HP OpenVMS Systems Documentation

Table 1-4 describes the optional networking software that enables OpenVMS Cluster system nodes to communicate and share resources with other OpenVMS Cluster nodes.

Optional storage enhancement software improves the performance or availability of storage subsystems.

Examples include:

• Compaq Volume Shadowing for OpenVMS (redundant arrays of independent disks [RAID] level 1)
• Compaq DECram for OpenVMS (random access memory [RAM] disk)
• StorageWorks RAID Software for OpenVMS (supports RAID level 0 arrays (disk striping) and RAID level 5 arrays (disk striping with parity)
• Compaq Hierarchical Storage Manager (HSM)

For the latest information about StorageWorks products, refer to the StorageWorks web site, which you can select from the Compaq World Wide Web site:

http://h18000.www1.hp.com/

System management software helps you manage your OpenVMS Cluster system.

Examples include:

• DIGITAL Availability Manager for Distributed Systems (DECamds) and Compaq Availability Manager
  Note: Both DECamds and Availability Manager are provided with the OpenVMS operating system.
• POLYCENTER Capacity Planner
• Compaq Archive/Backup System for OpenVMS
• Compaq Code Management System for OpenVMS
• POLYCENTER Network Manager

Business applications are optional software packages that help you perform your business function.

Examples include:

• Database systems, such as Oracle Rdb, Oracle CDD/Repository, and Sybase
• Transaction processing systems, such as Compaq ACMSxp for OpenVMS Alpha
• Transaction routing systems, such as Reliable Transaction Router

To take advantage of OpenVMS Cluster features and benefits, proper configuration is essential. An ideal OpenVMS Cluster configuration meets the following criteria:

• Provides the best combination of hardware and software components to meet your business requirements.
• Strategically distributes your budget dollars to return maximum value in the areas that are high priority for your business.
• Meets your current needs and retains your investment as your business needs grow and change.

Configuring your OpenVMS Cluster system requires careful planning because you need to consider many factors. You will probably modify plans as new factors arise. As your design evolves, you can weigh advantages against tradeoffs and make decisions that best meet your needs.

1.4.1 General Configuration Rules

The following general rules apply to OpenVMS Cluster systems:

• A node is an OpenVMS system. An OpenVMS Cluster system cannot contain more than 96 Alpha and VAX (combined total) nodes.
• An Alpha and a VAX system can not boot from the same system disk. System disks are architecture specific and can be shared only by systems of the same architecture.
  Cross-architecture satellite booting is supported. Alpha satellites (clients) can boot from a VAX boot server, and VAX satellites (clients) can boot from an Alpha boot server.
• Every OpenVMS node must be able to communicate directly with every other OpenVMS Cluster node.
  Configurations that use a shared (multihost) SCSI bus or a shared (multihost) Fibre Channel interconnect must also be configured with any of the other supported OpenVMS Cluster interconnects, because node-to-node communication does not occur across the SCSI bus or the Fibre Channel.
  Reference: See Section 4.8 for more information about the SCSI interconnect and Section 4.6 for more information about the Fibre Channel interconnect.
  Configurations that use a MEMORY CHANNEL interconnect must also be configured with any of the other supported OpenVMS Cluster interconnects for access to storage. Storage cannot be configured on MEMORY CHANNEL.
  Reference: See Section 4.7 for more information about MEMORY CHANNEL.
• An OpenVMS Cluster node, storage controller, or storage device can participate in only one OpenVMS Cluster system at a time.
• DECnet--Plus software is not required in an OpenVMS Cluster configuration. However, DECnet--Plus is necessary if internode process-to-process communication using DECnet mailboxes is needed. Starting with OpenVMS Version 7.0, the Monitor utility uses a TCP/IP or DECnet transport, as appropriate, for intracluster communication.

In addition to these general rules, more detailed guidelines apply to different configurations. The rest of this manual discusses those guidelines in the context of specific configurations.

This chapter contains information about how to determine your OpenVMS Cluster business and application requirements.

2.1 Determining Business Requirements

The kinds of business requirements that you have affect the way that you configure your OpenVMS Cluster. Typical business requirements for an OpenVMS Cluster system include:

• Budget
• Availability
• Scalability and future growth
• Physical location requirements
• Security

Some of these requirements may conflict with each other, such as scalability and physical location. For example, you may want to grow your OpenVMS Cluster, but you are limited by physical space or by the location of your systems. In situations like this, determine what your primary requirements are and where you are willing to make tradeoffs.

2.1.1 Budget

As with most business decisions, many of your choices will be determined by cost. Prioritizing your requirements can help you apply your budget resources to areas with the greatest business needs.

When determining your budget, plan for the initial system cost as well as the cost of ownership, which includes:

• Service and update
• Power consumption
• Cooling
• System management

Determine how available your computing system must be. Most organizations fall into one of the three broad (and sometimes overlapping) categories shown in Table 2-1.

Reference: For more information about availability, see Chapter 8 in this guide.

2.1.3 Scalability and Future Growth

Scalability is the ability to expand an OpenVMS Cluster in any system, storage, and interconnect dimension and at the same time fully use the initial configuration equipment. Scalability at the node level means being able to upgrade and add to your node's hardware and software. Scalability at the OpenVMS Cluster level means being able to increase the capacity of your entire OpenVMS Cluster system by adding processing power, interconnects, and storage across many nodes.

Among the low-end PCs and workstations, midrange departmental systems, and high-end data center systems offered by Digital, each level has different processing, storage, and interconnect characteristics. Investing in the appropriate level means choosing systems that meet and perhaps exceed your current business requirements with some extra capacity to spare. The extra capacity is for future growth, because designing too close to your current needs can limit or reduce the scalability of your OpenVMS Cluster.

If you design with future growth in mind, you can make the most of your initial investment, reuse original equipment, and avoid unnecessary upgrades later.

Reference: See Chapter 10 for more help with analyzing your scalability requirements.

2.1.4 Physical Location Requirements

Physical restrictions can play a key role in how you configure your OpenVMS Cluster. Designing a cluster for a small computer room or office area is quite different from designing one that will be spread throughout a building or across several miles. Power and air-conditioning requirements can also affect configuration design.

You may want to allow room for physical growth and increased power and cooling requirements when designing your cluster.

Reference: See Section 8.6 and Section 10.7.7 for information about multiple and extended local area network (LAN) configurations.

2.1.5 Security

A secure environment is one that limits physical and electronic access to systems by unauthorized users. Most businesses can achieve a secure environment with little or no performance overhead. However, if security is your highest priority, you may need to make tradeoffs in convenience, cost, and performance.

Reference: See the OpenVMS Guide to System Security for more information.

2.2 Determining Application Requirements

Applications require processing power, memory, storage, and I/O resources. Determining your application requirements allows you to design an OpenVMS Cluster system that will meet your application needs. To determine your application requirements, follow the steps described in Table 2-2.

Systems require approximately 5% more memory to run in an OpenVMS Cluster than to run standalone. This additional memory is used to support the shared cluster resource base, which is larger than in a standalone configuration.

With added memory, a node in an OpenVMS Cluster generally can support the same number of users or applications that it supported as a standalone system. As a cluster configuration grows, the amount of memory used for system work by each node may increase. Because the per-node increase depends on both the level of data sharing in the cluster and the distribution of resource management, that increase does not follow fixed rules. If the node is a resource manager for a heavily used resource, additional memory may increase performance for cluster users of that resource.

Reference: For more information about using additional memory to improve performance, refer to the OpenVMS Performance Management manual.

2.2.2 Balancing Processor, Memory, and I/O Resources

Application performance depends on adequate processor, memory, and I/O resources. Depending on your applications, one of these resources may be more important than the others. Consider your application requirements, and find a balance among these three resources that meets your requirements. Table 2-3 provides some guidelines on the resource requirements of different application types.

The OpenVMS operating system supports a number of utilities and tools that help you determine your business and application requirements in OpenVMS Cluster configurations. Table 2-4 describes many of these products and indicates whether each is supplied with the OpenVMS operating system or is an optional product.

OpenVMS Partners offer a wide selection of tools to meet diverse system management needs, as shown in Table 2-5. The types of tools are described in the following list:

• Schedule managers
  Enable specific actions to be triggered at determined times, including repetitive and periodic activities, such as nightly backups.
• Event managers
  Monitor a system and report occurrences and events that may require an action or that may indicate a critical or alarming situation, such as low memory or an attempted security breakin.
• Console managers
  Enable a remote connection to and emulation of a system console so that system messages can be displayed and commands can be issued.
• Performance managers
  Monitor system performance by collecting and analyzing data to allow proper tailoring and configuration of system resources. Performance managers may also collect historical data for capacity planning.

For current information about OpenVMS Partners and the tools they provide, access the OpenVMS web site: http://www.openvms.compaq.com/.