- Posted
- Jul 17, 2006
- | By
Traffic management key to enterprise networks
Technology is intended to create efficiencies - not headaches. Emerging network technologies enable increased productivity, but adopting them may seem - at first glance - to increase the complexity of both the network infrastructure and the network administrator's job. Gurkirat Singh, from ProCurve Networking by HP, writes about the importance of traffic management to overcome the complexity.
This complexity can appear to intensify as more remote users log on, additional applications are extended to partners and customers, and different traffic types are flowing across a single network.
As the enterprise network infrastructure expands to support different types of traffic and users, traffic management becomes critical. Complete visibility into a network's behaviour becomes more important and more challenging.
The importance of traffic management
The LAN has evolved from a transport mechanism to a strategic business tool for many companies. Three major and interdependent forces have driven this transformation: the internet, an increasingly mobile workforce and the convergence of multiple types of data traffic running through the same network.
As a result, networks are becoming more public, more extended and more complex. Enterprises are being forced to understand and support new applications and new connection management solutions for their various constituencies - employees, partners, and customers.
In today's connected business environment, straightforward and effective traffic management from the network core to the network edge is essential. Enterprises need a network infrastructure that scales to meet new business needs and manages added complexity in a cost-effective manner.
In addition, network administrators are expected to control the network in such a way that it is transparent to users. Essential information assets need to be instantly available around the clock. However, this is impossible to achieve without the right tools to make smart, informed decisions. Most network administrators do not have simple, affordable tools that can quickly answer the following questions, regardless of the size of the network:
- Is network performance slowing down or becoming congested?
- Is a network interface card (NIC) chattering, effectively clogging the network?
- What is the current network usage, and what has it been in the past hour?
- Which network routers are most active or over-utilised?
- Why is a server slow or inaccessible?
- Which users and applications are driving network traffic?
- Which users and applications are starving for bandwidth?
- How much bandwidth do I need for new applications?
Types of network traffic
Today's networks must accommodate an increasingly complex set of data traffic. Network traffic management tools enable network administrators to identify traffic types, users and applications, and facilitate the optimisation of the network so all three components interact appropriately.
The first step towards network optimisation is to understand where inefficiencies and problems are located, and why they are occurring. The following examples represent several types of network traffic that must be considered and supported:
Bursty traffic
- Example: Large file downloads such as FTP, multimedia content (.wmv, .swf, .mov files) and graphic content (.jpg, .gif files).
- Problem: Can result in spikes in bandwidth consumption, effectively starving other applications of bandwidth for a brief period of time. Interactive traffic and latency sensitive traffic are particularly susceptible to problems caused by bursty traffic.
- Solution: Set a maximum constraint to limit access to bandwidth.
Interactive traffic
- Example: Secure socket layer (SSL) transactions, instant messenger and telnet sessions all consist of relatively short request/response, and generally support real-time interaction with end users.
- Problem: Susceptible to competition for bandwidth, which can result in poor and unpredictable application response time.
- Solution: Prioritise over less essential traffic and traffic that is less dependent on real-time response (such as email).
Latency sensitive traffic
- Example: Streaming applications, voice over IP and video conferencing all generate a steady stream of traffic, which consumes a significant amount of bandwidth.
- Problem: Susceptible to competition for bandwidth, which can result in poor and unpredictable application response time. These applications can also easily saturate available bandwidth, effectively starving other applications.
- Solution: Set minimum guarantees of access to bandwidth prioritised by business need and set maximums to prevent any application from consuming too much bandwidth.
Non-real time traffic
- Example: Email and batch processing applications are the predominant sources of non-real-time traffic within the enterprise.
- Problem: Can consume bandwidth that could be used by more business-critical applications.
- Solution: Schedule bandwidth assignment for non-business hours; set a maximum bandwidth constraint and low prioritisation during business hours.
Traffic management: standard vs best practices
Making well-informed decisions on how to monitor, manage and fine-tune a business network begins with a traffic management infrastructure strategy built on industry best practices. Understanding and implementing best practices to create assured service levels, and to create efficiencies across network resources, will help network administrators provision traffic and create intelligent networks.
By taking a look at how many businesses provision resources, it's easy to understand how service levels can be assured. Different network practices and outcomes for network service levels and efficient use of resources follow.
Standard practice: Over provision
High capacity reduces possibility of network slowdown. (Downside: unnecessary expense).
Good practice: Static policies
Eliminate unnecessary traffic. (Downside: doesn't adapt to changing business needs.)
Better practice: Dynamic management
Identify historical trends and patterns in order to appropriately set and enforce priorities. (Upside: meets business needs and centrally sets priorities with efficient, distributed enforcement.)
Best practice: Event-driven provisioning
Smart, real-time assignment of resources based on user and application needs. (Upside: provides a hands-free, optimal allocation of network resources that tightly matches business requirements.)
Traffic management technologies
Understanding the traffic traversing an enterprise network has always been a complex task and the solutions for managing traffic have historically been very expensive. Traditionally, network traffic monitoring has been achieved using probes. This worked very effectively in shared networks where a single instrument can monitor all the traffic. However, with switched point-to-point networks, every port on a switch needs to be monitored to achieve the same visibility to network traffic. In addition, switches and routers make packet-forwarding decisions that affect the flow of traffic through a network. Understanding these traffic flows is critical to maintaining visibility to network use and misuse.
More than a decade ago, ProCurve Networking by HP set out to address these problems by inventing embedded sampling technology for network traffic, which is the basis of both its XRMON and sFlow technology today.
By embedding a packet sampling technology into the network devices, ProCurve enabled network administrators to gather data across the entire network, instead of merely a few places where they could afford to install a network probe. This sampling technique also provided exact traffic level measurements and an accurate representation of who was causing the traffic, while adding minimal overhead to the network. Finally, packet sampling made it possible to provide a traffic management solution that scaled in accordance with increasing network speeds.
Meeting today's requirements
The ProCurve Networking Adaptive EDGE architecture was developed to secure the future of ProCurve customers by delivering the best central command with control to the edge. With control to the edge, companies can provide secure, robust functionality to support all current and future traffic and application types.
By definition, it is the network edge where users and applications connect, where traffic enters and exits the network, and where the network must determine how that traffic is handled.
The edge is where security priorities and policies must be enforced, where the user connects after being authenticated at a central command resource.
Without control to the edge, decisions about security and traffic must be deferred to the network core, impacting core performance and scalability, while at the same time requiring more bandwidth in all parts of the network - driving up cost and complexity. In addition, this opens the network to security attacks between the points where access is physically attained and where authorisation is granted.
Control to the edge places selected functions from Layers 2, 3 and 4, and higher functions in edge switches. These switches control access and traffic flows to ensure the increasing set of applications can function correctly and concurrently without interference. Adaptive EDGE networks support both centralised cores and distributed cores with equal facility, because both core configurations can enforce the key decisions made at the edge.
Ultimately, the Adaptive EDGE Architecture enables highly available meshed networks - a grid of functionally uniform switching devices - to scale out to virtually unlimited dimensions and performance due to the distributed decision-making of control to the edge. Moreover, the EDGE architecture facilitates comprehensive traffic management throughout an enterprise infrastructure.
The ProCurve Networking Adaptive EDGE architecture focuses on:
- Traffic monitoring to control bandwidth optimisation throughout the network grid.
- Integration of industry-standard virtual LAN (VLAN) and industry-standard security constructs to provide 'out-of-the-box' access management unavailable on non-EDGE ports.
- Complementary implementation of industry-standard traffic routing and Layer 2 meshing, providing path failover and multi-path load balancing for robust reliability.
- Unparalleled range of traffic prioritisation features to provide traffic type coexistence and quality of services functions, virtually eliminating the need for custom network design architectures to ensure support of current and future voice, video and content delivery applications.
- Centralised command of the edge for easy implementation of user security and application policies.
ProCurve network management systems provide traffic management modules that utilise embedded packet sampling technologies to provide the network administrator with an in-depth view of the traffic levels across all ProCurve ports. The traffic management modules also offer the ability to drill down to see who is responsible for network traffic, so that it is possible to not only see when a problem exists, but also who is causing the problem.
As enterprises add new network applications, support remote users, implement different types of traffic and extend to partners and customers, having complete visibility into an infrastructure's behaviour becomes increasingly important and challenging. More than ever, straightforward and effective traffic management from the network core to the network edge is essential to business success.
"In today's connected business environment, straightforward and effective traffic management from the network core to the network edge is essential."
"Understanding the traffic traversing an enterprise network has always been a complex task and the solutions for managing traffic have historically been very expensive."
Gurkirat Singh has been country manager, South Pacific, ProCurve Networking by HP, since 2004 and an HP employee since 2001. He has more than 20 years of experience in the ICT and engineering industries.
Related Articles
Recent News
