June 2003
Making Business Realities Work For You
Part VI: Why �best effort� isn�t the best for your
enterprise network.
BY TONY RYBCZYNSKI
Business Reality: The rules of the game have changed.
For at least three years, our economy has been on a wild ride that has
challenged all notions about �business as usual.� Global network
connectivity increased corporate reach and revenues, but also blurred the
lines between internal and external resources, between trusted and insecure,
private and virtually private. The communications advancements that spawned
new opportunities for collaboration also accelerated the pace of change and
ratcheted up every baseline expectation for productivity and turnaround. The
eCommerce applications that generated gigabytes of data at every phase of
the process also drowned the systems that were supposed to capture, digest,
and share them. The proliferation of new contact channels produced
disconnects that led customers to wonder if they were in fact dealing with
one company.
These realities highlight the urgent need to increase IT agility in
serving business objectives, and to implement a networking infrastructure
that can meet evolving application demands.
When �Best Effort� Isn�t Good Enough
Traditional IP networks were built on hubs, bridges, and routers with
limited or no redundancy. End-to-end reliability was achieved through a
combination of dynamic routing and application-level TCP error recovery. IP
networks offered best-effort networking only, relying on IT�s ability to
stay ahead of the traffic growth curve.
The challenge is to create IP networks that support a full spectrum of
business application needs, including transactional, interactive, and
collaborative services and file transfers. However, meeting application
needs goes beyond nodal reliability and quality of service (QoS)
considerations, as is illustrated by these real-world examples:
� Routing protocol hop count limits can result in logical disconnects
that result in a user not �seeing� a network resource such as a printer. In
such a case, the network may be 100 percent up, and the application 100
percent down.
� Network congestion can increase packet loss, which in turn impacts
audio and video streaming and IP telephony applications. IP telephony users
may have to hang up the call and try again.
� Back-end processing delays may prompt customers to abandon Web
transactions and visit competitive Web sites instead.
� Excessive file transfer delays could conflict with financial book
closing intervals or customer contractual commitments.
� Users may get sluggish server response, while other servers that
support the same application are underutilized.
� High participation in a Webcast can degrade performance to all users
due to network congestion.
Taking months to redesign the network to accommodate a new application that
will deliver competitive advantage is the wrong answer in today�s world.
Most enterprises recognize the need to evolve their networks from a
best-effort IP world. Some will respond by according special treatment to
selected applications, such as customer care, transaction services, or IP
telephony. Others will run the network as an internal service or utility,
offering a range of service level agreements (SLAs) to their business units.
Whatever the strategy, the foundation is to adopt an application-oriented
view of end-to-end network performance, rather than a network-oriented view.
Technology Response: Instead of �best-effort,� think
�content-aware intelligence.�
When building and designing enterprise networks, the IT organization must
foster an application-oriented view. Application-optimized networks:
� Provide the agility to support diverse data, voice, and video
applications;
� Eliminate network bottlenecks that inhibit introduction of new
applications;
� Consistently and reliably deliver required levels of connectivity,
latency and throughput; and
� Embrace data centers, server farms, and databases, as well as
communications links among them.
Let�s take a look at some key considerations:
Rapid recovery. Real-time, delay-intolerant applications such as IP
telephony and interactive collaboration require rapid recovery from
equipment and physical link failures. At the nodal level, key core elements
must be engineered for redundancy and rapid recovery. Simplified network
architectures (fewer tiers and fewer boxes) using Layer 2/3 switching ease
networking engineering and design, while minimizing fault isolation and
recovery times. For example, multi-link trunking and resilient packet rings
provide highly reliable Ethernet networking.
Quality of service. Even when bandwidth is over-engineered, QoS
mechanisms are required to ensure performance requirements are met under
normal, congested, and failure conditions. Very low latency across network
switches and a manageable number of user traffic categories across the
network can support various classes, such as real-time, delay-intolerant
(premium); real-time, delay tolerant (platinum, gold); non-real-time,
mission-critical (silver, bronze); and non-real-time non-mission critical
(standard). QoS classification functionality should be implemented as close
to the application as possible, ideally in the application itself.
Policy management. �Closed loop� policy management ensures that QoS and
security policies are consistently applied across the network, and includes
configuration of edge devices, enforcement of policies in the network, and
verification of performance and security controls in line with corporate
policies. Enforcement of policies in the network also includes admission
controls of applications vying for access to network and application
resources.
Reconsidering The Network Edge, Core, And Operations Center
Extending application-optimized enterprise network performance and
reliability across service provider networks requires a high degree of
flexibility at the enterprise edge. QoS has to be mapped to the appropriate
service provider protocol mechanisms, while optimizing price/performance.
Reliability options include Layer 1, 2, and 3 mechanisms such as SONET and
DWDM redundancy, ATM multilink, IP dual homing, and ISDN backup. Rapid
detection of access and service failures on IP, frame relay, and ATM are
also a necessity. Emerging optical Ethernet MAN/WAN networking eliminates
many of the barriers to end-to-end application-optimized networking, by
establishing a reliable consistent Ethernet-based networking infrastructure.
Finally, on the networking front, more rigorous operational and
engineering procedures within the enterprise need to be adopted, recognizing
the transition from best-effort networking to always-on,
application-optimized networking.
These considerations apply across the enterprise network -- and to the
data centers, server farms, and databases that fuel business-critical
applications. Here�s where Layer 4-7 networking comes in -- extending
intelligent security and networking technologies across entire application
infrastructure to enhance reliability and performance.
The Role Of Layer 4-7 Networking In The Application-Optimized Network
Layer 1-3 routers make decisions based on physical, link, and
network-level information -- which port, which interface, which IP address.
That�s fine and well for best-effort routing, but the new Business Realities
call for the advantages of Layer 4-7 traffic management and application
switching.
Layer 4-7 application switches and networking devices know link and
network information, just as Layer 1-3 devices do, but they also know
application-level information, such as what type of user or device is
requesting the content (handheld device, frequent shopper, first-time
visitor, etc.), what type of content the user is requesting (executable
script, static content, streaming Web cast, shopping cart, etc.), and more.
These capabilities reduce global network load and costs, improve server and
network response time, and enable differentiated services that cannot be
achieved any other way.
Application switches and intelligent networking devices support a number
of advanced services, such as:
� Intelligent load balancing across severs, firewalls, and many other
devices, and content-intelligent application redirection, based on full
Layer-7 inspection of URLs and cookies.
� Content-aware security that protects servers and applications against
attacks and unwanted intrusion and SSL (Secure Socket Layer) acceleration,
relieving servers of the processing-intensive tasks associated with
encryption.
� Intelligent bandwidth management capabilities that can meter, control,
and account for resource utilization by any Layer 2-7 attribute, including
client, server farm, filters, service, application, user class, URL,
cookies, and content type.
� Content caching capabilities, that cache frequently used content close
to requesting users and apply other �smart� storage strategies to deliver
content in a bandwidth-efficient fashion.
Purpose-built networking platforms, selectively operating across Layers
1-7, including integrated routing, are starting to replace legacy
multi-protocol routers as the workhorse of enterprise networks.
Application-optimized networking, include content-aware intelligence,
deliver the functionality required to win, now that the rules of the game
have irrevocably changed.
Tony Rybczynski is director of strategic enterprise technologies for
Nortel Networks with 30 years experience in networking. For more
information, visit the company�s Web site at
www.nortelnetworks.com.
[ Return
To The June 2003 Table Of Contents ]
|