VMS Clusters

In 1983, Digital announced VAXclusters to answer the growth of interactive users.  The initial VAX cluster designs grouped VAX computers together in a loosely coupled system allowing them to operate as a single system, extending the characteristics of VAX to high capacity and high availability.

In 1984, with the release of VAX/VMS v4.0 Digital added support for clustering using Digital’s proprietary Computer Interconnect (CI) technology.  This technology used a star coupler at the center of each VMS cluster that connected to every node (computer) and data storage device by using one or two pairs of CI cables.  A pair of cables would transmit at a rate of 70 megabits while two pairs would give an aggregate transmission rate of 140 megabits per second.  Redundancy would be achieved when using two pairs of cables in case one of the cables failed to operate in accordance to the cluster.  For better availability the star couplers had redundant wiring and acted as a passive device requiring no electricity when in use.

In 1986,  Digital added support for ethernet clustering.  This was popular with customers who couldn’t afford the high-availability advantages of the CI hardware.  This option also provided the ability for a diskless satellite node to boot over the network from a shared boot device.  Terminal servers using the LAT protocol over ethernet allowed ordinary serial terminals to access the host nodes making it easier to accomplish fail-over from one node to another.

In 1988, Digital supported “mixed interconnect” clusters using both CI and Ethernet along with other clustering options for DSSI (Digital Systems and Storage Interconnect), SCSI (Small Computer System Interface) and FDDI (Fiber Distributed Data Interface).  Support was extended with Clustering for satellite data links and long-distance terrestrial links when high-bandwidth wide area networking became available.  Disaster-tolerant clusters were created by using a single cluster spread across several diverse geographical areas, that could survive infrastructure failures and natural disasters.  

In 2010, HP with the use of Gigabit Ethernet networking technologies added support for OpenVMS clustering over TCP/IP using standard networking cables and cards.