Aspera Professional Services typically handles the installation of Aspera software and
tools for high availability (HA) systems. However, it is the customer’s responsibility to
provide and configure infrastructure hardware and the required operating systems. In
particular, the shared storage required for HA systems must be configured prior to Aspera
personnel installing the Aspera products.
The systems requirements for HA installation of Orchestrator on Linux are the following:
- Hardware: Normal HA operation requires two servers. Virtual machines can be
used as long as enough resources are allocated to each of them.
- Operating system:
Note: Red Hat high-availability packages (such as ricci, luci,
rgmanager, and cman) are not used, and therefore must not
be installed or activated in the environment.
See also, List of System Commands Used by Aspera Cluster Manager (ACM).
- Software:
- Aspera Orchestrator
- Aspera Cluster Manager (ACM) for Orchestrator (the latest version, provided by
Aspera Professional Services)
- Time synchronization:The system clocks for all nodes in the HA cluster
(including the shared storage) must always be kept in sync in order for ACM to operate
correctly. Aspera recommends using the ntpd daemon (for the NTP
(Network Time Protocol)), but any time-synchronization mechanism is acceptable.
- Database: Typically, the MySQL database runs on each Aspera Orchestrator node
across shared MySQL data. This MySQL database is delivered with the Aspera Common
Components and is supported by the Aspera Support team. An external MySQL database with
an high availability solution of its own is also a possible solution, although IBM
Aspera will not support that option.
Note: The Oracle MySQL cluster solution comes with
the NDB engine; this will not interoperate with Orchestrator because only MySQL with
the InnoDB engine is supported.
- Load balancer: A load balancer that implements a Virtual IP address (VIP) is
required. For more information about what is expected from the load balancer, see Load Balancer Behavior.
- Single point of failure: Aspera strongly recommends that customers identify
single points of failure (SPOF) in the environment and recognize the source of these
risks. Often, these are situations where all nodes are plugged into the same power
source. Another possible SPOF is shared storage or a load balancer that is not also
HA.