HA-OSCAR: the Birth of Highly Available OSCAR

As clusters reach thousands of nodes, eliminating single points of failure becomes critical. The Open Cluster Group's HA-OSCAR is a solution.
Dual Master Nodes and Redundancy

A typical cluster computing architecture consists of several nodes that can provide some degree of availability. However, it normally has a single-head node that is a simplex architecture and prone to single points of failure. The current release of OSCAR falls into this architectural category, which is unsuitable for mission-critical systems as it contains several individual system elements that have no redundancy for a backup or failover. In order to support HA requirements, clustered systems must provide ways to eliminate single points of failure.

Hardware duplication and network redundancy are common techniques utilized for improving the reliability and availability of computer systems. To build an HA-OSCAR cluster system, we first must provide a duplication of the cluster head node. Such an architecture can be implemented in different ways, including active-active, active-warm standby and active-cold standby.

The active-active model enables both performance and availability, because both head nodes simultaneously can provide services. However, its implementation is quite complicated and leads to data inconsistency when failures occur. Active-standby options mostly are adopted solutions. The standby server watches the primary server health and can take over control when it detects an outage. Currently, the active-warm standby configuration is the initial model of choice.

Figure 1. HA-OSCAR Cluster System Architecture

Figure 1 shows the HA-OSCAR cluster system architecture. We experimented with and planned to incorporate Linux Virtual Server and Heartbeat mechanisms into our initial active-hot standby HA-OSCAR distribution. Now, we plan to extend our initial architecture to support active-active HA after we release the hot-standby distribution. The active-active architecture can better utilize resources, because both head nodes can be simultaneously active to provide services. The dual master nodes then can run redundant DHCP, NTP, TFTP, NFS and SNMP servers. In the event of a head node outage, all functions provided by that node failover to the second redundant head node and are served at a reduced performance rate (in theory, 50% at the peak or busy hours).

Another HA functionality to support in HA-OSCAR is providing a high-availability network using redundant Ethernet ports on every machine. In addition, duplicate switching fabrics (network switches, cables, etc.) are used for the entire network configuration. This enables every node in the cluster to be present on two or more data paths within its networks. Backed with this Ethernet redundancy, the cluster achieves higher network availability. Furthermore, when both networks are up, improved communication performance may be achieved by using techniques such as channel bonding of messages across the redundant communication paths.

HA-OSCAR aims to reuse features from other implementations and existing projects, including the High-Availability Linux, Kimberlite and Linux Virtual Server projects. We then plan to contribute the added enhancements and functionalities back to the community.

Support for Dual IP Stack (IPv4 and IPv6)

IPv6 is the next-generation protocol designed by IETF to replace the current version of the Internet Protocol, IPv4. Most of today's Internet uses IPv4, which has been remarkably resilient in spite of its age, but it is beginning to have problems. Most importantly, there is a growing shortage of IPv4 addresses, which are needed by all new devices connecting to the Internet. As a result, IETF defined IPv6 to fix the problems in IPv4 and to add many improvements for the future Internet. These improvements come in different areas, such as routing, autoconfiguration, security, QoS and mobility.

HA-OSCAR has support for IPv6 activated by default. Most of the ISPs and telecom companies already are experimenting with co-existence schemes for IPv4 and IPv6. All cluster nodes installed with HA-OSCAR provide support for IPv6 and basic IPv6 capabilities compiled directly in the network utilities and binaries.

Recovering from Corrupted Disks

OSCAR assumes the client node disks on which it is installing are faultless. But, this is not always the case; some nodes may have corrupted disks. HA-OSCAR considers this issue and does not assume that all disks on all nodes are a good installation base. To this end, we support special scripts in our installs and software RAID in the kernel, in parallel with developing the necessary set of scripts needed to synchronize disk contents. As such, if a disk fails, data is not lost. In addition, our installation wizard first tries to fix the corrupted disk. HA-OSCAR also supports synchronous operation, disk removal and disk insertion. In addition, HA-OSCAR supports software RAID by default. By enabling software RAID, clusters powered by HA-OSCAR have increased data redundancy and better performance.

______________________

Webinar
One Click, Universal Protection: Implementing Centralized Security Policies on Linux Systems

As Linux continues to play an ever increasing role in corporate data centers and institutions, ensuring the integrity and protection of these systems must be a priority. With 60% of the world's websites and an increasing share of organization's mission-critical workloads running on Linux, failing to stop malware and other advanced threats on Linux can increasingly impact an organization's reputation and bottom line.

Learn More

Sponsored by Bit9

Webinar
Linux Backup and Recovery Webinar

Most companies incorporate backup procedures for critical data, which can be restored quickly if a loss occurs. However, fewer companies are prepared for catastrophic system failures, in which they lose all data, the entire operating system, applications, settings, patches and more, reducing their system(s) to “bare metal.” After all, before data can be restored to a system, there must be a system to restore it to.

In this one hour webinar, learn how to enhance your existing backup strategies for better disaster recovery preparedness using Storix System Backup Administrator (SBAdmin), a highly flexible bare-metal recovery solution for UNIX and Linux systems.

Learn More

Sponsored by Storix