Clustering Management Tools Support Growing Applications.

Over the past several years, clusters of workstations and persona computers have become increasingly popular. The incremental scalability of processors, memories, and mass storage systems, together with the high-performance interconnection networks, have made clusters a cost-effective platform for distributed and parallel computing.

A cluster is a group of systems linked by a high-speed interconnect and functioning as a single resource. It differs from multiprocessing in that a multiprocessing machine runs a single copy of the operating system. Clustering, on the other hand, connects several systems, each running its own copy of the operating system.

Cluster interconnect is slower than multiprocessing. To try to alleviate bottlenecks in the interconnect, some try to give all machines in the cluster direct access to a shared resource such as a disk. This eliminates the need for one server to go through another server to access a disk resource and also reduces the amount of data going over a shared path (bus contention).

Clustering is used for several reasons: availability, scalability, and unifying management. Any given cluster technology may offer one or more of these advantages in any combination, but not all cluster technologies offer all three advantages.

Clustering is also a key strategy for departmental computing, putting it as a solution of choice for such applications as e-commerce and Internet-related uptime. Software management options have been growing for this space and two major software developers recently announced new products in this space.

The rapid acceptance of e-commerce is increasing the level of availability required for Internet infrastructures to obtain optimal uptime. For an e-commerce site, downtime means lost revenue, reduced customer loyalty, and can affect shareholder value. Traditionally, proprietary fault-tolerant systems have been used to implement optimal uptime solutions, but rapidly growing and more open Internet infrastructures are being deployed on clustered Unix and Windows/NT servers.

VERITAS Software Corporation, one of the industry's leading enterprise-class application storage management software providers, announced VERITAS Global Cluster Manager, which provides monitoring and fail-over of clusters distributed around the world. Managing a virtually unlimited number of VERITAS Cluster Server installations, VERITAS Global Cluster Manager is a unique solution for customers building optimal uptime e-commerce platforms, implementing disaster recovery processes, or requiring monitoring of all mission critical systems from a single operations center.

VERITAS has not been slow about partnering with major hardware vendors in implementing cluster management solutions. Although EMC Corporation considers itself more and more a software business, they have been active participants with VERITAS in e-business applications.

"EMC and VERITAS Software have many joint customers utilizing EMC Symmetrix Remote Data Facility (SRDF) and VERITAS Cluster Server to maximize their application uptime," according to Linda Wright, director of the E-Infostructure Developers Program at EMC. "As part of our program, which integrates new applications with EMC's open Enterprise Storage solutions, we are building links for SRDF business continuance software to integrate with VERITAS Global Cluster Manager. By doing so, we provide our mutual customers with a tool that allows them to efficiently manage and control availability of critical applications."

There is no doubt that the e-applications are growing in importance. This means clustering is no longer a data center or large corporate preserve.

"We find that more and more organizations, even those outside the e-commerce space, are discovering a need to architect global or intercontinental disaster recovery solutions to protect their most-critical applications," said Harvey Hindin, vice president, High Availability and Clusters Service, DH Brown Associates.

Significant Downtime Losses

To achieve this near-continuous level of availability, administrators must build on their single site high clustering solutions by replicating critical data to one or more geographically remote sites. In the event of a major failure or disaster at a primary site, the entire cluster can, then, be failed-over to one of the redundant remote sites. This capability can also be used to implement a "follow-the-sun" operation, in which a cluster is intentionally failed-over to a location closest to the peak network demand, optimizing customer service.

A typical e-commerce site can lose over $100,000 per hour due to downtime, according to Forrester Research, and some observers suggest that most sites achieve only 99.4% uptime, leaving them vulnerable to competitors and lost revenue opportunities in some of these skyrocketing markets. The most successful Internet retailing sites must strive to achieve increased uptime through geographic distribution of services running on modular, open-systems building blocks. "We have a large installed base of current VERITAS Cluster Server customers who have told us they need to take high availability to the next level for Internet applications," said Fred van den Bosch, executive vice president of engineering at VERITAS.

The new VERITAS Global Cluster Manager is a cluster management product for VERITAS Cluster Server, the company's flagship, and heterogeneous high-availability product for Unix and NT applications. Using a single web browser console, a cluster administrator may monitor a virtually unlimited number of VERITAS Cluster Server clusters anywhere in the world both within and across firewalls. In addition to supporting fail-over of a VERITAS Cluster Server cluster from one site to another, VERITAS Global Cluster Manager is scheduled to offer the flexibility of integrating with other VERITAS software for host-based data replication or major third-party replication products.

Lest We Forget Unnx

"Linux has attracted quite a bit of early attention because of its development model, its licensing model, its capability, and [its] single machine reliability," said Dan Kusnetzky, vice president of system software research for International Data Corporation. "Although Linux has a wealth of load balancing and computer farm software, high availability or clustering software has been missing."

Legato Systems is addressing this gap in the growing Linux software surge with its new eCluster solution. Legato Cluster Enterprise and this eCluster product are the first heterogeneous, high-availability software solutions to support the Linux operating system. The two clustering applications immediately provide support for Caldera OpenLinux eServer 2.3 and Red Hat Linux 6.1. Legato's solutions represent a recognition of Linux, providing a high suite of benefits required for Linux to establish itself within enterprise business-critical data centers and e-commerce deployments.

Legato also announced that its Legato Cluster Enterprise and Legato eCluster have been tested on IBM's Netfinity Linux servers and are IBM "Server Proven" solutions. Legato's availability solutions are the first clustering solutions of any type on Linux to attain this important benchmark on IBM Netfinity servers.

"Legato Cluster Enterprise provides exceptional capabilities for customers of IBM Netfinity for Linux," stated Sandy Carter, director of Partnerworld for Developers at IBM. "This will enable customers to deploy Netfinity running Linux within enterprise data centers, as well as e-commerce solutions. IBM is pleased that Legato has selected Netfinity as its premier Linux solution platform and has announced Netfinity Server Proven solutions."

Legato Cluster Enterprise is designed to protect applications running in back office environments, such as ERP, RDB and messaging solutions. Legato eCluster is designed to protect applications and overall load capacity within Web farm environments running Cisco Local Director or other types of Web load balance solutions. To enable this, Legato eCluster includes a Cluster Toolkit for Apache, a component providing fine-tuned protection for Apache Web services on Linux Web servers.

Any of the Legato clustering solutions running on Linux servers can be used on their own, or as part of a larger cluster configuration that includes Windows NT/2000 or Unix servers. All of the servers running Legato eCluster or Legato Cluster Enterprise are centrally manageable from a single GUI console, which can run on Linux, Windows NT/2000 or Unix platforms to manage and control the entire mixed-platform configuration. Caldera Systems is one of the leading Linux software developers and is committed to an e-server strategy.

Product suites like Legato's "advance our OpenLinux eServer strategy which is to provide ready-to-deploy server class solutions," said Benoy Tamang, vice president of marketing for Caldera. "Our customers benefit immediately from the mature availability capabilities that Legato is providing for Web environments with products like Legato eCluster, as well as data center deployments."

"At Red Hat, we know that high availability is a key component to enabling Linux deployments within enterprise data centers and e-commerce configurations," stated Paul McNamara, senior director of business development for Red Hat, another major name in Linux software development. "We support Legato because Legato Cluster Enterprise and Legato eCluster are designed to meet the requirements of Red Hat customers in both of these critical environments."

Legato Cluster solutions extend the Linux free-code philosophy to high availability by providing a Perl scripting environment that independent programmers can use to create high availability solutions for specific applications and deployments. Legato Cluster Enterprise provides the backbone capabilities for defining and managing resources. Programmers can use Perl scripting rules that determine how those resources are managed and in response to which sensed events. Legato plans to encourage independent programmers to share their code along with tips and tricks for setting up high-availability configurations with Legato Cluster solutions on Linux through Cluster-for-Linux user groups and online forums.

Nora Denzel, executive vice president of Legato and a well-known name in mass storage software, observed, "We are also pleased to embrace the open-code philosophy of Linux by providing our Perl scripting environment for high availability. This is a strong indication of Legato's commitment to the Linux platform and to the community of Linux programmers that have made Linux the fastest maturing platform in history."