Almost all the VMWare and IBM or Microsoft cluster architectures call for LAN connectivity as a baseline requirement. Hence, emulating LAN service while still connected via a WAN poses a lot of challenges for convergence (ideally within a few seconds).
The solution requirements in Table 1-7 call for an infrastructure that has the features outlined in Table 1-8.
Table 1-7 DCI Feature Requirements
Feature | Details |
Layer 2 extensions | Typically using pseudowires. |
Spanning Tree Protocol (STP) isolation | Spanning-tree isolation is one of the must-haves, where each DCI does not extend the STP to avoid any loop. Having redundant links functioning at the same time without STP in the core. |
HA | The DCI edge must deal with node and link failures. |
Faster convergence | This needs to be as small as possible in case of node or link failure. Ideally, anything less than a few seconds. |
Secure communication | Encryption, such as IPsec-based solutions. |
QoS | Hierarchical QoS for DCI. |
WAN optimization | DCI WAN optimization using WAAS technologies. |
Maximum transmission unit (MTU) requirements | Jumbo frame support. |
The solution requirements in Table 1-7 call for an infrastructure that has the features outlined in Table 1-8.
Table 1-8 Router/Switch Feature Requirements Needed to Meet the DCI Solution Requirements
Feature | Details |
Layer 2 extensions | Using EoMPLS (p2p), or VPLS (point to multipoint). |
STP isolation per data center | Capability to terminate the STP at the given data center itself. Redundant links functioning at the same can be provisioned using Cat 6500 Virtual Switching Systems / Multichassis Ethernet Channel (VSS/MEC) and/or Nexus 7K vPC (virtual port channel). |
HA | Usage of redundant routers (ASR 1000, for example) or switches (6500/Nexus 7K). |
Faster convergence | There are two broader approaches: EoMPLS remote port shutdown via laser off (supported on ASR 1000). Using Embedded Event Manager (EEM) or undirectional link detection (UDLD) on 6500, Nexus 7K, or ASR 1000. |
Secure communication | GRE over IPsec solution, or Nexus TrustSec (Cisco TrustSec based on IEEE 802.1AE link-layer encryption). |
QoS | Hierarchical QoS at the DCI edge. |
WAN optimization | WAN optimization using WCCPv2 or PBR using existing Cisco WAAS appliances. |
MTU requirements | Jumbo frames are supported on Cat 6500, Nexus 7K, and ASR 1000 GE / 10 GigE links. |
~Large Branch WAN
As universally understood, not all branches are equal. This is not only true for the size of branch (as in number of users or perhaps application servers residing at the branch) but also for how critical the branch is to the overall business function. Consider bank branches, for example. Not all branches provide the entire portfolio of services. In the real world, some provide only basic banking services, whereas others provide full-blown services, including home mortgage, small business loans, and investment services to commercial customers.
Large branches (those that provide more services or services that are critical to the business, or in most cases both) tend to have slightly different requirements for a WAN infrastructure that connects them to the corporate backbone. Table 1-9 outlines the large branch WAN requirements.
Table 1-9 Large Branch Office Deployment Requirements
Requirements | Details |
Larger bandwidth uplink | OC3, or even Metro Ethernet. |
Ability to handle both WAN and Internet traffic | Because of the volume of traffic, large branches are connected directly to the Internet. |
Multitenancy | Capability to support multiple departments or even customers or partners that use the common physical infrastructure along with employees. |
QoS | Hierarchical QoS to support multiple levels of classes of service. |
Class-based WFQ, fair queuing, LLQ, WRED Traffic shaping. | |
Services requirements | Services such as NAT, firewall, and NetFlow at high speeds and scale. |
HA | Intra and interbox HA supporting basic traffic forwarding and services. |
Table 1-10 maps the requirements onto the infrastructure needed to support such requirements.
Table 1-10 Large Branch Office Requirements/Traits
Requirements | Infrastructure Traits to Meet Them |
Larger bandwidth uplink | Interface diversity |
Ability to handle both WAN and Internet traffic | Modular data and control plane to deal with the increasing set of requirements |
Multitenancy | Capability to support virtualization of interfaces, services, and routing/forwarding tables |
QoS | Flexible architecture being able to adopt to changing QoS requirements via software upgrade |
Services requirements | Capability to support the existing and newer services with the existing hardware via software upgrades |
HA | Inherently highly available system |
Summary
This opening chapter covered the basic building blocks of WAN architectures:
Branch aggregation
Internet edge
Data center interconnect
Large branch office
Although the basic requirements are common across the various roles, they differ significantly enough that you need to understand how they are architected, deployed, and troubleshot. If there were one word to describe the hardware required to meet these needs, that word would be flexibility. Infrastructure needs to be very flexible in terms of feeds and speeds, scale and performance, service richness, and interface diversity (to name a few).
The next chapter covers the various business drivers, and the underlying technical requirements that they are generating. It concludes with an analysis of how these are driving requirements for next-generation WAN infrastructure.
Review Questions
What are the four usual WAN architectures?
What does optimized WAN mean?
What are the few fundamental requirements for WAN aggregation?
Why is service richness so important to enterprises?
Why would an enterprise connect directly to a service provider or be multihomed?
What are the core business drivers for DCI?
Answers
Usual WAN architectures typical in today’s networks include the following:
Branch and private WAN aggregation
Internet edge
Data center interconnect
Large branch WAN
Optimization here refers to the capability of the network infrastructure to provide the voice, video, and data traffic optimization before it goes over the WAN links. This helps reduce the need for more bandwidth every time a new application is added to the network. Cisco WAAS and IOS provide such services.
WAN aggregation requires, at a minimum, infrastructure to support the following:
Flexible routing/switching architecture that can evolve with the changing business requirements
Capability to combine various types and speeds of interfaces into one common infrastructure
Modular and highly available carrier-class design with the separation of control, data, and I/O planes
Capability to add basic services without requiring new hardware
Service richness refers to an in structure (hardware and software) that can introduce a basic service and some of the advanced services into baseline hardware with simple software upgrades. Enterprises and their businesses thrive on applications, and that in turn requires network-based services for them to work on common physical infrastructure and with the ability for virtualization for today’s typical multitenant requirements.
Enterprises connect to multiple Internet providers (or what is generally known as multihoming) for a few key reasons, including the following:
Fault tolerance and resiliency to failure in one provider’s network
Granular routing control
Path selection based on features such as performance-based routing or PfR
Core business drivers behind DCI include the following:
Layer 2 extensions
Data center consolidation
Data center disaster recovery site
Virtualization and clustering applications such as VMWare’s VMotion
Further Reading
Introduction to WAN, document: http://tinyurl.com/6g8cym
Cisco Validated Designs, document: http://tinyurl.com/lnnyjt
Unified WAN Services, document: http://www.cisco.com/en/US/netsol/index.html
DC Interconnect, document: http://tinyurl.com/rclv2f
© Copyright Pearson Education. All rights reserved.