Yesterday we had our last class on our subject in Deployment of Free Software, and this class was about how to create an high availability cluster using free software.
I’m going to talk about the elements and terms that are used in a cluster of this type:
- Active/active: this clustering mode is using when all nodes are active at the same time separating the services which offer.
- Active/passive: this clustering mode is using, for example with two nodes, when one of them is active and runs all services, but when this downs, then the other change to active mode and try to put all services online but now in this another node.
- VirtualBox: this is necessary for the virtualization, in the class example we haven’t used different machines, so we have used VirtualBox for simulate the different machines. But in a real environment VirtualBox is also used many times, is very important in every node to separate different services that run on every node, ie: we can having different services as MySQL, Postgree, Apache, etc. and is good custom to separate this services in different virtual machines. This allow us to separate critical services, better management of the memory and storage, and other improvements. Once I mounted a simple cluster in my computer to serve different Apache Tomcat in the same machine, but I didn’t use VirtualBox, so I mounted different Apache Tomcat in the same machine, in the same operating system, and this is a bad way to mount this, because it can’t separate the system resources.
- FileSystem: this is a question than before I always asked me, how I can serve the same files on different machines? of course I knew that a file system is mounted in different machines through the net, but I didn’t know about the database, I thought that the database was in the database server, so the replication of the database would be completely made regarding a database machine, but no, the database management different that the database storage, and the storage can be only a set of HD on raid (for high availability) and the database sgbd use the data through the net, so in conclusion is different a database server that the database storage. As I’ve said, very interesting.
- FreeNAS: is the encharge to manage the storage, as the Wikipedia says, is a free network-attached storage.
- Corosync: The Corosync cluster engine is an open source which mission is to develop, release, and support a community de-facto open source cluster executive for use by multiple open source and commercial cluster projects or products.
- Heartbeat: this is a term using as the pulse which is sending from a node to another node. This pulse is using to check if the other node is alive, the nodes check another nodes state sending this signal. If a node detects that the other is down, then it try to switch on the services (well this is for the active/passive mode).
- Pacemaker: is an open source, high availability resource manager suitable for both small and large clusters.
- Resource Agent: A resource agent is a standardized interface for a cluster resource. In translates a standard set of operations into steps specific to the resource or application, and interprets their results as success or failure.
- Single point of failure (SPOF): is a part of a system that, if it fails, will stop the entire system from working.
Telling also that this model simulated in class is based in the Linux-HA model that is a model to build HA Clusters based on linux and other platforms.
For more info see this links:
Resource agent: http://www.linux-ha.org/wiki/Resource_Agents
Single point of failure: http://en.wikipedia.org/wiki/Single_point_of_failure
And that’s all, see you my friends!!!