Novell Inc used its technical and user conference in Utah this week to outline a roadmap for the roll-out of its next generation clustering services, code-named Orion II. The Provo, Utah based networking giant said the technology, which will allow up to 12 or more Netware 5.0 servers to be linked together, will be available in the second half of this year, although it will be renamed Novell Cluster Services when it ships. Novell currently has two products in the clustering space, both of which are co- developed with Vinca Corp.
The first, called Stand-By enables a single Netware 4.0 server, or multiple servers (up to four or five) to be connected to a passive failover server, which takes over should the Netware server(s) go down. The second, called High Availability Server, enables two active Netware servers to be connected, which means either server can take over the other one’s functions should a failure occur. Michael Bryant, Novell’s director of product marketing for high availability and clustering solutions said Orion II was the next logical evolution to the networking giant’s clustering line-up.
At this week’s Brainshare show in Utah, both Compaq and Dell showed 12-node, configurations based on the software but the final version will likely scale to even larger numbers, said Bryant, it’s just a question of what hardware there will be to support it. Apart from the ability to scale up to and beyond 12 nodes, Bryant said that the other main difference in Orion II is tighter integration with Novell’s directory services, NDS. The integration will enable administrators to see the Netware cluster as an object on the NDS tree. It will let them click on individual servers and see what resources are assigned to each server, Bryant said. But one thing the software won’t do is enable automatic load balancing. That feature will still have to be carried out manually, through NDS, although Bryant added that Novell will incorporate that feature in a future version of its cluster services software.
The integration with NDS will also enable companies to plan downtime better by manually shifting resources to other servers in the cluster while updating software in others, and so on, he said. Other areas of improvement for future cluster services releases include support for the Microsoft-Intel-Compaq Virtual Interface architecture, for high speed, low latency interconnects between servers, improved manageability and salability, mirroring of data in memory and the ability to handle distributed Java applications. Bryant said he envisaged the majority of early cluster services users to link four or six nodes together, customers who are doing it for server consolidation, who want to centralize their data on a small number of highly reliable servers. But ultimately, he said the software was designed as a front-end for users that want to access data held in a fibre channel based storage area network (SAN) back end environment. á
