I came across implementation with appfabric using region based caching and had issues with application crashing when cache cluster is down or any of the cache host is down.
As it was my first experience with Appfabric, I started understanding the basics of caching and how clusters are setup.
There are two main type of cache implementation possible as following:-
1) Named caching
2) Region based caching
Region based caching is used when availability is not a concern and search need to be faster. For making search faster region based cache is associated with only one host.
On the other hand named caching is used to spread across available cache hosts in the cluster. There is a system defined region created for this as well but it's completely controlled by appfabric out of the box.
If we want to make the appfabric highly available, then it can be done only with named caching. We can create the named cache with number secondary nodes where the data can be replicated.
One more consideration you need to take while defining appfabric cluster for their application. The deployment topology should be defined keeping the application usage and server availability. There are two type of deployment appfabric offer i.e. separate cluster server for appfabric and appfabric installed on application server along with application.
There are pros and cons to both model.
There is a very useful blog I found which explains about the implementation of cache
http://cgeers.com/2010/07/04/windows-server-appfabric-caching/
As it was my first experience with Appfabric, I started understanding the basics of caching and how clusters are setup.
There are two main type of cache implementation possible as following:-
1) Named caching
2) Region based caching
Region based caching is used when availability is not a concern and search need to be faster. For making search faster region based cache is associated with only one host.
On the other hand named caching is used to spread across available cache hosts in the cluster. There is a system defined region created for this as well but it's completely controlled by appfabric out of the box.
If we want to make the appfabric highly available, then it can be done only with named caching. We can create the named cache with number secondary nodes where the data can be replicated.
One more consideration you need to take while defining appfabric cluster for their application. The deployment topology should be defined keeping the application usage and server availability. There are two type of deployment appfabric offer i.e. separate cluster server for appfabric and appfabric installed on application server along with application.
There are pros and cons to both model.
There is a very useful blog I found which explains about the implementation of cache
http://cgeers.com/2010/07/04/windows-server-appfabric-caching/
Comments
Post a Comment
Thanks for your valuable comment