Transcript SD-B307: Optimize Your Data Center with Datacenter Services from

Adam Fazio
Solution Architect
WW Datacenter COE
Microsoft Services
David Ziembicki
Solution Architect
Office of the Americas CTO
Microsoft Services
Lead Architect
for Microsoft
Services Private
Cloud
Solutions
Contributor to:
Blog: http://davidzi.com
Twitter: http://twitter.com/davidzi
Lead Architect
for Microsoft
Services Private
Cloud
Solutions
Contributor to:
Blog:
http://technet.microsoft.com/privatecloud
Twitter: http://twitter.com/adamfazio
• Who is Microsoft Services and what is the Datacenter Services
Portfolio?
• What is the IaaS Product Line Architecture?
•
•
• Management Architecture (System Center)
• Deployment Scenarios
• Availability and Call to Action
World Class Expertise Globally, innovating to
demonstrate what is possible with Microsoft
Distinctive Intellectual Property (IP) to allow
our people and partners to deliver impact
Long-term Customer Relationships
committed to long-term customer success
Connection with our Product Groups to
resolve problems and influence product
roadmaps
A Relentless Focus on Driving Customer
Impact through architecture, deployment,
adoption, and support, helping to make sure our
customers use and get value from their
investments in Microsoft, providing a
competitive differentiator for Microsoft
•
13 areas / 82 countries
•
17,341 employees
•
44 languages
•
35,000 partners
•
715 million customer
touch points per year
Canada
349
WE
1939
CEE
886
UK
919
US
6588
France
519
Germany
870
Japan
789
Latam
516
MEA
547
Strategy
GCR
770
Consulting
India
1891
APAC
758
Support
World Class Expertise Globally, through
datacenter architecture, virtualization,
management, and operations
•
13 areas / 28 countries
•
100 Architects
Distinctive Intellectual Property (IP) to enable
rapid and consistent deployment of the
Microsoft cloud platform
•
300 Consultants
•
Hundreds of partners
Deep Industry Expertise in the Commercial and
Public Sectors
•
Hundreds of customer
projects delivered per year
Connection with our Product Groups to map
customer requirements to product design
A Relentless Focus on Driving Customer
Impact through architecture, deployment,
adoption, and support, helping to make sure our
customers use and get value from their
investments in the Microsoft cloud platform
Private
Cloud
Public
Cloud
Hybrid
Cloud
Current Architecture Landscape
We need a consistent
way to deploy…
• Mixed guidance on how to architect,
plan, design Microsoft solutions
Virtually unlimited number of
configuration variations
• “Islands of Best Practice” are
commonplace
• Delivering various levels of cloud
attributes across implementations
• Long planning and deployment cycles
Packaged
Services
Deployment
Frameworks
Product Line
Architectures
Extensible
Platform
Packaged
solution for
deploying highlyautomated
datacenter and
cloud services
Frameworks for
assessing,
planning, and
deploying
datacenter
capabilities
Best practices for
architecture,
design, and
operations of
datacenter and
cloud
infrastructures
Solution platform
provides
capability for
extensions and
value-added
services
Intellectual Property
IaaS Fabric
Architecture
IaaS Management
Architecture
Hybrid Cloud
VDI
Large/Complex Environments
Multi-Site / Disaster Recovery
Services Offerings
Content
Datacenter Infrastructure &
Management
Reference
Architectures
Datacenter Consolidation &
Migration
Project Plans
Datacenter Automation
Deployment Guides
Datacenter Modernization
Tools and
Automation
Datacenter Operations
Operations Guides
What is a PLA?
A Product Line Architecture (PLA) is a set of rules and architectural
patterns that provide a method for predictable and supportable
deployments
Microsoft evolves the Product Line Architecture as we capture
learnings from customer deployments and support incidents to
continuously improve the architecture and our deployment practices.
The Benefits of using the PLA include:
Rulebook
• Proven deployment model
• Deployed right the first time
• Reduced time to deployment
•
•
•
Predictable and achievable SLA’s
The best user experience
All workload capabilities realized
The PLA delivers the highest quality deployment by following design principles to support/enable …
Solution
Offerings
MCS
Deployments
Early
Adoption
and OEM
Programs
Guidance
(Mandatory, Recommended, Optional)
Engagement
Delivery / PG
Updates
Request
change /
new rule
Approve
Updates /
Release
OEM
IaaS PLA
• Fabric PLA
• Management PLA
Fast Track (OEM)
Fabric Architecture
Validation Guidance
Windows Server 2012
Design Patterns
Fast Track (OEM)
Fabric Management
Architecture Validation
Guidance
System Center 2012 SP1
Design Pattern
Order Hardware in
Days vs. Weeks
Workloads
…
Fabric Architecture
Guidance
IaaS PLA
Fast Track version 3
Private Cloud Fast Track
• Fabric Validation Guide
• Mgmt Validation Guide
Service Provider/Enterprise
Windows Server 2012
Design Patterns
Fabric Management
Architecture Guidance
System Center 2012 SP1
Design Pattern
Design and Deploy in
Weeks vs. Months
…
Scenarios
Hybrid
Cloud
Cloud Packs
Multi-Site
Datacenters
and DR
End-to-End
Monitoring
Service
Mgmt
Datacenter
Chargeback
/Showback
Complex
System
Center
IT Process
Automation
Deploy Enable workloads and complex scenarios in
Months vs. Years
Continuous
Availability
over SMB
Storage
Pattern
Converged
Infrastructure
Pattern
NonConverged
Infrastructure
Pattern
Teamed Network
Interfaces (LBFO)
Volume1
Volume-n
Cluster Shared Volumes (CSV2)
Hyper-V Cluster
FC
Volumes
SAN Based Storage
iSCSI
Optimal for
leveraging an
existing SAN
investment
FC
Large or recent
investment in
server/blade
hardware with
HBAs
iSCSI
Emphasis on
minimizing risk
and cost of new
hardware and
designs
Hyper-V Host Cluster(s)
SAN Storage
Volumes
Fibre Channel/
iSCSI
Fibre Channel/
iSCSI
Fibre Channel/
iSCSI
Fibre Channel/
iSCSI
VMs
VMs
Hyper-V
Extensible
Switch
VHDs
Cluster Shared Volumes
(CSV2) + CSV Cache
Eth
NIC
Eth
NIC
NIC
Teaming
(LBFO)
LAN
Non-Converged -Physical Host and Network Design
Converged Infrastructure Pattern
Cost savings and
operational
efficiency of a
single common
Ethernet network
vs. multiple
physical networks
and HBAs for
storage traffic
Large or recent
investment in
blade servers with
convergednetwork and
storage-network
infrastructure
Optimal for
leveraging a
converged-network
architecture
Hyper-V Cluster
Volume1
Volume-n
Cluster Shared Volumes (CSV2)
Converged
Network
Architecture
Volumes
SAN Based Storage
Converged Infrastructure Pattern
Hyper-V Host Cluster(s)
SAN Storage
Volumes
Fiber Channel /
iSCSI
Fiber Channel /
iSCSI
Fiber Channel
iSCSI
SMB Direct
LAN
CNA
NIC
Teaming
CNA
Hyper-V
Extensible
Switch
VMs
VMs
VHDs
Cluster Shared Volumes
(CSV2) + CSV Cache
Converged -Physical Host and Network Design
Continuous Availability over SMB Storage Pattern
Teamed Network
Interfaces (LBFO)
Optimal for low
cost JBOD
strategy enabled
by networking
capabilities of
SMB
Multichannel and
RDMA
Hyper-V Cluster
RDMA
\\cno\share-1
\\cno\share-n
SMB Direct Shares
(SMB3 over RDMA)
RDMA
Clustered
File Servers
Storage
Spaces
SAS
Requires
network
infrastructure
between the
Hyper-V servers
and file servers
SAS
Hyper-V clusters
consuming
storage from
scale-out file
server clusters or
SMB3 enabled
storage devices
Shared SAS Storage
Variation A
SMB Direct using
Shared SAS/Storage
Spaces
Variation B
SMB Direct using
SAN
Variation A
SMB Direct using
Shared SAS/Storage
Spaces
Hyper-V Host Cluster(s)
Scale-Out File Server Cluster
VHDs
Cluster Shared Volumes
(CSV2) + CSV Cache
Storage Spaces
SAS
Controller
SAS
Controller
SAS Disks
RDMA
NIC
RDMA
NIC
SMB3 Direct
SMB3 Transparent Failover
SMB3 Multi-Channel
RDMA
NIC
RDMA
NIC
VMs
VMs
Hyper-V
Extensible
Switch
Eth
NIC
Eth
NIC
NIC
Teaming
LAN
10Gb-E
10Gb-E
RDMA Port
RDMA Port
Scale-Out File Server Cluster Node
10Gb-E
10Gb-E
RDMA Port
RDMA Port
Scale-Out File Server Cluster Node
VHDs
VHDs
CA Over SMB
Variation A
Storage Design with
Scale-Out File
Server Cluster,
Storage Spaces, SAS
Cluster Shared Volumes (CSV v2) + CSV Cache
Storage Spaces
Storage Pool(s)
SAS HBA
SAS HBA
SAS HBA
SAS HBA
SAS SAS
Port Port
SAS SAS
Port Port
SAS SAS
Port Port
SAS SAS
Port Port
SAS Expander
SAS Expander
SAS Disks
SAS JBOD Array with Dual Expander/Dual Port Drives
10Gb-E
10Gb-E
RDMA Port
RDMA Port
Scale-Out File Server Cluster Node
10Gb-E
10Gb-E
RDMA Port
RDMA Port
Scale-Out File Server Cluster Node
VHDs
VHDs
Cluster Shared Volumes (CSV v2) + CSV Cache
CA Over SMB
Variation A
Storage Design with
Scale-Out File
Server Cluster,
Storage Spaces,
Switched SAS
Storage Spaces
Storage Pool(s)
SAS HBA
SAS HBA
SAS HBA
SAS HBA
SAS SAS
Port Port
SAS SAS
Port Port
SAS SAS
Port Port
SAS SAS
Port Port
SAS Switch
SAS
Expander
SAS
Expander
SAS Switch
SAS
Expander
SAS
Expander
SAS
Expander
SAS
Expander
SAS Disks
SAS Disks
SAS Disks
SAS JBOD Array with Dual
Expander/Dual Port Drives
SAS JBOD Array with Dual
Expander/Dual Port Drives
SAS JBOD Array with Dual
Expander/Dual Port Drives
10Gb-E
10Gb-E
RDMA Port
RDMA Port
Scale-Out File Server Cluster Node
10Gb-E
10Gb-E
RDMA Port
RDMA Port
Scale-Out File Server Cluster Node
VHDs
VHDs
CA Over SMB
Variation A
Storage Design with
Scale-Out File
Server Cluster,
Clustered HW RAID
(no spaces), SAS
Cluster Shared Volumes (CSV v2) + CSV Cache
PCI Clustered RAID Volumes
PCI RAID
PCI RAID
PCI RAID
PCI RAID
SAS SAS
Port Port
SAS SAS
Port Port
SAS SAS
Port Port
SAS SAS
Port Port
SAS Expander
SAS Expander
SAS Disks
SAS JBOD Array with Dual Expander/Dual Port Drives
Continuous Availability over SMB Storage Pattern
Service
Delivery
Fabric
Management
Component
Scalability Reference
Virtual Machine Manager
800 hosts/25,000 virtual machines per instance
App Controller
Operations Manager
Orchestrator
Service Manager
Notes
A VMM instance is defined as a standalone or cluster
installation. Scalability is limited to 5000 virtual machines
when Service Provider Foundation (SPF) is installed. A single
SPF installation can support up to five Virtual Machine
Manager instances.
Scalability is proportional to Virtual Machine Manager Supports 250 virtual machines per Virtual Machine
(VMM)
Manager User Role
3,000 agents per management server, 15,000 agents
per management group
Simultaneous execution of 50 runbooks per Runbook
server
Large deployment supports up to 20,000 computers
Topology dependent. Note that in Fast Track Service
Manager is used solely for private cloud virtual machine
management. An advanced deployment topology can
support up to 50,000 computers
SCDB
SCSMDB
SCSMDW
SCSMAS
SCVMMDB
SCOMDB
SCOMDW
SCOMASRS
SharePoint_Config
ServiceManager
CMDWDataMart
SCSM SSAS
VirtualManagerDB
OperationsManager
OperationsManagerDW
SSAS and SSRS
Installed
Remotely on
the SCOM
Reporting
Server
SharePoint_Content
DBs
OMDWDataMart
WSS DBs
DWDataMart
Orchestrator
DWStagingAndConfig
AppController
DWSRepository
WSUS DB
ReportServer
Optional Component
ReportServerTempDB
ReportServer
ReportServerTempDB
LUN2: Data
LUN4: Data
LUN6: Data
LUN8: Data
LUN10: Data
LUN12: Data
LUN14: Data
LUN3: Logs
LUN5: Logs
LUN7: Logs
LUN9: Logs
LUN11: Logs
LUN13: Logs
LUN15: Logs
LUN1: Quorum
Why make VMM highly available?
•
•
•
VMM is a key part of the Private Cloud management toolset
VMM HA is a fault tolerant service feature, but does not increase scale/performance
For complete highly availability, all SCVMM components need to be HA
•
•
•
SQL
Library
VMM setup is integrated with Failover Clustering
•
•
VMM console connections should use the cluster name of the HA VMM service instead of an
individual node
Connectors
•
•
Ports , Service Accounts
TechNet guidance is focused on deployment of the Component scenario
•
•
Not developed for Private Cloud deployment
SharePoint considerations
•
•
•
•
SQL/SharePoint and Portal
Certificates
Version
Default behavior of the connector will synchronize ALL objects from
Active Directory
• Solution – Be targeted
•
Example :
1.
Use wildcard for group name prefixes in AD Connector properties
I.E. – “cn=SM*” for ‘SMAdmins’, ‘SMUsers’, etc.
2.
Select “Automatically add users of AD groups imported by this connector”
http://SCO2012SQL:81/Orchestrator2012/Orchestrator.svc
Azure
App
Controller
Portal
Service
Manager
Service
Manager
Portal
CMDB
SM Data
Warehouse
Integration
Pack
CI Connector
Active
Directory
OM Data
Warehouse
Orchestrator
Virtual
Machine
Manager
VMM/OM Integration
Hyper-V
Operations
Manager
Reporting
Data
• The IaaS Product Line Architecture is the Microsoft best practices reference
architecture for Windows Server, Hyper-V, and System Center Deployment
for private cloud IaaS
• The PLA is the basis for Microsoft, OEM, and Partner Programs
• The PLA was created by a team of experts from across the world:
Do-it-Yourself
Microsoft Consulting Services
Partner
Private Cloud Fast Track
Executive and Technical Briefings Practice Accelerators
PLA Reference Architecture &
Deployment Guide
blogs.technet.com/privatecloud
(coming soon!)
Workshops
Microsoft Virtual Academy
microsoftvirtualacademy.com
Proof-of-Concept
Jumpstart Engagement
(Rapid, 5-week deployment)
Full Engagement
Datacenter Services Partner
Program