is how big of a facility can an operator lose before the lost resources and the
massive network access pattern changes on failure can’t be hidden from customers.”
1.Microsoft: 35 MW with CloudHQ in Manassas, Virginia
2.Microsoft: 30 MW with EdgeConneX in Elk Grove Village, Illinois
3.Microsoft: 22 MW with CyrusOne in Ashburn, Virginia
4.Microsoft: 16 MW with DuPont Fabros Technology in Santa Clara, California
5.Microsoft: 13.5 MW with CyrusOne in Phoenix, Arizona
6.Microsoft: 9 MW with CyrusOne in San Antonio, Texas
7.Oracle: 6 MW with CyrusOne in Sterling, Virginia
8.Salesforce: 6 MW with QTS in Dallas, Texas
9.Oracle: 5 MW with CyrusOne in Ashburn, Virginia
10.Oracle: 5 MW with Digital Realty Trust in Ashburn, Virginia
http://www.datacenterknowledge.com/archives/2017/01/18/who-leased-the-most-data-center-space-in-2016/
Microsoft、AWSともに自社建設、保有分は入っていないので雰囲気だけ
Colocation Density
2.0+ Power Usage
Effectiveness (PUE) 1.4 – 1.6 PUE
Discrete servers
Capacity
20 year technology
Rack
Density & deployment
Minimized resource
impact
Generation 1 Generation 2
Containment Modular Hyper-scale
1.2 – 1.5 PUE
1.12 – 1.20 PUE 1.07 – 1.19 PUE
Containers, PODs
Scalability &
sustainability
Air & water
Economization
Differentiated SLAs
Deployment Areas &
ITPACs
No more traditional IT
Right-sized
Faster time-to-market
Outside air cooled
Fully integrated
Resilient software
Common
infrastructure
Operational simplicity
Flexible & scalable
Generation 3 Generation 4 Generation 5
http://natick.research.microsoft.com/
電気代
Standard_D1_v2 10.404 East US 2 / -44.1%
Standard_D2_v2 20.91 East US 2 / -44.4%
Standard_D3_v2 41.718 East US 2 / -44%
Standard_D4_v2 83.436 East US 2 / -44%
Standard_D5_v2 166.872 East US 2 / -44%
Standard_D11_v2 23.46 East US 2 / -35.2%
Standard_D12_v2 46.818 East US 2 / -34.9%
Standard_D13_v2 93.636 East US 2 / -34.9%
Standard_D14_v2 187.17 East US 2 / -34.8%
Standard_D15_v2 233.988 East US 2 / -34.7%
Standard_D1_v2 6.936 East US 2 / -16.2%
Standard_D2_v2 13.872 East US 2 / -16.2%
Standard_D3_v2 27.744 East US 2 / -15.8%
Standard_D4_v2 55.488 East US 2 / -15.8%
Standard_D5_v2 110.874 East US 2 / -15.7%
Standard_D11_v2 19.38 East US 2 / -21.6%
Standard_D12_v2 38.658 East US 2 / -21.1%
Standard_D13_v2 77.418 East US 2 / -21.2%
Standard_D14_v2 154.836 East US 2 / -21.1%
Standard_D15_v2 193.494 East US 2 / -21.1%
Standard_D1_v2 8.976 East US 2 / -35.2%
Standard_D2_v2 17.952 East US 2 / -35.2%
Standard_D3_v2 35.802 East US 2 / -34.8%
Standard_D4_v2 71.604 East US 2 / -34.8%
Standard_D5_v2 143.31 East US 2 / -34.8%
Standard_D11_v2 23.868 East US 2 / -36.3%
Standard_D12_v2 47.94 East US 2 / -36.4%
Standard_D13_v2 95.574 East US 2 / -36.2%
Standard_D14_v2 191.148 East US 2 / -36.1%
Standard_D15_v2 238.986 East US 2 / -36.1%
Standard_D1_v2 10.914 East US 2 / -46.7%
Standard_D2_v2 21.828 East US 2 / -46.7%
Standard_D3_v2 43.656 East US 2 / -46.5%
Standard_D4_v2 87.414 East US 2 / -46.6%
Standard_D5_v2 174.828 East US 2 / -46.6%
Standard_D11_v2 23.46 East US 2 / -35.2%
Standard_D12_v2 46.818 East US 2 / -34.9%
Standard_D13_v2 93.636 East US 2 / -34.9%
Standard_D14_v2 187.17 East US 2 / -34.8%
Standard_D15_v2 233.988 East US 2 / -34.7%
Azure 仮想マシン(Linux)の価格(円)を、最安値リージョンと比較すると
東日本 西ヨーロッパ
西イギリス 東アジア(香港)
m4.xlarge $0.258 -29%
m4.2xlarge $0.516 -29%
m4.4xlarge $1.032 -29%
"Eventually it just make better sense to offer customers alternative regions rather than attempting to scale a
single region and the argument in favor of multiple regions become even stronger when other factors are
considered.”
Azure Regionオンプレミス(ユーザー占有型)
User User
User User
Storage
Cluster
Storage
Cluster
Server
Cluster
Server
Cluster
共用部分が大きい
Azure Regionオンプレミス(ユーザー占有型)
User User
User User
Storage
Cluster
Storage
Cluster
Server
Cluster
Server
Cluster
影響範囲
ストレージクラスターが故障した場合
影響範囲が大きい
Azure Regionオンプレミス(ユーザー占有型)
User User
User User
Storage
Cluster
Storage
Cluster
Server
Cluster
Server
Cluster
データセンター全体で共有している設備の障害影響は、
オンプレもAzureも同様
どれも冗長化、自己回復機能を有するが、不具合や作業ミスの可能性はゼロではない
ユーザーがそれに対処するにはマルチリージョン構成が有効
(規模の違いはありますが)
Azure
Storage
Cluster
Storage
Cluster
Server
Cluster
Server
Cluster
Traditional on-premises Modern cloud
Monolithic, centralized
Design for predictable scalability
Relational database
Strong consistency
Serial and synchronized processing
Design to avoid failures (MTBF)
Occasional big updates
Manual management
Snowflake servers
Decomposed, de-centralized
Design for elastic scale
Polyglot persistence (mix of storage technologies)
Eventual consistency
Parallel and asynchronous processing
Design for failure (MTTR)
Frequent small updates
Automated self-management
Immutable infrastructure
Azure Application Architecture Guide
https://docs.microsoft.com/ja-jp/azure/architecture/guide/
"But, without a more fundamental solution to the speed of light problem, many regions are the only practical
way to effectively serve the entire planet for many workloads.”
車線数(帯域)は
増えている
すでに僕のかんがえる
最速の車(光)
"Imagine a $20,000 car in one market costing far more than $200,000 in another market.”
海底ケーブルへの投資
https://blogs.technet.microsoft.com/hybridcloud/2016/05/26/microsoft-
and-facebook-to-build-subsea-cable-across-atlantic/
https://azure.microsoft.com/ja-jp/blog/microsoft-invests-in-subsea-
cables-to-connect-datacenters-globally/
160Tbps
80Tbps
$azure location list --details
info: Executing command location list
+ Getting ARM registered providers
info: Getting locations...
data:
data: Location : japaneast
data: DisplayName : East Japan
data:
data:
[…]
12.88ミリ秒18.92ミリ秒
9.00ミリ秒
測ってみました
" In addition, some national jurisdictions will put in place legal restrictions than make it difficult to fully serve
the market without a local region. ”
Standard_D1_v2 10.404 East US 2 / -44.1%
Standard_D2_v2 20.91 East US 2 / -44.4%
Standard_D3_v2 41.718 East US 2 / -44%
Standard_D4_v2 83.436 East US 2 / -44%
Standard_D5_v2 166.872 East US 2 / -44%
Standard_D11_v2 23.46 East US 2 / -35.2%
Standard_D12_v2 46.818 East US 2 / -34.9%
Standard_D13_v2 93.636 East US 2 / -34.9%
Standard_D14_v2 187.17 East US 2 / -34.8%
Standard_D15_v2 233.988 East US 2 / -34.7%
Standard_D1_v2 6.936 East US 2 / -16.2%
Standard_D2_v2 13.872 East US 2 / -16.2%
Standard_D3_v2 27.744 East US 2 / -15.8%
Standard_D4_v2 55.488 East US 2 / -15.8%
Standard_D5_v2 110.874 East US 2 / -15.7%
Standard_D11_v2 19.38 East US 2 / -21.6%
Standard_D12_v2 38.658 East US 2 / -21.1%
Standard_D13_v2 77.418 East US 2 / -21.2%
Standard_D14_v2 154.836 East US 2 / -21.1%
Standard_D15_v2 193.494 East US 2 / -21.1%
Standard_D1_v2 8.976 East US 2 / -35.2%
Standard_D2_v2 17.952 East US 2 / -35.2%
Standard_D3_v2 35.802 East US 2 / -34.8%
Standard_D4_v2 71.604 East US 2 / -34.8%
Standard_D5_v2 143.31 East US 2 / -34.8%
Standard_D11_v2 23.868 East US 2 / -36.3%
Standard_D12_v2 47.94 East US 2 / -36.4%
Standard_D13_v2 95.574 East US 2 / -36.2%
Standard_D14_v2 191.148 East US 2 / -36.1%
Standard_D15_v2 238.986 East US 2 / -36.1%
Standard_D1_v2 10.914 East US 2 / -46.7%
Standard_D2_v2 21.828 East US 2 / -46.7%
Standard_D3_v2 43.656 East US 2 / -46.5%
Standard_D4_v2 87.414 East US 2 / -46.6%
Standard_D5_v2 174.828 East US 2 / -46.6%
Standard_D11_v2 23.46 East US 2 / -35.2%
Standard_D12_v2 46.818 East US 2 / -34.9%
Standard_D13_v2 93.636 East US 2 / -34.9%
Standard_D14_v2 187.17 East US 2 / -34.8%
Standard_D15_v2 233.988 East US 2 / -34.7%
発想の逆転: 日本に置く必要のないシステムをUSに置けたら?
東日本 西ヨーロッパ
西イギリス 東アジア(香港)
m4.xlarge $0.258 -29%
m4.2xlarge $0.516 -29%
m4.4xlarge $1.032 -29%
" Oracle is hardly unique in having their own semiconductor team. Amazon does custom ASICs, Google
acquired an ARM team and has done custom ASIC for machine learning. Microsoft has done significant work
with FPGAs and is also an ARM licensee. ”
汎用・柔軟 効率・性能
( https://docs.microsoft.com/ja-jp/azure/virtual-machines/virtual-machines-linux-sizes )
Host
SmartNIC
FPGA
ToR
NIC ASIC
SmartNIC
CPU
Web search
ranking
Traditional software (CPU) server plane
QPICPU
QSFP
40Gb/s ToR
FPGA
CPU
40Gb/s
QSFP QSFP
Hardware acceleration plane
Web search
ranking
Deep neural
networks
SDN offload
SQL
• Cavium ThunderX2
• Qualcomm Centriq 2400
• モバイルデバイス向けに数が伸
び続けるARMへの期待
• 電力あたり性能
• エコシステム
• Windows ServerをARM対応中
• まずはAzure内、ユーザーが意識
しないところで
• ソフトウェアとしての販売は未定
https://schd.ws/hosted_files/ocpussummit2017/f2/OCP17%20Workshop_Microsoft%20Project%20Olympus%20Servers_3_8_2017.pdf
http://files.opencompute.org/oc/public.php?service=files&t=5e232f151532e2af3b9754af4ca79f0b
http://files.opencompute.org/oc/public.php?service=files&t=14ab3cf25170b7a0a439e11a3d818c96
https://azure.microsoft.com/en-us/blog/ecosystem-momentum-positions-microsoft-s-project-olympus-as-de-facto-open-compute-standard/
• “Hyperscale GPU Accelerator”
• NVIDIAとの共同開発
• Tesla P100 SMX2 x8 &
NVLink
• 4シャーシまで拡張可
• Internal PCIe Fabric
Interconnect
• 16サーバーまで共用可
“Price reductions on L Series and
announcing next generation Hyper-
threaded virtual machines”
(2017/4/3)
データセンターは世界にいくつ必要か
データセンターは世界にいくつ必要か
データセンターは世界にいくつ必要か
データセンターは世界にいくつ必要か

Weitere ähnliche Inhalte

Ähnlich wie データセンターは世界にいくつ必要か

Performance Schema and Sys Schema in MySQL 5.7
Performance Schema and Sys Schema in MySQL 5.7Performance Schema and Sys Schema in MySQL 5.7
Performance Schema and Sys Schema in MySQL 5.7Mark Leith
 
Molecular Shape Searching on GPUs: A Brave New World
Molecular Shape Searching on GPUs: A Brave New WorldMolecular Shape Searching on GPUs: A Brave New World
Molecular Shape Searching on GPUs: A Brave New WorldCan Ozdoruk
 
Consistency vs. Flexibility in Design Systems: A GE Case Study by Ken Skistimas
Consistency vs. Flexibility in Design Systems: A GE Case Study by Ken SkistimasConsistency vs. Flexibility in Design Systems: A GE Case Study by Ken Skistimas
Consistency vs. Flexibility in Design Systems: A GE Case Study by Ken Skistimasuxpin
 
Consistency vs Flexibility in Design Systems : A GE Case Study
Consistency vs Flexibility in Design Systems : A GE Case StudyConsistency vs Flexibility in Design Systems : A GE Case Study
Consistency vs Flexibility in Design Systems : A GE Case StudyKen Skistimas
 
Tales from the Postgres Front - and What We Can Learn
Tales from the Postgres Front - and What We Can LearnTales from the Postgres Front - and What We Can Learn
Tales from the Postgres Front - and What We Can LearnEDB
 
NetApp VM .tr-3749
NetApp VM .tr-3749NetApp VM .tr-3749
NetApp VM .tr-3749Accenture
 
3DConsulting_Presentation
3DConsulting_Presentation3DConsulting_Presentation
3DConsulting_PresentationJoseph Baca
 
How Service-Oriented Drive Deployments improve VSD Driveline Uptime
How Service-Oriented Drive Deployments improve VSD Driveline UptimeHow Service-Oriented Drive Deployments improve VSD Driveline Uptime
How Service-Oriented Drive Deployments improve VSD Driveline UptimeSchneider Electric
 
Migrations, Health Checks, and Support Experiences - Postgres from the Servic...
Migrations, Health Checks, and Support Experiences - Postgres from the Servic...Migrations, Health Checks, and Support Experiences - Postgres from the Servic...
Migrations, Health Checks, and Support Experiences - Postgres from the Servic...EDB
 
Ways to minimise performance risks in continuous delivery
Ways to minimise performance risks in continuous deliveryWays to minimise performance risks in continuous delivery
Ways to minimise performance risks in continuous deliverya32an
 
Oracle_Patching_Untold_Story_Final_Part2.pdf
Oracle_Patching_Untold_Story_Final_Part2.pdfOracle_Patching_Untold_Story_Final_Part2.pdf
Oracle_Patching_Untold_Story_Final_Part2.pdfAlex446314
 
dassault-systemes-catia-application-scalability-guide
dassault-systemes-catia-application-scalability-guidedassault-systemes-catia-application-scalability-guide
dassault-systemes-catia-application-scalability-guideJason Kyungho Lee
 
IRJET- Smart Tracking System for Healthcare Monitoring using GSM/GPS
IRJET- Smart Tracking System for Healthcare Monitoring using GSM/GPSIRJET- Smart Tracking System for Healthcare Monitoring using GSM/GPS
IRJET- Smart Tracking System for Healthcare Monitoring using GSM/GPSIRJET Journal
 
IRJET- A Review on Aluminium (LM 25) Reinforced with Boron Carbide (B4C) & Tu...
IRJET- A Review on Aluminium (LM 25) Reinforced with Boron Carbide (B4C) & Tu...IRJET- A Review on Aluminium (LM 25) Reinforced with Boron Carbide (B4C) & Tu...
IRJET- A Review on Aluminium (LM 25) Reinforced with Boron Carbide (B4C) & Tu...IRJET Journal
 
Dev and Ops Collaboration and Awareness at Etsy and Flickr
Dev and Ops Collaboration and Awareness at Etsy and FlickrDev and Ops Collaboration and Awareness at Etsy and Flickr
Dev and Ops Collaboration and Awareness at Etsy and FlickrJohn Allspaw
 
Ceph optimized Storage / Global HW solutions for SDS, David Alvarez
Ceph optimized Storage / Global HW solutions for SDS, David AlvarezCeph optimized Storage / Global HW solutions for SDS, David Alvarez
Ceph optimized Storage / Global HW solutions for SDS, David AlvarezCeph Community
 
Mk epn seminar-panel-for-public
Mk epn seminar-panel-for-publicMk epn seminar-panel-for-public
Mk epn seminar-panel-for-publicMiya Kohno
 
Disk health prediction for Ceph
Disk health prediction for CephDisk health prediction for Ceph
Disk health prediction for CephCeph Community
 
Supply chain design and operation
Supply chain design and operationSupply chain design and operation
Supply chain design and operationAngelainBay
 

Ähnlich wie データセンターは世界にいくつ必要か (20)

Performance Schema and Sys Schema in MySQL 5.7
Performance Schema and Sys Schema in MySQL 5.7Performance Schema and Sys Schema in MySQL 5.7
Performance Schema and Sys Schema in MySQL 5.7
 
Molecular Shape Searching on GPUs: A Brave New World
Molecular Shape Searching on GPUs: A Brave New WorldMolecular Shape Searching on GPUs: A Brave New World
Molecular Shape Searching on GPUs: A Brave New World
 
Consistency vs. Flexibility in Design Systems: A GE Case Study by Ken Skistimas
Consistency vs. Flexibility in Design Systems: A GE Case Study by Ken SkistimasConsistency vs. Flexibility in Design Systems: A GE Case Study by Ken Skistimas
Consistency vs. Flexibility in Design Systems: A GE Case Study by Ken Skistimas
 
Consistency vs Flexibility in Design Systems : A GE Case Study
Consistency vs Flexibility in Design Systems : A GE Case StudyConsistency vs Flexibility in Design Systems : A GE Case Study
Consistency vs Flexibility in Design Systems : A GE Case Study
 
Tales from the Postgres Front - and What We Can Learn
Tales from the Postgres Front - and What We Can LearnTales from the Postgres Front - and What We Can Learn
Tales from the Postgres Front - and What We Can Learn
 
NetApp VM .tr-3749
NetApp VM .tr-3749NetApp VM .tr-3749
NetApp VM .tr-3749
 
3DConsulting_Presentation
3DConsulting_Presentation3DConsulting_Presentation
3DConsulting_Presentation
 
How Service-Oriented Drive Deployments improve VSD Driveline Uptime
How Service-Oriented Drive Deployments improve VSD Driveline UptimeHow Service-Oriented Drive Deployments improve VSD Driveline Uptime
How Service-Oriented Drive Deployments improve VSD Driveline Uptime
 
Migrations, Health Checks, and Support Experiences - Postgres from the Servic...
Migrations, Health Checks, and Support Experiences - Postgres from the Servic...Migrations, Health Checks, and Support Experiences - Postgres from the Servic...
Migrations, Health Checks, and Support Experiences - Postgres from the Servic...
 
Ways to minimise performance risks in continuous delivery
Ways to minimise performance risks in continuous deliveryWays to minimise performance risks in continuous delivery
Ways to minimise performance risks in continuous delivery
 
Oracle_Patching_Untold_Story_Final_Part2.pdf
Oracle_Patching_Untold_Story_Final_Part2.pdfOracle_Patching_Untold_Story_Final_Part2.pdf
Oracle_Patching_Untold_Story_Final_Part2.pdf
 
dassault-systemes-catia-application-scalability-guide
dassault-systemes-catia-application-scalability-guidedassault-systemes-catia-application-scalability-guide
dassault-systemes-catia-application-scalability-guide
 
IRJET- Smart Tracking System for Healthcare Monitoring using GSM/GPS
IRJET- Smart Tracking System for Healthcare Monitoring using GSM/GPSIRJET- Smart Tracking System for Healthcare Monitoring using GSM/GPS
IRJET- Smart Tracking System for Healthcare Monitoring using GSM/GPS
 
IRJET- A Review on Aluminium (LM 25) Reinforced with Boron Carbide (B4C) & Tu...
IRJET- A Review on Aluminium (LM 25) Reinforced with Boron Carbide (B4C) & Tu...IRJET- A Review on Aluminium (LM 25) Reinforced with Boron Carbide (B4C) & Tu...
IRJET- A Review on Aluminium (LM 25) Reinforced with Boron Carbide (B4C) & Tu...
 
Machine Learning Impact on IoT - Part 2
Machine Learning Impact on IoT - Part 2Machine Learning Impact on IoT - Part 2
Machine Learning Impact on IoT - Part 2
 
Dev and Ops Collaboration and Awareness at Etsy and Flickr
Dev and Ops Collaboration and Awareness at Etsy and FlickrDev and Ops Collaboration and Awareness at Etsy and Flickr
Dev and Ops Collaboration and Awareness at Etsy and Flickr
 
Ceph optimized Storage / Global HW solutions for SDS, David Alvarez
Ceph optimized Storage / Global HW solutions for SDS, David AlvarezCeph optimized Storage / Global HW solutions for SDS, David Alvarez
Ceph optimized Storage / Global HW solutions for SDS, David Alvarez
 
Mk epn seminar-panel-for-public
Mk epn seminar-panel-for-publicMk epn seminar-panel-for-public
Mk epn seminar-panel-for-public
 
Disk health prediction for Ceph
Disk health prediction for CephDisk health prediction for Ceph
Disk health prediction for Ceph
 
Supply chain design and operation
Supply chain design and operationSupply chain design and operation
Supply chain design and operation
 

Mehr von Toru Makabe

インフラ廻戦 品川事変 前夜編
インフラ廻戦 品川事変 前夜編インフラ廻戦 品川事変 前夜編
インフラ廻戦 品川事変 前夜編Toru Makabe
 
Ingress on Azure Kubernetes Service
Ingress on Azure Kubernetes ServiceIngress on Azure Kubernetes Service
Ingress on Azure Kubernetes ServiceToru Makabe
 
細かすぎて伝わらないかもしれない Azure Container Networking Deep Dive
細かすぎて伝わらないかもしれない Azure Container Networking Deep Dive細かすぎて伝わらないかもしれない Azure Container Networking Deep Dive
細かすぎて伝わらないかもしれない Azure Container Networking Deep DiveToru Makabe
 
Demystifying Identities for Azure Kubernetes Service
Demystifying Identities for Azure Kubernetes ServiceDemystifying Identities for Azure Kubernetes Service
Demystifying Identities for Azure Kubernetes ServiceToru Makabe
 
Azure Blueprints - 企業で期待される背景と特徴、活用方法
Azure Blueprints - 企業で期待される背景と特徴、活用方法Azure Blueprints - 企業で期待される背景と特徴、活用方法
Azure Blueprints - 企業で期待される背景と特徴、活用方法Toru Makabe
 
ミッション : メガクラウドを安全にアップデートせよ!
ミッション : メガクラウドを安全にアップデートせよ!ミッション : メガクラウドを安全にアップデートせよ!
ミッション : メガクラウドを安全にアップデートせよ!Toru Makabe
 
俺の Kubernetes Workflow with HashiStack
俺の Kubernetes Workflow with HashiStack俺の Kubernetes Workflow with HashiStack
俺の Kubernetes Workflow with HashiStackToru Makabe
 
Resilience Engineering on Kubernetes
Resilience Engineering on KubernetesResilience Engineering on Kubernetes
Resilience Engineering on KubernetesToru Makabe
 
Real World Azure RBAC
Real World Azure RBACReal World Azure RBAC
Real World Azure RBACToru Makabe
 
Azure Kubernetes Service 2019 ふりかえり
Azure Kubernetes Service 2019 ふりかえりAzure Kubernetes Service 2019 ふりかえり
Azure Kubernetes Service 2019 ふりかえりToru Makabe
 
インフラ野郎AzureチームProX
インフラ野郎AzureチームProXインフラ野郎AzureチームProX
インフラ野郎AzureチームProXToru Makabe
 
NoOps Japan Community 1st Anniversary 祝辞
NoOps Japan Community 1st Anniversary 祝辞 NoOps Japan Community 1st Anniversary 祝辞
NoOps Japan Community 1st Anniversary 祝辞 Toru Makabe
 
ZOZOTOWNのCloud Native Journey
ZOZOTOWNのCloud Native JourneyZOZOTOWNのCloud Native Journey
ZOZOTOWNのCloud Native JourneyToru Makabe
 
Essentials of container
Essentials of containerEssentials of container
Essentials of containerToru Makabe
 
インフラ野郎 Azureチーム at クラウド boost
インフラ野郎 Azureチーム at クラウド boostインフラ野郎 Azureチーム at クラウド boost
インフラ野郎 Azureチーム at クラウド boostToru Makabe
 
ダイ・ハード in the Kubernetes world
ダイ・ハード in the Kubernetes worldダイ・ハード in the Kubernetes world
ダイ・ハード in the Kubernetes worldToru Makabe
 
半日でわかる コンテナー技術 (応用編)
半日でわかる コンテナー技術 (応用編)半日でわかる コンテナー技術 (応用編)
半日でわかる コンテナー技術 (応用編)Toru Makabe
 
インフラエンジニア エボリューション ~激変する IT インフラ技術者像、キャリアとスキルを考える~ at Tech Summit 2018
インフラエンジニア エボリューション ~激変する IT インフラ技術者像、キャリアとスキルを考える~ at Tech Summit 2018インフラエンジニア エボリューション ~激変する IT インフラ技術者像、キャリアとスキルを考える~ at Tech Summit 2018
インフラエンジニア エボリューション ~激変する IT インフラ技術者像、キャリアとスキルを考える~ at Tech Summit 2018Toru Makabe
 

Mehr von Toru Makabe (20)

インフラ廻戦 品川事変 前夜編
インフラ廻戦 品川事変 前夜編インフラ廻戦 品川事変 前夜編
インフラ廻戦 品川事変 前夜編
 
Ingress on Azure Kubernetes Service
Ingress on Azure Kubernetes ServiceIngress on Azure Kubernetes Service
Ingress on Azure Kubernetes Service
 
細かすぎて伝わらないかもしれない Azure Container Networking Deep Dive
細かすぎて伝わらないかもしれない Azure Container Networking Deep Dive細かすぎて伝わらないかもしれない Azure Container Networking Deep Dive
細かすぎて伝わらないかもしれない Azure Container Networking Deep Dive
 
Demystifying Identities for Azure Kubernetes Service
Demystifying Identities for Azure Kubernetes ServiceDemystifying Identities for Azure Kubernetes Service
Demystifying Identities for Azure Kubernetes Service
 
Azure Blueprints - 企業で期待される背景と特徴、活用方法
Azure Blueprints - 企業で期待される背景と特徴、活用方法Azure Blueprints - 企業で期待される背景と特徴、活用方法
Azure Blueprints - 企業で期待される背景と特徴、活用方法
 
ミッション : メガクラウドを安全にアップデートせよ!
ミッション : メガクラウドを安全にアップデートせよ!ミッション : メガクラウドを安全にアップデートせよ!
ミッション : メガクラウドを安全にアップデートせよ!
 
俺の Kubernetes Workflow with HashiStack
俺の Kubernetes Workflow with HashiStack俺の Kubernetes Workflow with HashiStack
俺の Kubernetes Workflow with HashiStack
 
Resilience Engineering on Kubernetes
Resilience Engineering on KubernetesResilience Engineering on Kubernetes
Resilience Engineering on Kubernetes
 
俺とHashiCorp
俺とHashiCorp俺とHashiCorp
俺とHashiCorp
 
Real World Azure RBAC
Real World Azure RBACReal World Azure RBAC
Real World Azure RBAC
 
Azure Kubernetes Service 2019 ふりかえり
Azure Kubernetes Service 2019 ふりかえりAzure Kubernetes Service 2019 ふりかえり
Azure Kubernetes Service 2019 ふりかえり
 
インフラ野郎AzureチームProX
インフラ野郎AzureチームProXインフラ野郎AzureチームProX
インフラ野郎AzureチームProX
 
NoOps Japan Community 1st Anniversary 祝辞
NoOps Japan Community 1st Anniversary 祝辞 NoOps Japan Community 1st Anniversary 祝辞
NoOps Japan Community 1st Anniversary 祝辞
 
ZOZOTOWNのCloud Native Journey
ZOZOTOWNのCloud Native JourneyZOZOTOWNのCloud Native Journey
ZOZOTOWNのCloud Native Journey
 
Ops meets NoOps
Ops meets NoOpsOps meets NoOps
Ops meets NoOps
 
Essentials of container
Essentials of containerEssentials of container
Essentials of container
 
インフラ野郎 Azureチーム at クラウド boost
インフラ野郎 Azureチーム at クラウド boostインフラ野郎 Azureチーム at クラウド boost
インフラ野郎 Azureチーム at クラウド boost
 
ダイ・ハード in the Kubernetes world
ダイ・ハード in the Kubernetes worldダイ・ハード in the Kubernetes world
ダイ・ハード in the Kubernetes world
 
半日でわかる コンテナー技術 (応用編)
半日でわかる コンテナー技術 (応用編)半日でわかる コンテナー技術 (応用編)
半日でわかる コンテナー技術 (応用編)
 
インフラエンジニア エボリューション ~激変する IT インフラ技術者像、キャリアとスキルを考える~ at Tech Summit 2018
インフラエンジニア エボリューション ~激変する IT インフラ技術者像、キャリアとスキルを考える~ at Tech Summit 2018インフラエンジニア エボリューション ~激変する IT インフラ技術者像、キャリアとスキルを考える~ at Tech Summit 2018
インフラエンジニア エボリューション ~激変する IT インフラ技術者像、キャリアとスキルを考える~ at Tech Summit 2018
 

Kürzlich hochgeladen

Unraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfUnraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfAlex Barbosa Coqueiro
 
My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024The Digital Insurer
 
Powerpoint exploring the locations used in television show Time Clash
Powerpoint exploring the locations used in television show Time ClashPowerpoint exploring the locations used in television show Time Clash
Powerpoint exploring the locations used in television show Time Clashcharlottematthew16
 
What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024Stephanie Beckett
 
Unleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubUnleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubKalema Edgar
 
CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):comworks
 
Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 3652toLead Limited
 
"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii SoldatenkoFwdays
 
SIP trunking in Janus @ Kamailio World 2024
SIP trunking in Janus @ Kamailio World 2024SIP trunking in Janus @ Kamailio World 2024
SIP trunking in Janus @ Kamailio World 2024Lorenzo Miniero
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationRidwan Fadjar
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek SchlawackFwdays
 
Training state-of-the-art general text embedding
Training state-of-the-art general text embeddingTraining state-of-the-art general text embedding
Training state-of-the-art general text embeddingZilliz
 
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024BookNet Canada
 
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr LapshynFwdays
 
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Mark Simos
 
Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Enterprise Knowledge
 
Developer Data Modeling Mistakes: From Postgres to NoSQL
Developer Data Modeling Mistakes: From Postgres to NoSQLDeveloper Data Modeling Mistakes: From Postgres to NoSQL
Developer Data Modeling Mistakes: From Postgres to NoSQLScyllaDB
 
DevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsDevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsSergiu Bodiu
 

Kürzlich hochgeladen (20)

Unraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdfUnraveling Multimodality with Large Language Models.pdf
Unraveling Multimodality with Large Language Models.pdf
 
My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024My INSURER PTE LTD - Insurtech Innovation Award 2024
My INSURER PTE LTD - Insurtech Innovation Award 2024
 
DMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special EditionDMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special Edition
 
Powerpoint exploring the locations used in television show Time Clash
Powerpoint exploring the locations used in television show Time ClashPowerpoint exploring the locations used in television show Time Clash
Powerpoint exploring the locations used in television show Time Clash
 
What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024
 
Unleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubUnleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding Club
 
CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):
 
Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365
 
"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko
 
SIP trunking in Janus @ Kamailio World 2024
SIP trunking in Janus @ Kamailio World 2024SIP trunking in Janus @ Kamailio World 2024
SIP trunking in Janus @ Kamailio World 2024
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 Presentation
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
 
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptxE-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
 
Training state-of-the-art general text embedding
Training state-of-the-art general text embeddingTraining state-of-the-art general text embedding
Training state-of-the-art general text embedding
 
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
 
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
"Federated learning: out of reach no matter how close",Oleksandr Lapshyn
 
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
 
Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024Designing IA for AI - Information Architecture Conference 2024
Designing IA for AI - Information Architecture Conference 2024
 
Developer Data Modeling Mistakes: From Postgres to NoSQL
Developer Data Modeling Mistakes: From Postgres to NoSQLDeveloper Data Modeling Mistakes: From Postgres to NoSQL
Developer Data Modeling Mistakes: From Postgres to NoSQL
 
DevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platformsDevEX - reference for building teams, processes, and platforms
DevEX - reference for building teams, processes, and platforms
 

データセンターは世界にいくつ必要か

  • 1.
  • 2. { “名前” : “真壁 徹(まかべ とおる)”, “所属” : “日本マイクロソフト株式会社”, “役割” : “クラウド ソリューションアーキテクト”, “経歴” : “大和総研  HP Enterprise”, “特技” : “クラウド & オープンソース” }
  • 4. http://fortune.com/2017/04/12/mark-hurd-oracle-data-centers/ "If I have two-times faster computers, I don't need as many data centers. If I can speed up the database, maybe I need one fourth as may data centers. I can go on and on about how tech drives this.“
  • 5.
  • 6.
  • 7.
  • 8. [Microsoft Global Datacenters and Network Infrastructure] https://www.youtube.com/watch?v=bqZrejosqWU
  • 10.
  • 11. リージョンペア リージョン リージョン リージョン AWS re:Invent 2014 “AWS Innovation at Scale with James Hamilton” より
  • 13.
  • 14. "There is some absolute data center size where the facility becomes “too big to fail.” This line is gray and open to debate but the limiting factor is how big of a facility can an operator lose before the lost resources and the massive network access pattern changes on failure can’t be hidden from customers.”
  • 15. 1.Microsoft: 35 MW with CloudHQ in Manassas, Virginia 2.Microsoft: 30 MW with EdgeConneX in Elk Grove Village, Illinois 3.Microsoft: 22 MW with CyrusOne in Ashburn, Virginia 4.Microsoft: 16 MW with DuPont Fabros Technology in Santa Clara, California 5.Microsoft: 13.5 MW with CyrusOne in Phoenix, Arizona 6.Microsoft: 9 MW with CyrusOne in San Antonio, Texas 7.Oracle: 6 MW with CyrusOne in Sterling, Virginia 8.Salesforce: 6 MW with QTS in Dallas, Texas 9.Oracle: 5 MW with CyrusOne in Ashburn, Virginia 10.Oracle: 5 MW with Digital Realty Trust in Ashburn, Virginia http://www.datacenterknowledge.com/archives/2017/01/18/who-leased-the-most-data-center-space-in-2016/ Microsoft、AWSともに自社建設、保有分は入っていないので雰囲気だけ
  • 16. Colocation Density 2.0+ Power Usage Effectiveness (PUE) 1.4 – 1.6 PUE Discrete servers Capacity 20 year technology Rack Density & deployment Minimized resource impact Generation 1 Generation 2 Containment Modular Hyper-scale 1.2 – 1.5 PUE 1.12 – 1.20 PUE 1.07 – 1.19 PUE Containers, PODs Scalability & sustainability Air & water Economization Differentiated SLAs Deployment Areas & ITPACs No more traditional IT Right-sized Faster time-to-market Outside air cooled Fully integrated Resilient software Common infrastructure Operational simplicity Flexible & scalable Generation 3 Generation 4 Generation 5
  • 17.
  • 20. Standard_D1_v2 10.404 East US 2 / -44.1% Standard_D2_v2 20.91 East US 2 / -44.4% Standard_D3_v2 41.718 East US 2 / -44% Standard_D4_v2 83.436 East US 2 / -44% Standard_D5_v2 166.872 East US 2 / -44% Standard_D11_v2 23.46 East US 2 / -35.2% Standard_D12_v2 46.818 East US 2 / -34.9% Standard_D13_v2 93.636 East US 2 / -34.9% Standard_D14_v2 187.17 East US 2 / -34.8% Standard_D15_v2 233.988 East US 2 / -34.7% Standard_D1_v2 6.936 East US 2 / -16.2% Standard_D2_v2 13.872 East US 2 / -16.2% Standard_D3_v2 27.744 East US 2 / -15.8% Standard_D4_v2 55.488 East US 2 / -15.8% Standard_D5_v2 110.874 East US 2 / -15.7% Standard_D11_v2 19.38 East US 2 / -21.6% Standard_D12_v2 38.658 East US 2 / -21.1% Standard_D13_v2 77.418 East US 2 / -21.2% Standard_D14_v2 154.836 East US 2 / -21.1% Standard_D15_v2 193.494 East US 2 / -21.1% Standard_D1_v2 8.976 East US 2 / -35.2% Standard_D2_v2 17.952 East US 2 / -35.2% Standard_D3_v2 35.802 East US 2 / -34.8% Standard_D4_v2 71.604 East US 2 / -34.8% Standard_D5_v2 143.31 East US 2 / -34.8% Standard_D11_v2 23.868 East US 2 / -36.3% Standard_D12_v2 47.94 East US 2 / -36.4% Standard_D13_v2 95.574 East US 2 / -36.2% Standard_D14_v2 191.148 East US 2 / -36.1% Standard_D15_v2 238.986 East US 2 / -36.1% Standard_D1_v2 10.914 East US 2 / -46.7% Standard_D2_v2 21.828 East US 2 / -46.7% Standard_D3_v2 43.656 East US 2 / -46.5% Standard_D4_v2 87.414 East US 2 / -46.6% Standard_D5_v2 174.828 East US 2 / -46.6% Standard_D11_v2 23.46 East US 2 / -35.2% Standard_D12_v2 46.818 East US 2 / -34.9% Standard_D13_v2 93.636 East US 2 / -34.9% Standard_D14_v2 187.17 East US 2 / -34.8% Standard_D15_v2 233.988 East US 2 / -34.7% Azure 仮想マシン(Linux)の価格(円)を、最安値リージョンと比較すると 東日本 西ヨーロッパ 西イギリス 東アジア(香港) m4.xlarge $0.258 -29% m4.2xlarge $0.516 -29% m4.4xlarge $1.032 -29%
  • 21.
  • 22. "Eventually it just make better sense to offer customers alternative regions rather than attempting to scale a single region and the argument in favor of multiple regions become even stronger when other factors are considered.”
  • 23.
  • 24.
  • 25. Azure Regionオンプレミス(ユーザー占有型) User User User User Storage Cluster Storage Cluster Server Cluster Server Cluster 共用部分が大きい
  • 26. Azure Regionオンプレミス(ユーザー占有型) User User User User Storage Cluster Storage Cluster Server Cluster Server Cluster 影響範囲 ストレージクラスターが故障した場合 影響範囲が大きい
  • 27. Azure Regionオンプレミス(ユーザー占有型) User User User User Storage Cluster Storage Cluster Server Cluster Server Cluster データセンター全体で共有している設備の障害影響は、 オンプレもAzureも同様
  • 31.
  • 32. Traditional on-premises Modern cloud Monolithic, centralized Design for predictable scalability Relational database Strong consistency Serial and synchronized processing Design to avoid failures (MTBF) Occasional big updates Manual management Snowflake servers Decomposed, de-centralized Design for elastic scale Polyglot persistence (mix of storage technologies) Eventual consistency Parallel and asynchronous processing Design for failure (MTTR) Frequent small updates Automated self-management Immutable infrastructure Azure Application Architecture Guide https://docs.microsoft.com/ja-jp/azure/architecture/guide/
  • 33.
  • 34.
  • 35. "But, without a more fundamental solution to the speed of light problem, many regions are the only practical way to effectively serve the entire planet for many workloads.”
  • 37. "Imagine a $20,000 car in one market costing far more than $200,000 in another market.”
  • 38.
  • 40. $azure location list --details info: Executing command location list + Getting ARM registered providers info: Getting locations... data: data: Location : japaneast data: DisplayName : East Japan data: data: […]
  • 41.
  • 42.
  • 44.
  • 45. " In addition, some national jurisdictions will put in place legal restrictions than make it difficult to fully serve the market without a local region. ”
  • 46. Standard_D1_v2 10.404 East US 2 / -44.1% Standard_D2_v2 20.91 East US 2 / -44.4% Standard_D3_v2 41.718 East US 2 / -44% Standard_D4_v2 83.436 East US 2 / -44% Standard_D5_v2 166.872 East US 2 / -44% Standard_D11_v2 23.46 East US 2 / -35.2% Standard_D12_v2 46.818 East US 2 / -34.9% Standard_D13_v2 93.636 East US 2 / -34.9% Standard_D14_v2 187.17 East US 2 / -34.8% Standard_D15_v2 233.988 East US 2 / -34.7% Standard_D1_v2 6.936 East US 2 / -16.2% Standard_D2_v2 13.872 East US 2 / -16.2% Standard_D3_v2 27.744 East US 2 / -15.8% Standard_D4_v2 55.488 East US 2 / -15.8% Standard_D5_v2 110.874 East US 2 / -15.7% Standard_D11_v2 19.38 East US 2 / -21.6% Standard_D12_v2 38.658 East US 2 / -21.1% Standard_D13_v2 77.418 East US 2 / -21.2% Standard_D14_v2 154.836 East US 2 / -21.1% Standard_D15_v2 193.494 East US 2 / -21.1% Standard_D1_v2 8.976 East US 2 / -35.2% Standard_D2_v2 17.952 East US 2 / -35.2% Standard_D3_v2 35.802 East US 2 / -34.8% Standard_D4_v2 71.604 East US 2 / -34.8% Standard_D5_v2 143.31 East US 2 / -34.8% Standard_D11_v2 23.868 East US 2 / -36.3% Standard_D12_v2 47.94 East US 2 / -36.4% Standard_D13_v2 95.574 East US 2 / -36.2% Standard_D14_v2 191.148 East US 2 / -36.1% Standard_D15_v2 238.986 East US 2 / -36.1% Standard_D1_v2 10.914 East US 2 / -46.7% Standard_D2_v2 21.828 East US 2 / -46.7% Standard_D3_v2 43.656 East US 2 / -46.5% Standard_D4_v2 87.414 East US 2 / -46.6% Standard_D5_v2 174.828 East US 2 / -46.6% Standard_D11_v2 23.46 East US 2 / -35.2% Standard_D12_v2 46.818 East US 2 / -34.9% Standard_D13_v2 93.636 East US 2 / -34.9% Standard_D14_v2 187.17 East US 2 / -34.8% Standard_D15_v2 233.988 East US 2 / -34.7% 発想の逆転: 日本に置く必要のないシステムをUSに置けたら? 東日本 西ヨーロッパ 西イギリス 東アジア(香港) m4.xlarge $0.258 -29% m4.2xlarge $0.516 -29% m4.4xlarge $1.032 -29%
  • 47.
  • 48. " Oracle is hardly unique in having their own semiconductor team. Amazon does custom ASICs, Google acquired an ARM team and has done custom ASIC for machine learning. Microsoft has done significant work with FPGAs and is also an ARM licensee. ”
  • 51.
  • 53. Web search ranking Traditional software (CPU) server plane QPICPU QSFP 40Gb/s ToR FPGA CPU 40Gb/s QSFP QSFP Hardware acceleration plane Web search ranking Deep neural networks SDN offload SQL
  • 54. • Cavium ThunderX2 • Qualcomm Centriq 2400 • モバイルデバイス向けに数が伸 び続けるARMへの期待 • 電力あたり性能 • エコシステム • Windows ServerをARM対応中 • まずはAzure内、ユーザーが意識 しないところで • ソフトウェアとしての販売は未定 https://schd.ws/hosted_files/ocpussummit2017/f2/OCP17%20Workshop_Microsoft%20Project%20Olympus%20Servers_3_8_2017.pdf http://files.opencompute.org/oc/public.php?service=files&t=5e232f151532e2af3b9754af4ca79f0b
  • 55. http://files.opencompute.org/oc/public.php?service=files&t=14ab3cf25170b7a0a439e11a3d818c96 https://azure.microsoft.com/en-us/blog/ecosystem-momentum-positions-microsoft-s-project-olympus-as-de-facto-open-compute-standard/ • “Hyperscale GPU Accelerator” • NVIDIAとの共同開発 • Tesla P100 SMX2 x8 & NVLink • 4シャーシまで拡張可 • Internal PCIe Fabric Interconnect • 16サーバーまで共用可
  • 56. “Price reductions on L Series and announcing next generation Hyper- threaded virtual machines” (2017/4/3)