- Operating mode of an HFSQL cluster
- Benefits of an HFSQL Cluster in relation to a stand-alone HFSQL server
Overview of HFSQL Cluster
Available only with this kind of connection
HFSQL Cluster is a database server intended to be deployed on several servers. A database cluster favorably replaces a stand-alone server when the availability is an essential criterion.
For more details, see:
A specific setup package is required to use HFSQL Cluster.
This pack can be downloaded at no extra cost on PC SOFT site, in the download section of modules common to version 27.
Operating mode of an HFSQL cluster
The diagram below presents the deployment of an HFSQL cluster that is using 4 computers: a coordinator and three nodes.
The following elements are represented in this diagram:
|Cluster||A cluster is kind of a super database server.|
Internally, it includes a set of nodes and a coordinator.
From the client side, it operates like a stand-alone database server.
|Nodes||A node is one of the database servers that constitutes the cluster.|
Each node contains a copy of the databases hosted by the cluster. All the nodes strictly contain the same data and they are replicated among themselves in synchronous mode.
|Coordinator||The coordinator is the server used to manage the cluster via the ClusterManager service.|
It is used when starting and stopping the cluster.
It is also used when one or more nodes fail in order to regulate the operating mode of the cluster.
|Client||A client is an application (WINDEV or WINDEV Mobile) or a WEBDEV site accessing the data hosted by the cluster.|
|DNS||The DNS server is used to orientate the clients among the different nodes of the cluster.|
The architecture for storing the HFSQL Cluster data is used to guarantee a high availability of the service. It is based on the following principles:
- The storage structures are not shared (in most cases, this mechanism is called "shared-nothing"): the content of the databases is entirely replicated on each node of the cluster.
- Synchronous replication of the nodes: the data is continuously updated on the entire cluster.
- No Single Point of Failure (SPoF): no system element compromises the operating mode of the cluster.
The standard features of HFSQL Client/Server are still available. These features already guarantee on the stand-alone servers a high availability and low maintenance and ownership costs (TCO):
- Automatically recalculates the statistics on live databases.
- Live modification of data structures.
- Automatic reconnection of the client applications in case of network outage.
Benefits of an HFSQL Cluster in relation to a stand-alone HFSQL server
The main benefits of an HFSQL Cluster in relation to a stand-alone server are as follows:
- High availability
As the data is not located on a single physical server but on a cluster of computers, the database is more insulated from technical problems. If one of the cluster computers fails or if its network connection is interrupted, the clients that were connected to it are automatically re-routed to another node of the cluster: an uninterrupted service is guaranteed.
Use case: databases used by applications that operate on a permanent basis (production lines, sites accessible in the entire world) or for which the interruption of services are costly.
- Load balancing
When the number of clients increases, it can be difficult to make a single physical server evolve to comply with the request. In case of a cluster, the client connections are distributed among the different nodes. Therefore:
- each cluster node can be hosted by a less powerful server.
- new nodes can be easily added when the number of clients increases.
Use case: the databases whose number of clients can be very high (Web sites for example) or whose number of clients can evolve very quickly in time.
- Automatic resynchronization of failing nodes
If an incident occurs on one of the cluster computers, the failing node is automatically excluded from the cluster. When the problem is corrected (computer fixed or changed, network link restored, ...), all you have to do is restart the computer. HFSQL Cluster will automatically re-associate the node with the cluster and re-synchronize its content. After the re-synchronization, the node will become accessible to the all cluster clients.
- Dynamic modification of the cluster structure
The nodes of a cluster can be stopped or restarted at any time from the management interface (the HFSQL Control Center). This allows for a progressive evolution of the cluster configuration by replacing the nodes one by one without any service interruption.
This page is also available for…