Welcome on my SharePoint 2010 blog

Hello,

My name's Roy. I'm a dutch SharePoint & BI consultant/architect at Advantive B.V. At this moment I'm busy with some big SharePoint 2010 projects in The Netherlands. In all the projects I've got different roles, like: Business consultancy, Lead Consultant, Architect (logical and technical), Development and Teaching/courses.

Products where you can ask me about are: SharePoint, Visual Studio, SQL Server, PowerPivot, Analysis and Reporting Services, Visio Services, InfoPath, PerformancePoint Services, Team Foundation Server, Office line.

I love to work and to write about Microsoft SharePoint 2010 so, feel free and read/comment my Blogs!

Greetz.

Monday, October 11, 2010

Failover principal SP2010 service applications

I hope I will help a lot of people with this new blog about the OOB failover principal that comes with SharePoint 2010. Please read the following:

You can find a lot of information about the failover principal for OOB service applications, things like: "It is based on Round Robin...", "if one service app is down, the first available one will be used..." and  "This whole principle is embedded in the application discovery and load balancer service application....".



All of the above statements are true, BUT....How does it really work under the hood....That answer you can find from now on in this blog and the answer is pretty simple!!

There is a timer job OOB within the central administration that checks which services on the specific appservers are up and running (application discovery). If so, it cleans and re-creates the connections/addresses to the working services.

SharePoint will directly act when a service that represents a specific service application is down. Example:

2 servers that have an active metadata service. The services are connected to the metadata service application. When the service on the first service (order of the services for the SP NLB) goes down (or both services go down), you cannot use the metadata tree throughout your SharePoint UI. The load balancer service application will not directly change the address to the other server when, because this service application does'nt discover immediatly. This will be done by the timer job that will run every 15 minutes (by default), it is called: Application Addresses Refresh Job

So, when one or all of the services for a specific service application are down, you can see it immediately throughout the UI of SharePoint. But the timerjob will discover the services again to reconnect to the first available one. Of course, because it's just a normal timer job, you can reschedule it.

No comments:

Post a Comment