SP2013 Traditional Topology Model
SP2013 Traditional Topology Model
Example topologies
Limited deployments (1-2 servers) Small multipurpose farms (3-4 servers) Medium farm architectures (6+ servers) Medium farm with Office Web Apps Server and articulated search roles
Limited deployments are typically used for product Small farm architectures serve a larger number of users and scale out based on how heavily services are used. Not all small farms are fault-tolerant. Medium farm architectures can be multi-purpose or optimized for specific purposes. This farm illustrates a fully fault-tolerant, virtual environment that includes Office Web Apps Server VMs and a plan for
evaluation, development and testing, or for environments Medium-size farms are fully fault-tolerant. Some environments might require more web scaling out databases. This diagram also calls out all search application roles.
that have limited numbers of users and don t require Three-server virtualized farm Four-server physical farms Smallest fault-tolerant farm utilizing virtualization servers. Factor 10,000 users per web server as a starting point.
fault-tolerance. Use virtualization to maximize the potential Add a dedicated application server for All farm server roles virtualized and distributed across two or four host servers (depending
of a smaller number of servers. environments with moderate service usage. on the operating system) to provide fault tolerance using the minimum number of servers. Six-server physical farm Six-server virtualized farm Windows Server 2008 R2
One-server farm Two-tier farm Windows Server 2008 R2
Evaluation or <100 users Up to 10,000 users Two web servers are predicted to serve Windows Server 2008 R2 Windows Server 2012 Load Balancer
Host A Host B
10,000-20,000 users. Web Servers
All Web and Host A Host B Host A Host B Host A Host A
All roles on one Web Server Web Server
server, including application
SQL Server server roles Web/Query server Web Server Web Server Web Server Web Server Web Office Web Office Office Web Apps Server VMs can share the same host as a SharePoint Web or
Host A Host A Application servers Server Web Server Web application server. If more than one Office Web Apps Server VM is included in the
Apps Apps architecture, add a load balancer or configure Application Request Routing in IIS.
running all service Server Server
application roles Web Server Dedicated
Databases Web Server Web Server Application Server Web Server
All All All All for Crawling
Application Application Application Application Host C Host D
Server Roles Server Roles Server Roles Server Roles All SharePoint
All All Databases The index is stored across replicas.
All SharePoint Databases Application Server Application Server
Application Application Host C Host D Each replica for a given index
Server Roles Server Roles Host C Host D Query Processing Query Processing partition contains the same data. The
Development Environments Query and Query and
Replica Index Partition 0 Replica
data within index replicas is stored in
the file system on the server.
Index Index Search Databases
Mimic a three-tier environment by using virtualization. If fault-tolerance of services is more important than user
performance, configure the farm with redundant
All
Databases
All
Databases
All
Databases
All
Databases
application servers instead of redundant Web servers. Eight-server physical farm optimized Application Server – All other Application Server – All other The rest of the search components: Crawl DB
Application Server Roles Application Server Roles Crawl, Search Admin, Content
Host A All SharePoint Databases for search All other All other
Crawl Crawl
Processing Component, Analytics Admin DB
Application Application Processing Component. Link DB
Web Server Web server SQL Server installed and configured to support SQL SQL Server installed and configured to support SQL Server Roles Server Roles
clustering, mirroring, or AlwaysOn. AlwaysOn requires clustering, mirroring, or AlwaysOn. AlwaysOn requires Web Servers Analytics DB
High availability — For environments above 1,000 users, two SQL Server 2012. SQL Server 2012.
servers utilizing SQL Server clustering, mirroring, or AlwaysOn Host E Host F All other Services DB
is recommended. AlwaysOn requires SQL Server 2012. Host E Host F
All Application Server Application Servers Virtualizing with Windows Server 2008 R2 versus Windows Server 2012 All SharePoint Databases All SharePoint Databases
Roles Windows Server 2012 allows a greater number of resources to be allocated to each virtual machine, and thus allows running a Application servers All All Scale out the database layer by first Config DB
SharePoint farm with fewer virtual machines than Windows Server 2008 R2. Due to the virtual architectures that are possible, this running all service Databases Databases Crawl DB
separating content DBs from services
application roles DBs. The Config DB can share a server Crawl DB
topology model indicates which operating system a virtual topology is designed for.
with the content DBs.
Database Server All SharePoint Databases
Query processing SQL Server installed and configured to support SQL clustering, mirroring, or AlwaysOn
All Databases components and index SQL Server installed and configured to support SQL clustering, across both of the hosts.
components mirroring, or AlwaysOn. AlwaysOn requires SQL Server 2012.
Load Balancer
starting point. In large environments, the specific groups that evolve for a farm depend
on the specific demands for each service. Dedicated The number of users will affect the requirement for web
Physical Host A Physical Host B Web servers web server servers. Factor 10,000 users per web server as a starting
Up to four VMs can be combined onto one Note: Server groups is a planning concept. This term and concept is not found in Central for crawling point. Adjust the number based on how heavily the servers
Web servers for all incoming Dedicated web server(s) for
Web Office Web Office physical host if the host has sufficient CPU are utilized. Heavy use of client services will increase the
Server Web Server Web cores and RAM.
Administration. load on web servers. Physical Host 3 Physical Host 4
requests crawling and administration
Apps Apps Query processing
Server Server
Dedicated application servers for the Application Server - Index Application Server - Index
Scaling out search
components and
distributed cache service are recommended index components Start with two application servers dedicated to the query Application Application server group 1 Application server group 2 Application server group 3 Application server group 4
Application Server – Application Server – in environments that use social features, such processing component and index partitions and place all other Servers
Distribution Cache Service Distribution Service as social feeds and microblogging. service application components on a separate application
As illustrated, the farm on the right is designed to crawl 10 million items (~800 GB of All other application server server. Based on utilization, consider either adding all-purpose Replica Index Partition 0 Replica
Combining all application server roles onto components and services application servers that are configured similarly, or adding
one VM requires Windows Server 2012 which data). The correlation between items and volume of data will vary depending on the application servers to dedicate resources to specific service
Application Server - Index Application Server - Index
Application Server – All Application Server – All allows a greater amount of resources to be types of data that are crawled. It is important to understand the characteristics of the applications. For example, if performance data indicates that
Excel Services is using a disproportionate amount of Host B
Application Server Roles Application Server Roles allocated to individual VMs. Query and index servers All other search components All other services (use these servers Servers for running sandboxed code
data within the environment. Above 40 million items, consider a dedicated search farm. resources, offload this service to a dedicated server. for the Central Admin site)
Database
Replica 1a Index Partition 0 Replica 1b
The following table provides starting-point numbers for search components based on Servers Replica Index Partition 1 Replica
When scaling at the database layer, start separating
Crawl DB Crawl DB
number of items. databases by role. First, separate the content databases Database Database group 1 Database group 2 Database group 3
from the rest of the databases (as illustrated). Add database When scaling out search, typically one index partition is spread across two servers or Servers
Content databases All other SharePoint
servers based on the volume of content in your environment VMs. In this configuration, a server or VM hosts only one index replica. Index replicas
For more information and examples on scaling search farms, see the following model: databases and sizing targets for your organization. for the same partition must run on separate physical hosts (whether virtualized or not)
Example Search Architectures for Enterprise Search. to achieve fault tolerance. In SharePoint 2013, more index components are required
Physical Host C Physical Host D than query processing components.
All SharePoint Databases All SharePoint Databases Content databases and configuration Search databases All other SharePoint databases
database
Crawl DB
Starting-point numbers for search components based on number of items* Storage groups
SQL Server installed and configured to support SQL clustering, mirroring, or AlwaysOn Storage groups is a concept in which similar types of databases are grouped together and scaled out independent of the rest
across both of the hosts. NUMBER INDEX QUERY PROCESSING CONTENT ANALYTICS CRAWLERS CRAWL LINK ANALYTICS REPORTING SEARCH ADMNISTRATION of the databases based on need. All databases within a storage group are treated the same with backup procedures and
restore protocols. The best practice is to include the configuration database with the content database group.
OF ITEMS COMPONENTS COMPONENTS PROCCESING PROCCESING DATABASES DATABASE DATABASE COMPONENT
AND PARTITIONS COMPONENTS COMPONENTS
General Add 1 index Use 2 query processing Add 1 crawl Add 1 crawl Add one analytics reporting database for Use 2 search administration components for
Guidance partition per 10 components for redundancy. database per 20 database per 60 each 500K unique items viewed each day redundancy, for all farm sizes
million items Above 80 million items, increase million items million items or every 10-20M total items
to 4.
* This guidance is intended for enterprise search within an organization. Guidance for Internet sites is provided in a different document.
© 2013 Microsoft Corporation. All rights reserved. To send feedback about this documentation, please write to us at ITSPdocs@microsoft.com.