Juniper VIRTUAL CHASSIS TECHNOLOGY User Manual

White Paper

NETWORK SIMPLIFICATION WITH JUNIPER NETWORKS VIRTUAL CHASSIS TECHNOLOGY

©2020, Juniper Networks, Inc.

Network Simplification with Juniper Networks Virtual Chassis Technology

TABLE OF CONTENTS
Introduction..................................................................................................................................	3
Data Center Network Challenges..............................................................................................................................	3
Enterprise Network Challenges.................................................................................................................................	5
Juniper Networks Virtual Chassis Technology.......................................................................	6
Simplifying the Data Center ......................................................................................................................................	6
Reduced Complexity ...................................................................................................................................................	7
Simplifying the Enterprise Network ........................................................................................................................	7
Cost-Efficient Use of Resources................................................................................................................................	9
Improved Performance ...............................................................................................................................................	9
Product Portfolio.........................................................................................................................	9
Conclusion..................................................................................................................................	11
About Juniper Networks..........................................................................................................	11

©2020, Juniper Networks, Inc.	2

Network Simplification with Juniper Networks Virtual Chassis Technology

EXECUTIVE SUMMARY

The proliferation of mobile devices—combined with trends such as cloud, analytics, and social media—is fundamentally changing user behavior and network usage patterns, driving demand for bandwidth and a growing need for resiliency and security. These changes are making enterprise and data center networks increasingly complex, adding to IT’s economic burden by creating environments that are inherently difficult to manage and operate.

Network complexity is by far the single biggest roadblock to scalability for data center and enterprise networks. Juniper Networks® Virtual Chassis technology—available on Juniper Networks EX Series Ethernet Switches and QFX Series switches—offers an innovative, unique solution for deploying and growing data center and enterprise networks, helping organizations address this growing complexity by making network architectures simpler, more reliable,

and more cost-effective to manage and maintain. This paper describes how Virtual Chassis technology on EX Series and QFX Series switches reduces network complexity by simplifying network operations and architecture, increasing scalability and performance while reducing operational costs.

Introduction

Data Center Network Challenges

Most data center networks today employ legacy three-tier architectures. These networks are plagued by complexity and inflexibility in the various network layers, leading to poor performance and excessive resource consumption (Figure 1).

L2/L3	L2/L3
Switch	Switch

	SSL VPN
	Firewall
	IPsec VPN	L2/L3	L2/L3
	IPS
		Switch	Switch

	Servers	NAS
		Storage

FC SAN

Figure 1: Legacy data center architecture

©2020, Juniper Networks, Inc.	3

Network Simplification with Juniper Networks Virtual Chassis Technology

Design and Operations

Data center networks typically use a variety of design architectures to provide access layer connectivity. Two of the methods are top-of-rack and end-of-row deployments. While these deployment methods have their

advantages, they also suffer from excessive resource consumption, inefficient bandwidth utilization, and costly cable management. In addition, server virtualization, which has been adopted in today’s data center, demands networks that are nimble and adaptable enough to react quickly to changes and maintain efficient delivery of mission-critical services.. Virtual servers and the workloads running on them need to be migrated either within or across data centers based on demand and load. The network has to be able to support this migration and adapt to the changes quickly with no risk of service disruption.

Such changes require that devices appear to be seamlessly connected on the same network, regardless of their physical proximity to each other. Legacy network architectures lack the flexibility to react to these types of changes, and this adversely affects the operational efficiency of the entire data center.

In addition to the way boxes are physically connected and oriented, maintaining the network becomes a challenge when existing technologies prevent network engineers from doing their jobs efficiently. Routine tasks such as monitoring devices, troubleshooting, providing configuration management, and maintaining software upgrades become increasingly difficult as the number of independent devices in the network increases.

Such operational challenges are further compounded if these devices are running different versions of software or have different configurations, since software must be carefully managed across devices to ensure consistent

functionality and limit exposure to defects or other vulnerabilities. Special training or expertise might also be needed to support these configurations. As a result, the effort and resources required to adequately operate, maintain, and troubleshoot the unique requirements of each network device can be enormously time-consuming and expensive.

Performance

The complexities of today’s data center architectures lead to increased latency, delays in network convergence, and limited bandwidth availability.

•Latency caused by the network architecture: Approximately 75% of all traffic in today’s data center is server- to-server, which means it travels laterally, or east to west, across the infrastructure. However, due to the multilayered architecture employed by most data center networks, this traffic must first travel north and south from the access layer up to the aggregation and core layers and then back down again before it reaches its final destination—a costly, inefficient use of network assets that adds latency and complexity to each transaction.

•Suboptimal use of access and uplink ports: In today’s data center, approximately 50% of access layer switch ports are used for inter-switch connections to higher-layer devices in the hierarchal tree, limiting the bandwidth available for supporting customer connections.

•Layer 2 control plane scaling: Spanning Tree Protocol (STP) is typically employed to prevent network loops from occurring in the data center. However, STP can take up to 50 seconds to converge in a network following a failure—even the Rapid Spanning Tree Protocol (RSTP) can require tens of seconds to converge in some topologies. Plus, both STP and RSTP render half the ports in the core and aggregation layers unusable, leading to inefficient bandwidth utilization.

Virtualized servers compound these problems, since they, too, require high performance and low latency.

Resource Consumption

Typically, the quest for higher bandwidth in modern data centers involves operators adding more network devices that end up consuming additional rack space, power, and cooling. However, this extra resource consumption does not necessarily translate to the required scale in bandwidth. This is an inefficient approach since the data center can rapidly run out of power due to excessive resource consumption and the bandwidth challenge now becomes multifold while the operating costs rapidly escalate.

©2020, Juniper Networks, Inc.	4