|
Introduction
Today's bandwidth hungry intranet and multimedia applications drive the demand for more and more bandwidth, while mission-critical applications call for network designs that offer high
availability. Enterprises, faced with increasing deployments of switched Ethernet to the desktop, are searching for higher bandwidth solutions than Fast Ethernet technology can currently provide.
Solutions implemented today need to protect existing investments, while providing a stable migration path to Gigabit Ethernet bandwidths and beyond.
Cisco's Fast EtherChannel and Hewlett-Packard Company's Auto-Port Aggregation software address these business needs. Fast EtherChannel technology builds upon IEEE 802.3-compliant Fast Ethernet by
grouping multiple, full-duplex, point-to-point links together as one logical link. Bandwidth aggregation of up to 800 Mbps can be achieved in multiples of 200 Mbps with each full-duplex Fast Ethernet
link. Fast EtherChannel's Port Aggregation Protocol (PAgP) aids in the automatic creation of these Fast EtherChannel connections, and now, through close collaboration with Cisco, HP has implemented
Cisco's PAgP in its Auto-Port Aggregation software on the HP 9000 series of Enterprise servers. By drawing on their respective strengths, Cisco and HP have come together to provide a solution that
offers increased performance, scalability, manageability, and availability.
Fast etherchannel applications
Cisco's Fast EtherChannel is well suited for environments with high bandwidth and high availability requirements. In today's rapidly changing environments, managers are witnessing phenomenal
increases in network growth and usage.
Frequently, thousands of clients require access to mission-critical servers that may be distributed or centrally located. Fast EtherChannel can be implemented to aggregate PC connectivity to
application and database servers in a client/server environment.
Web access is proving to be the new information distribution technique of the 1990s. Internet/intranet information distribution and transaction processing requires fast response times and high
availability. The use of multimedia applications involving voice, video, and data are also on the rise pushing the need for greater bandwidth, improved performance, and reliability.
Data from smaller servers and PCs are frequently archived to special, dedicated backup servers. In the event of a link failure, transmission of large backup files must be restarted. Fast
EtherChannel's high bandwidth and redundant capabilities can ensure reliable delivery to these backup servers.
Fast etherchannel topologies
Some common network applications for Fast EtherChannel are shown in Figure 1. Fast EtherChannel can be used to increase performance and resiliency in the data center and wiring closets between
various devices:
- Multiple switches
- Multiple routers
- Switches and routers
- Switches and servers
Fast etherchannel components
Cisco's Fast EtherChannel is a trunking technology based on grouping together multiple full-duplex 802.3 Fast Ethernet connections to provide fault-tolerant, high-speed links between switches,
routers, and servers. It consists of the following key elements:
- Fast EtherChannel links
- Redundancy
- Management
Fast EtherChannel links use the same standard 802.3 mechanisms for full-duplex auto-negotiation and auto-sensing. Each Fast EtherChannel connection can be composed of two, three, or four
industry-standard Fast Ethernet links to provide load sharing of traffic with up to 800 Mbps of usable bandwidth.
Load balancing capabilities mean that links can be added to a channel with no performance degradations. High throughput and low latencies can be maintained while gaining more total available
bandwidth. Load sharing of traffic across a Fast EtherChannel group is based upon source and destination MAC or IP address.
Fast EtherChannel connections can be used on any type of media, whether Unshielded Twisted Pair (UTP), single mode, or multimode fiber. These connections are fully compatible with Cisco IOS routing
technologies and are supported on Cisco Catalyst 1900, 2820, 2900XL, 5000/5500, and 8500-series switches and Cisco 7500-series routers. Fast EtherChannel is also supported on stackable HP 1600 and
8000 switches as well.
Fast EtherChannel is also supported on multiple operating system platform and local area network (LAN) adapter products, including operating systems HP-UX version 11.0 and later, and a select number
of HP 9000 Fast Ethernet adapters.
When attached to servers, a Fast EtherChannel connection to a switch has only one Layer 3 address, that is, one IP address. As shown in Figure 2, a single address is used even when a server has a
multiple port adapter (Example 1) or multiple adapters (Example 2). The operating system sees only one logical link.
Resiliency and Redundancy are other key features of Fast EtherChannel. If a link should fail within the Fast EtherChannel group, Fast EtherChannel provides automatic fail-over for the lost link by
redistributing traffic loads across the remaining links within the channel.
Fast EtherChannel also uses an approach that leverages the recovery features of the network without adding complexity or creating incompatibilities with third-party equipment or software.
Fast EtherChannel uses a peer-to-peer control protocol that provides autoconfiguration and sub-second convergence times for parallel links, yet allows higher-level protocols, such as Spanning Tree
Protocol, or existing routing protocols, to maintain topology. Since it operates with the existing network framework, network managers can leverage their existing network topologies, augmenting
bandwidth by installing Fast EtherChannel where single Fast Ethernet links were previously installed.
Manageability is also included among the many features of Fast EtherChannel. It can easily be configured using the Command Line Interface or by SNMP management applications such as CiscoWorks.
Cisco Works for Switched Internetworks will graphically display Fast EtherChannels between devices as well as collect statistics for both individual Fast Ethernet links within the channel and
aggregate statistics for the Fast EtherChannel.
Another integral management benefit of Fast EtherChannel is the ability to detect, report, and prevent the use of incorrectly paired interfaces within the channel. Consistency checks are completed
before activation of a channel to ensure network integrity.
A key enabler of these capabilities is PAgP, which aids in the automatic creation of Fast EtherChannel links. Upon power-on or reconnection, a port sends PAgP packets between Fast
EtherChannel-capable ports. Neighboring devices learn of each other's port capabilities and their ability to form channels.
After the protocol determines correctly paired, bidirectional, point-to-point links, it groups the ports based on device hardware, software, and user-configured capabilities into a channel.
With PAgP implemented both in the network and on mission-critical servers, Fast EtherChannel bundles can be automatically discovered and configured with transparent ease.
HP auto-port aggregation software and mc/serviceguard
HP's Auto-Port Aggregation software extends Fast EtherChannel's PAgP automatic configuration benefits to HP'server connections and enhances Fast EtherChannel's resilient features with HP's
MC/ServiceGuard's high availability capabilities.
MC/ServiceGuard is a specialized facility for protecting mission-critical applications from a wide variety of hardware and software failures. MC/ServiceGuard consists of a robust architecture of HP
9000 Enterprise Server clusters of up to eight nodes. These organized enterprise clusters deliver highly available application services to LAN-attached clients.
MC/ServiceGuard monitors the health of each node and quickly responds to failures in a way that eliminates or minimizes application downtime. MC/ServiceGuard is able to automatically detect and
respond to failures in the following components:
- System processors
- System memory
- LAN media and adapters
- System processes
- Application processes
MC/ServiceGuard increases server availability in the event of system or component failure by allocating resources within the Enterprise cluster and migrating applications to the appropriate servers.
When MC/ServiceGuard and HP's Auto-Port Aggregation software are implemented together, MC/ServiceGuard can also redirect traffic, in the event of channel failure, to other channels connected to
standby servers within the same cluster. Resources are reallocated automatically and transparently without service interruptions.
Figure 3 illustrates the recovery mechanisms enabled through the use of Fast EtherChannel, Auto-Port Aggregation software, and MC/ServiceGuard. If a link within a Fast EtherChannel group fails,
traffic is rerouted to another link within the channel in less than a few milliseconds.
In the event that a Fast EtherChannel is cut, Auto-Port Aggregation software and MC/ServiceGuard ensure application resources are migrated to standby servers within the cluster which have been
reconfigured accordingly, including the transfer of the failed connection's IP address to the new connection. MC/ServiceGuard uses a "heartbeat link" between servers to monitor the servers and their
connections. Shared databases enable resources to be accessed by other'servers in the case of failure. Fail over for both links and Fast EtherChannels occur transparently to both end user and
applications and require no operator intervention to reconfigure. Reducing operator intervention means reducing the total cost of ownership.
Benefits summary
Fast EtherChannel is a technology-leveraging, standards-based Fast Ethernet used in parallel to provide the additional bandwidth networks require today. It is readily adaptable in the future to
provide Gigabit EtherChannel as Gigabit Ethernet standardization is completed.
In the interim, it provides flexible, scalable bandwidth with resiliency and load sharing across links for switches, router interfaces, and servers. Fast EtherChannel provides the tools for network
managers to build high-speed solutions today for their mission-critical networks, while leveraging their existing cabling and network device infrastructure.
Further, with Fast EtherChannel's PAgP, Cisco and HP, through its Auto-Port Aggregation software, are the first to offer automatic discovery and configuration manageability benefits to Fast Ethernet
users. These capabilities also lead to lower total cost of ownership benefits. Fast EtherChannel's resiliency is further capitalized when used with HP's MC/ServiceGuard facilities to satisfy today's
requirements for high availability networks.
Increased performance, scalability, manageability, and availability are among a few of the many benefits Cisco and HP deliver to their customers in providing end-to-end Enterprise solutions.
Cisco fast etherchannel features at a glance
Compliant with IEEE standard 802.3
- Flexible, incremental bandwidth up to 800 Mbps
- Gigabit Ethernet ready
- Load-balancing
- Resiliency and fast, sub-second convergence
- Compatible with Cisco IOS, virtual LAN, and routing technologies
- Easy configuration and manageability
- Usable on any type of media whether Unshielded Twisted Pair (UTP), single mode, or multimode fiber
- Supported on Cisco Catalyst 1900, 2820, 2900XL, 5000/5500, and 8500-series switches and Cisco 7500-series routers
- Supported on stackable HP 1600 and 8000 switches
- Single IP address capability for Fast EtherChannel bundles
- Supported on HP-UX 11.0 and later
- Link failover transparent to end users and network applications
- PAgP automatic discovery and configuration features
HP auto-port aggregation software and mc/serviceguard features at a glance
Complete Fast EtherChannel autodiscovery and autoconfiguration capabilities with the implementation of PAgP in HP Auto-Port Aggregation software
- Automatic detection and response to a wide variety of system, LAN, and application component and process failures with MC/Service Guard
- In the event of a failed channel, traffic is redirected between Fast EtherChannel-connected servers
- Auto-Port Aggregation software use is interoperable with all of the following:
- Cisco Catalyst 1900, 2820, 2900XL, 5000/5500, and 8500-series switches and Cisco 7500-series routers
- HP-UX 11.0 and later
- Stackable HP 1600 and 8000 switches
Cisco and HP alliance
HP and Cisco together offer customers the comprehensive end-to-end enterprise solutions they need to address critical business objectives. The alliance has enabled close collaboration in the areas of
technology development, product integration, professional services, and customer service and support and ensures full internetworking compatibility and interoperability between products. More
information about the HP and Cisco alliance can be found at http://hp.cisco.com.
|
