Scaling SharePoint Topologies Session

Hosting_SharePointProducts2010

Simon Skaria gave a great session. I was very impressed with the architectural thought process and approach to designing SharePoint sites. This is a bit different from a typical Microsoft session, very refreshing. 

The session was great for intranets but didn’t cover internets at all. I would assume that these are similar but there are other considerations to support the usage profiles and security that is attributed to an internet type of site. There is follow-up topology information for SP2010 on the SharePoint Server 2010 TechNet site located here:

Download the model

This model helps you understand hosting environments and provides an overview of the following:

  • Hosting features and concepts such as multitenancy, site subscriptions, and service partitioning
  • Enterprise hosting environments
  • Shared hosting environments

 

Additionally the hardware requirements can be found here:

http://technet.microsoft.com/en-us/library/cc262485(office.14).aspx

 

Key Themes

  • Patterns can be used to accelerate deployments
  • Improved Security Model
    • Claims based authorization
    • Cross farm communication
  • Administration simplified
  • Service Isolation
  • Each app can use separate database and optionally separate app pool
  • Support mult applications with different accounts and databases
  • Multitenency support
  • Some service apps can be partitioned to handle multi-tenants

 

General Considerations

  1. Start with logical architecture
    1. Consolidated vs. Distributed
  2. Logical Topology Considerations
  3. Business Needs, Regulatory Impacts, IT considerations (information architecture)
  4. Physical Considerations
  5. Scale up vs. Scale Out
  6. Link Latency – If you are deploying over a WAN and how to handle this.
  7. Directory Architecture – Host services relationship with AD and how to handle sync
  8. Separate meta data repositories based on the need in the organization. Example HR could create tags that manages personnel. For example, pay grade or performance ratings. This metadata wouldn’t go outside the HR department.
  9. MOSS 2007 is not interoperable with 2010 farms
  10. Microsoft’s suggestions on upgrading
    1. Upgrade All – This seems a bit risky, especially if you have mission critical solutions on SharePoint. Additionally it could be time intensive.
    2. Phased Upgrade – This is what Microsoft is doing internally. They do weekend upgrades till they get all upgrade.

 

Scaling Services – General recommendations are to scale within the farm. Identify where there are bottlenecks are put them on their own dedicated server.

    1. Search and user profiles are SQL intensive and can be a bottle neck.
    2. Usage logging service (new) you want to put that on a separate server
    3. Scale out on each tier.
    4. Add Web front Ends (WFE) for content servers if additional capacity is needed
    5. Add apps servers for LOB centric / transactional needs.
    6. Adding multiple content farms will help for organizations that deal with a lot of content to scale
    7. Multiple service farms are used in international WAN based solutions.

 

Scenarios where given for implementation profiles of small to mid to large size companies. These profiles are great starting points but I see that there are a lot more patterns that will be more applicable. For example, the mid sized profile doesn’t account for Disaster Recovery or High Availability. These two non-functional requirements do not hinge on the number of users. 

The small company profile was fairly simplistic with WFE and app server combined with dedicated SQL Server.

For mid sized profile < 50k users.

  • Single farm
  • Isolated web apps
  • Multiple services apps (metadata)
  • Multiple proxy groups
  • WFE and Application Server is separated
  • SQL on separate server

 

Large Enterprise profile is >50k users that may be international distribution and leveraging most if not all advanced features of the SharePoint platform.

  • Multiple farms that could be distributed across datacenters.
  • Centralized IT services that manages the growth through governance and policy
  • Disaster Recovery support and High availability. These two where not talked about in length.

 

For more resources, take a look at:

SharePoint 2010 Website – to view SharePoint 2010 in action

SharePoint 2010 forum– for SharePoint 2010 questions

SharePoint 2010 PressPass– for the SPC 2009 keynote video, a Q&A with Jeff Teper, and more

SharePoint 2010 Developer Center – for developer info

http://www.mssharepointitpro.com – for IT Pro info

http://www.microsoft.com/sharepoint – for more SharePoint information

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s