The Arellia Management Server allows organisations to manage all of their endpoints from a single web console.
In order to facilitate this centralised management AMS has been designed with High Availability and Load Balancing features, so that there is no single point of failure within the product.
There are two components that make up the AMS:
Web front end
Using Microsoft Internet Information Services the web front end provides the communication channel between the server and clients and the web console that allows users to administer the system over HTTP and/or HTTPS.
A single AMS installation can have one more more web front end servers. The benefits of having multiple front ends are:
- Redundancy - in case of a server failure the other servers within the cluster will seamlessly take over without any interruption to clients
- Load Balancing - during normal operation the client traffic is distributed across all nodes in the cluster
Client server communication with the AMS platform is sessionless, thereby removing the need for client session affinity which greatly reduces the complexity of setting up load balancing / fail over.
Network traffic can be distributed between the nodes by using a variety of methods. Some example scenarios are:
- DNS round robin - very simple to setup and requires minimal configuration, load is balanced evenly between cluster nodes. Requires manually removing a failed node from the cluster.
- Network Load Balancing - a component of Windows Server operating systems that allows multiple IIS servers to appear as one. Using NLB provides a greater control over how the load can be balanced and provides for detection of failed nodes which can be automatically removed from the cluster.
- Hardware based appliance - provides highest level of reliability but requires more work setting up initial configuration.
AMS provides REST API's that can be utilised to determine a nodes health so that automatic fail over is possible.
The client/server communication channel is encrypted using SSL which will require that all nodes in the web front end cluster have a SSL certificate installed. Nodes within the cluster do not need to share the same certificate, and optionally the load balancer can terminate the SSL connections on its external interface and communicate with the web front end nodes using normal HTTP thereby freeing up CPU resources.
SQL database back end
AMS supports the use of SQL server clusters for High Availability scenarios. We support all cluster types including stretch clusters, however the latency between the web front end and each SQL cluster node must be no greater than 30 ms. AMS also supports SQL Server fail over cluster configurations. Note that an active/active configuration does not provide improved performance, just high availability.
Some customers have chosen to deploy a single standalone instance for production with database mirroring to a second standalone DR instance that can also be used for reporting. This type of configuration requires manually cutting over in a failure scenario but allows the second SQL server to be actively used instead of sitting underutilised.