With continuous performance improvements in PC and LAN technology-such as
gigabit Ethernet, multi-core processors, and better desktop software -delivery
of applications and data within a locally connected environment is quick and
easy. However, the same cannot be said when it comes to maintaining an
acceptable level of application performance across the WAN. In fact, this is a
consistent challenge for many organizations and one where throwing more
bandwidth at the problem is not the answer.
The Scourge of the Network
The key challenge for IT managers is the perennial networking issue known as
latency-the silent killer of applications. Latency is simply the result of
physics, 'namely, the speed of light over longer distances combined with
store-and-forward hops across routers. And because reducing latency itself is
not possible, IT managers need to consider reducing the impact latency has on
the behavior of enterprise applications.
In addition, there are four areas that have slowed application performance
over the WAN for years:
- Application performance visibility: If IT managers do not know what
is happening in the distributed enterprise network, they will not have a clear
idea how to improve performance. - Application contention: Competition between applications will
increase, causing business-critical applications to be slowed by non-critical
traffic. - Bandwidth limitations: WAN links offer limited bandwidth. New
application rollouts, which are inevitable in any large organization, will
consume even more. - Availability: Without a connection to the data-center or the rest
of the enterprise, work simply stops. Creating an application-fluent
infrastructure, which understands the content and transactions at network
layers 4 through 7, is an absolute necessity to deliver transaction
completeness.
Therefore, an application acceleration solution is not about guaranteeing
uptime or throughput or transactions per second, but providing the user with a
consistent, high-performance, reliable and secure application experience.
The Application Acceleration Attraction
As more organizations begin to understand that buying more bandwidth is not
the answer, they have turned their attention to address the real user
concern—faster transfer of files within the WAN.
On this premise, application acceleration promises to provide substantial
operational savings by eliminating the need to deploy servers in remote branch
offices, dramatically simplifying system maintenance, management and
administration with a smaller IT staff. With an application acceleration
solution, detailed insight into the WAN environment will allow the IT manager
the ability to monitor and record activity and understand how the network,
remote users, and applications interact.
WAN Best Practices
With increasing budget pressure, application acceleration solutions have
begun to be viewed as attractive in removing WAN bottlenecks. It is all about
knowing how to optimize the WAN environment. In our 'Best Practices for WAN
Optimization' whitepaper, we listed a set of best practices that enterprises may
want to consider when implementing a WAN optimizing solution. Briefly, the best
practices are:
- Consider compression and caching techniques rather than increase the
pipe size: The default option for increasing the size of the WAN link has
been to upgrade the capacity of that constrained link. However, this is
expensive. A cost-effective alternative is to use market-leading compression
and caching to gain instant WAN capacity on the existing network. Based on our
experience, and depending on the application mix, businesses typically gain a
two to four fold increase in capacity on existing WAN links by eliminating
repetitious data. Some enterprises have seen as much as a 10-fold increase. - Reduce inefficient protocols: To speed transmissions between
business applications, WAN optimization and application acceleration platforms
need to overcome the impact of latency. To do so, an application acceleration
platform requires both TCP acceleration to benefit a broad range of
applications and application-specific acceleration to speed applications whose
Layer-7 protocols are less efficient than TCP. - Quality of Service (QoS) support and configuration: Instead of
relying on per-flow manipulation of traffic with current QoS policies,
enterprises should consider approaching QoS and bandwidth allocation with a
pragmatic, solutions-based orientation, making sure that mission-critical
application traffic arrives where it needs to within the required service
level agreement (SLA). This approach strikes the needed balance between
performing QoS in the most appropriate device-the WAN optimization
platform-and keeps implementation simple. - Choice of deployment: Delivering applications throughout a
distributed enterprise requires great flexibility in options for network
placement and configuration, transport support, and capacity ranges. Thus, for
network placement, IT must have the choice to deploy WAN optimization
equipment either on the network between a LAN switch and WAN router, or
attached to a switch and router in a one-armed fashion. - Usability: Given the increasingly pervasive deployment of
application acceleration platforms throughout an enterprise's locations,
having a system that's easy to use is vital. A number of essential factors
determine a device's usability, including an intuitive interface, automated
configuration and deployment capabilities, and synchronized communications
among platforms. A good application acceleration solution should be able to
pre-stage configurations centrally via templates and then have remote WAN
platforms download them automatically. - Maximize the value of multiple WAN links: Enterprises increasingly
seek to take advantage of hybrid public/private WAN transports, but need the
assurance that key performance criteria will still be met. To make effective
use of both paths, IT needs to apply business policies to each link and
monitor performance. Most WAN optimization platforms overlook this WAN
deployment scenario and fail to help IT make full use of these dual links. - Get comprehensive insight to WAN traffic patterns: Defining
policies for optimizing traffic flows over the WAN requires that IT understand
the actual traffic flows. Monitoring tools that provide unified insight into
distributed applications and networks are essential to effective application
delivery. - Scalability: Because enterprises have a variety of locations that
range in size, applications, and link types, businesses need a wide range of
WAN optimization and application acceleration platforms. Thus, IT should look
for platforms that scale the performance of a wide cross-section of the
enterprise's applications. - Security: Enterprises no longer have the luxury of securing just
their private WANs and avoiding the Internet for business transactions.
Instead, they need to make use of both transports. Thus, IT needs tools to
make all transports secure enough for these business transmissions. For WAN
optimization platforms, two aspects of security are critical: securing the
device itself and securing the data that traverses the device.
Based on these best practices, the ideal WAN framework should include
compression and caching, acceleration, application control, and visibility. As a
result, this integrated approach allows IT to successfully provide LAN-quality
application delivery across distributed enterprises. Therefore, by recognizing
application acceleration as a strategic business asset, IT managers can address
problems of organizational efficiency with high-performance networking that
helps drive market share, revenue and profit.
Sanjay Jotshi,
The author is director of Enterprise and Channels, India & SAARC, Juniper
Networks
vadmail@cybermedia.co.in