E X E C U T I V E S U M M A R Y
The WAN is
critical to the business of modern enterprises. Despite technology
innovations such as MPLS and xDSL, WAN bandwidth is a finite
resource, while network delay is determined by fundamental physical
constraints. WAN Optimization Controllers (WOCs) have emerged over
the past few years as a way to address application performance
hurdles in selected sections of the network. Enterprises today tend
to tactically deploy WAN
optimization for networked business applications at sites that
suffer from poor end-user experience. This “pain point”
approach can work well, since it often solves the immediate problem
at these sites. However, most enterprises, especially large ones,
require WAN optimization benefits to be delivered globally to
consistently meet the Quality of Experience requirements of their
whole distributed workforce. But traditional WOC solutions
are
faced with four challenges that prevent the delivery of WAN
optimization on a global scale: scalability, efficiency, management
costs and investment costs.
INTRODUCTION
The WAN is critical to the business of modern enterprises. Despite
technology innovations such as MPLS and xDSL, WAN bandwidth is a
finite resource, while network delay is determined by fundamental
physical constraints. WAN Optimization Controllers (WOCs) have
emerged over the past few years as a way to address application
performance hurdles in selected sections of the network.
Enterprises today tend to tactically deploy WAN optimization for
networked business applications at sites that suffer from poor
end-user experience. This “pain point” approach can work well,
since it often solves the immediate problem at the troubled
sites.
However, not all networks will benefit sufficiently from such a
tactical approach to application performance. Most enterprises,
especially large ones, need WAN optimization benefits to be
delivered globally in order to consistently meet the Quality of
Experience requirements of their entire distributed
workforce. Very few large enterprises have generalized the
deployment of WOCs in their network as of yet. There are four key
reasons for that:
Limited scalability - Nowadays many vendors are able to enhance
application performance at a maximum of 10 or 20 sites. Only a
select few are able to scale WOC benefits to hundreds or thousands
of sites.
High management costs - WOCs are complex devices that need to be
individually configured. Each device needs to be configured in
relation to the others, and yet each must reflect local
requirements.
Limited effectiveness - Putting WOCs in selected sites in a large
network often does not improve performance. Modern networks have
meshed topologies that cannot be properly handled by traditional
point-to-point WOCs.
Significant investment costs - Even if the technology trend is to
be more affordable, WOCs still require significant budget
commitments, since they cost many times more than the cost of a
branch router.
In this paper, we consider each of these four challenges, how they
argue against the use of traditional solutions for strategic WAN
optimization deployments, and how next generation solutions can
overcome them all.
In addition to these four challenges, this paper discusses the
importance of application visibility in strategic WAN optimization
deployments. Enterprises should to understand what is at stake
before investing in global deployment. They should also understand
the impact of the technology on application
and business performance. Furthermore, enterprises need to gain
confidence in the operational effectiveness of WAN optimization to
understand the higher level of benefits this new technology can
provide to them. Application Service Level Agreements (SLAs) that
can be configured across the entire
WAN are a good example of what a next-generation solution such as
Ipanema's Business Network Optimization is able to provide to
enterprises that embrace WAN optimization at a strategic
level.
THE LIMITED SCALABILITY OF TRADITIONAL WOCS
Many vendors are today able to enhance application performance at
10 or 20 sites. Only a select few are able to scale WAN
optimization benefits to hundreds or thousands of sites. The
traditional approach to WAN optimization is centered on the notion
of a box, or a pair of boxes, deployed in a few locations and
serving the needs of a limited number of sites. Visibility features
of those solutions, when they are available at all, clearly show
the intrinsic limitations of such an approach.
Traditional visibility is based on the local collection and
analysis of local traffic information. With many WOCs, central
management software can be used to view each site's local data. But
the data is not available in real-time and since it is local to
each equipped point in the network, it cannot be used to
troubleshoot global application performance.
Optimization also suffers from serious scalability limitations.
Traditional QoS, such as found in all WAN optimization solutions on
the market, is based on the concept of “policies” that describe
a combination of flow characteristics and traffic management
actions. The number of policies is often limited, with a maximum of
just a few thousand on the highest end devices. When a traditional
WOC is placed in front of a datacenter talking to numerous branch
office sites, as is typical in medium to large enterprises, the
number of policies is rapidly exhausted. This leaves no option but
to drastically limit the granularity of the policies, and therefore
also limit their effectiveness.
Ipanema's Business Network Optimization solution does not share
these limitations because it has been designed from the ground for
global deployments on large networks. It provides local visibility,
but does so in real-time and it automatically combines the data to
provide coordinated, real-time visibility over the whole network.
Application performance can be easily monitored and troubleshot
globally.
With Ipanema, there are no policy-based limitations related to the
number of sites being optimized. The benefits of QoS can be
delivered with precision whatever the number of sites or directions
(tens, hundreds, or thousands) a datacenter is talking to.
1. How Ipanema scales visibility to the whole network
Captures real-time, synchronized active flow performance data
With Ipanema's system, the flow metrics are time-stamped using a
millisecond accurate system wide clock. The collation of flow
information is done once a minute and results are stored in a
synchronized central table containing all the active flows with all
their performance metrics. Any slice through the data can be done
to perform any kind of aggregation for reports. As a result, it is
possible to view real-time information on arbitrary groupings of
sites and/or applications as simply as if they were handled by a
single device. This flexible and powerful capability is key to
maximizing the efficiency of the troubleshooting process.
Centralizes network wide information using very little bandwidth
The Ipanema system uses a proprietary protocol to collect data from
devices. The protocol is built to scale to thousands of sites while
supporting minute by minute collection of data from all sites. The
reports can be instantiated over one hour intervals and are updated
minute by minute in real-time. Reports thus become powerful tools
for rapid troubleshooting.
The system employs a powerful graphical tool, the “treemap
display,” to implement Ipanema Maps. This provides a compact,
global real-time representation of all of the application flows in
the WAN. Flows are represented with a hierarchical display, with
areas that are proportional to the bandwidth or number of sessions,
and with colors (green/amber/red) that are proportional to the
end-user perceived performance of the constituent flows. At a
glance, an Ipanema map view shows the volume and performance of all
of the application flows on the network – a powerful and unique
capability. This tool takes advantage of the system's synchronized
real-time metric collection to provide a wholly new fault isolation
technology to help
desk, operations and IT personnel.
2. How Ipanema scales QoS to the whole network
The QoS capabilities of traditional WOCs are based on the concept
of policies that describe a combination of flow characteristics and
traffic management actions. The number of policies is limited, and
is often insufficient.
With Ipanema, there are no limitations related to the number of
flow characteristics and management actions. The benefits of QoS
can be delivered with the same high per-flow precision whatever the
number of sites or directions (tens, hundreds, or thousands) a
datacenter is talking to. The system makes use of traffic
forwarding policies, but unlike traditional WOCs, does not force
the user to manually configure and tune them on a site-by-site
basis. Once the performance objectives for an enterprise's
applications have been entered into the system, the system will use
its visibility measurements to dynamically monitor user demand and
the WAN's ability to satisfy it, and will compute globally optimal
forwarding policies on a second-by-second basis to ensure that the
maximum number of users have their performance objectives
met.
MANAGEMENT COSTS OF TRADITIONAL WOCS ARE HIGH
WAN Optimization Controllers are complex devices that need to be
configured individually. The configuration of each device needs to
be consistent with each other, yet all must reflect local
requirements.
With traditional solutions WAN optimization mechanisms need to be
individually configured on each device. This represents a very
large workload when deploying the solution for the first time.
Often a piece of central management software provides a templating
function, a simple way to replicate configurations across devices,
but this software does not really solve the workload issue. With
this approach, local site parameters such as the WAN access
bandwidth or limits in terms of users are combined with flow
management parameters. This means that identical configurations
across all devices are rare and the templates still require
per-site customization.
Furthermore, the configuration of each device needs to be
consistent with all of the others, and there is often no way to
automatically check for consistency, this leads to chronically high
rates of configuration errors. This architectural limitation also
has a tremendous impact on ongoing operations. WANs are living
entities that evolve with the enterprise and its IT changes. When a
new application is launched, when the number of users of an
application evolves, when the site bandwidth changes, or when
application usage patterns vary – WOC configurations need to be
modified across all the devices, with the same associated costs as
for the initial deployment.
Ipanema's performance objective based approach to WAN optimization
results in management costs that are nearly independent of the
number of devices being deployed. The only local site parameter
that needs to be configured is the access bandwidth for each
device. The only parameters required for WAN optimization are a set
of objectives that describes application performance needs to that
have be achieved with WAN optimization. Rather than a policy for
how traffic should be handled on each individual device local
device policies are automatically and dynamically computed based on
a real-time analysis of the network traffic mix. Not only does this
minimize the initial management cost, but ongoing changes in the
WAN are handled either automatically or with only a few simple GUI
operations.
1. The paradigm shift offered by Ipanema: objective based
management
Traditional WAN optimization requires device by device
configuration. And not only mandatory parameters such as the IP
address, but also of a complex set of parameters related to the
traffic management configuration. As a result, the effort to
configure the solution is at least proportional to the number of
sites, and can be very high when the number of sites grows beyond a
handful.
With Ipanema, only the IP address and the access bandwidth
parameters need to be configured locally. All the traffic
management parameters (the “policies”) are computed
automatically from the global application performance objectives
and from a real-time analysis of the traffic mix that takes into
account both the global and the local environment. With Ipanema,
the time to configure traffic management is independent of the
number of sites.
With most WAN optimization solutions, it is recommended to use
“partitions” for the best protection of the performance of
critical applications. Partitions are expressed in terms of
guaranteed bandwidth for all users of an application – taken as a
group. This mechanism has a crucial weakness: it requires knowledge
of the number of users of each application at each site. When the
number of users changes considerably, the parameters of the
partitions also need to be changed. If they are left unchanged, the
traffic management becomes highly sub-optimal and the performance
of applications will degrade.
With Ipanema's system, there is no need to know the number of users
of each application at each site. The dynamic bandwidth allocation
mechanism handles traffic on a flow by flow basis. It is able to
dynamically adapt to an evolving number of users by always
computing optimal traffic management
policies.
With traditional WOCs, a change in network topology or bandwidth
characteristics has significant consequences on operational costs.
Traffic management policy design requires the inclusion of topology
and bandwidth information as part of the policy parameters. Users
must define the amount of bandwidth available for each "direction"
each device is sending traffic to. When the bandwidth on a site
changes or when a new site is added, users must modify the policies
of each device that happens to communicate with this site. In a
fully meshed network this quickly turns into a nightmare of
interrelated complexity.
With Ipanema, a change of network topology or bandwidth
characteristics is handled independently from local device
parameters. To add a site or to change the bandwidth on a site, a
single configuration change is required in the central software,
regardless of the number of devices that are deployed in the
network.
The local device parameters are automatically computed by the
system from the global network topology information.
With many solutions, when a new device is integrated into the
system, users must change the configuration of every device it
exchanges traffic with.
The Ipanema system, adding a device is only a matter of adding the
new device's specifications in the global system parameters. If an
associated site didn’t previously exist in the system, it is a
matter of only a few mouse clicks to add it.
Traffic management mechanisms automatically enforce consistency of
device parameters across the WAN
With all WAN optimization solutions, traffic management
configuration involves defining parameters locally to each device.
In non-Ipanema systems, there is no mechanism that ensures the
consistency of parameters across all devices. Since network
topology information is a required part of the traffic
management policies, it is very easy to forget to modify the
parameters of a device in case of topology changes. This causes
performance problems that can be very difficult to
troubleshoot.
In the Ipanema system, the objective based approach to traffic
management automatically removes the need to check the consistency
of parameters across the network. The parameters are computed
automatically and dynamically by system level algorithms that
ensure that the configuration will not only be consistent but also
optimal.
TRADITIONAL WOCS ARE NOT FULLY EFFECTIVE IN MODERN NETWORKS
Modern networks have meshed topologies that cannot be handled
properly by traditional WAN optimization controllers designed for
point to point deployments.
Putting traditional WAN optimization devices in selected sites in a
large network does not improve performance. Modern networks have
meshed or oversubscribed topologies that cannot be handled properly
by a standalone-box-based and static-policies-based architecture.
Moreover, most WAN
optimization solutions address requirements for only a part of the
traffic. They often include only limited support for interactive
and real-time applications.
Ipanema's architecture is based on dynamic parameters and on
cooperation between devices to coordinate WAN optimization actions
across the whole network. The nature of the traffic is analyzed in
real-time to dynamically manage interactive and real-time flows
according to their requirements. Meshed flows are fully under
control even in the most complex environments, and application
performance can always be guaranteed. Oversubscribed topologies are
handled without sacrificing bandwidth and/or application
performance guarantees.
1. Ipanema sets new standards for QoS management
Differing time sensitivity of different flows is taken into account
with Smart Packet Forwarding
Not all flows have the same characteristics in terms of their
sensitivity to delay. Data transfer flows involve large packets
that do not necessarily need to be forwarded particularly quickly
with low latency. Bandwidth is the key parameter for these flows.
On the other hand, interactive traffic such as Citrix flows
requires packets to be forwarded with a minimum latency to ensure a
good interactive end-user experience. Citrix flows have the
additional complexity that they can change between data-transfer
and interactive behavior inside a single session. Real-time flows
also need to be handled with dedicated mechanisms, since they
require not only delay, but also jitter and packet loss to be
minimized.
With traditional WAN optimization solutions, all flows are handled
in the same way whatever their nature. Priority queuing mechanisms
can help somewhat by giving interactive and real-time flows the
highest priority so that they can be forwarded as fast as possible.
The problem is that priority queuing not only controls latency, but
also controls bandwidth. These solutions cannot handle time
sensitivity of flows and bandwidth requirements independently.
Furthermore, setting the highest priority for Citrix flows so that
their packets will be forwarded first can be dangerous, since, as
previously stated Citrix flows can change in nature and become file
transfers. Giving a high priority to file transfer flows will kill
other true interactive
or real-time flows.
With Ipanema, the bandwidth and latency requirements of flows are
separated. Ipanema’s APOs allow you to configure the traffic
types according to the nature of the flow; real-time transactional
or datatransfer. The per-session bandwidth requirement is described
in another, separate parameter. In
operation, the system's Smart Packet Forwarding feature will
automatically handle the case of a change in the nature of a flow
from interactive to data-transfer and will handle it properly,
preventing it from killing other applications. Ultimately, the
Quality of Experience can be maximized for each flow, whatever its
nature, while making full use of the available bandwidth.
Preventing congestion at the source and the destination for any
flows (UDP, TCP) over any topology (meshed, star)
A network containing a mix of UDP and TCP flows with traffic going
both from the branches to the central sites as well as from branch
to branch has a complex meshed traffic matrix. UDP flows are
particularly difficult to control in this situation. Ingress or
egress TCP flows can be controlled using only one device on either
end of the traffic path. This is not the case with UDP flows, which
can only be controlled at the source. When two sites are sending
UDP traffic to a third site, there is no way for the third site to
control the incoming flows. In fact what happens is that when the
incoming traffic exceeds the bandwidth at the third site,
congestion occurs in the PE router, causing uncontrolled
degradation of the performance of all the incoming flows.
Traditional WOCs' only answer to this problem is to use static
partitions to limit the UDP traffic from site to site. For example,
with VoIP flows, this means limiting the maximum number of
simultaneous calls at each site. Not only does this answer impose
significant limitations, it is also very
difficult to configure and maintain as it relies on site to site
traffic management policies that become nearly impossible to keep
current in networks with more than a handful of sites.
In contrast, Ipanema's devices at the receiving sites use a
patented mechanism to inform the devices at the multiple sending
sites that they are competing with each other. The information
transmitted allows the sending devices to share the destination
access bandwidth as efficiently as if all the flows were handled by
one unique device. The dynamic bandwidth allocation mechanism is
used to control UDP and TCP traffic on a flow by flow basis even in
this complex case. This avoids having to impose any static
limitation on UDP traffic and allows all applications to make full
use of the bandwidth when it is available. For VoIP flows, this
means that the maximum number of calls will only be limited by the
available bandwidth between the sites.
In the presence of “oversubscribed” topologies (a hub and spoke
design where the sum of the bandwidth of the branch sites exceeds
the bandwidth of the central site), traditional WAN optimization
solutions require that the hub site device be configured with the
amount of bandwidth that will be used to guarantee performance of
applications for each direction to/from each spoke site. The sum of
the guaranteed bandwidth cannot exceed the bandwidth of the central
site. This simply means that if a branch with some critical
applications requires bursting to the full bandwidth of the branch,
it is possible, but without any guarantee that it will not be in
competition with some non-critical flows using resources in the
central site.
In other words, network resources will not be optimally used in
terms of overall productivity. With Ipanema, even the simplest
“oversubscribed” topology case benefits from the objective
based approach to traffic management. With our solution there is no
need to define static tradeoffs in terms of
how the central site bandwidth will be used to attempt to guarantee
performance across all directions. All flows will be evaluated in
terms of their needs across all directions. From this analysis, the
system will intelligently allocate the network resources in full
compliance with the business criticality of the flows. If a branch
with a critical flow needs to burst to its full bandwidth, it is
possible, without the critical flow being endangered by some other
non critical traffic sent by another branch.
Though prices have been decreasing with time, traditional WOCs
still cost many times more than the cost of a branch router.
On non-hub and spoke, modern networks, traditional WAN optimization
devices need to be deployed at each site where WAN optimization
required. In many cases, current pricing makes it too expensive to
deploy devices to many of the branches that could benefit.
The Ipanema System supports progressive deployments of devices and
features. For example, companies can obtain visibility and
guarantees of flow performance over their whole network while only
equipping datacenters with physical devices. A first level of
acceleration can even be obtained using our patented asymmetric TCP
acceleration that does not require any device at the branch. Once
full visibility and initial optimization and acceleration is
achieved from the datacenter devices, branch devices are typically
deployed in several phases depending on performance issues
encountered at each site and/or to meet specific application
requirements. All sites can be covered with an investment that will
have a demonstrably good return.
Prevent congestion at the destination for meshed flows, without the
need for a box at the destination
In a network with at least two central sites and a multiplicity of
branches, meshed traffic occurs frequently. In such topologies, it
is common to have users at a branch requesting traffic from at
least two datacenters at the same time. This situation generates
meshed traffic flows. With traditional WAN optimization solutions,
there is no way of guaranteeing performance at the branch without
having a device there. The problem comes from the fact that the
devices at the datacenters do not know that they need to share the
bandwidth at the destination branch. This causes congestion in the
branch PE router and degrades the performance of the flows in a
completely uncontrolled way. A device in the branch tries to
address this situation by controlling the incoming flows on the
receiving end (although only for TCP flows, not UDP flows for
voice, video, etc.). However, putting a device at each branch is a
costly and at partial best solution, and thus is generally out of
the question.
With Ipanema, the devices at the datacenter share information about
the sites they are sending traffic to, using a patented set of
mechanisms. This means that, in the case of competition for the
bandwidth of a remote site, they share the PE bandwidth exactly as
if all the flows were handled by one unique device. With Ipanema,
there is no need to deploy devices everywhere, even in the case of
such meshed topologies. Devices are only required at the sources of
traffic, in central sites.
Traditional WAN optimization solutions use TCP acceleration
mechanisms that need devices on both ends of the traffic path; in
the central sites and in the branches. Due to their configuration
complexity, acceleration is often deployed only on only a few sites
that exhibit serious performance problems.
However acceleration has potential benefits for all the sites in
the network. With a paired-devices solution it is rarely
economically possible to deliver the benefits of acceleration to
the whole network.
Ipanema's TCP acceleration can be performed with devices deployed
only at central sites, near the sources of traffic. Thanks to this
technology, enterprises can get the benefits of acceleration
globally over the network without necessarily putting devices
everywhere.
VISIBILITY: A KEY FEATURE IN STRATEGIC WAN OPTIMIZATION
DEPLOYMENTS
Enterprises should understand what is at stake before investing in
the global deployment of WAN optimization. They also need to
understand the impact of the technology on application and business
performance.
Most WAN optimization solutions have no performance related
visibility features: there is no built-in way to monitor the impact
of WAN optimization on the network and on end-user Quality of
Experience.
Ipanema's Visibility features include metrics that give an
unprecedented insight into network and application performance.
They accurately measure the performance of all application flows in
real-time and report on usage and performance throughout the
organization. Ipanema combines proactive and
reactive helpdesk functions via a bird’s-eye view of performance
(Ipanema Maps), application performance-based alarming, and drill
down. Ipanema's rich metric set also enables allocating
responsibilities between the WAN and IT domains.
1. Accurately measure the performance of all application flows in
real-time
The Ipanema system measures application flow quality metrics such
as TCP RTT (Round Trip Time), TCP SRT (Server Response Time) and
TCP Retransmits.
The system also computes unique “one-way” metrics that provide
the most accurate information available about the performance of
any protocol, including UDP flows used in VoIP and video. These
one-way metrics include one-way transit delay, loss and jitter.
One-way metrics are unique in their ability to represent what
network performance is actually delivered to end-users and to
precisely show the network contribution to application performance
from LAN to server and back.
Ipanema also provides two application quality indicators: MOS and
AQS. The MOS (Mean Opinion Score) is computed using ITU’s E-Model
and the AQS (Application Quality Score) is computed using an
Ipanema proprietary model based on the comparison of multiple
metrics with per-application thresholds. These indicators take the
drudgery out of assessing application performance by having the
system itself do the hard work of compiling network metrics into
the crucial judgments about how each application is doing. This
makes it possible to provide “at a glance” reports of
application quality for the whole network in realtime.
If the MOS and AQS reports show green, all the users the network
are receiving good application performance.
Application quality Application flow quality metrics indicators
(MOS and AQS)
2. Report on usage and performance across the organization
Ipanema’s reporting tools include numerous report templates that
can be instantiated over any time, application and topology
criteria. Templates include detailed technical reports for network
or site managers as well as summary reports to communicate the
network and application usage and
performance across the organization.
Reports are updated in real-time (hourly reports are updated every
minute) and are interactive: they are clickable with hyperlinks
between them to support drill-down and they include the flexibility
to go back in time using intuitive slider controls.
Figure 8: Report relevant information to all the network
stakeholders
3. Combine proactive and reactive helpdesk functions via
bird’s-eye
views of performance (Maps), alarming and drill down
The Ipanema system includes sophisticated tools to leverage its
unique ability to capture a global real-time view of networked
application performance. Flexible alarms can be set on any local or
global performance event. For example, a dedicated alarm can be
triggered when the MOS of a VoIP call is
going below a threshold at the company headquarters.
The system also features unique “bird’s-eye views” of
application performance using a compact, hierarchical, graphical
real-time representation called “Ipanema Maps”. In these Maps,
each rectangle represents a collection of application flows. The
size of the rectangle is proportional to the volume of data flowing
and the color (green/amber/red) represents the quality of the flows
(using/based on AQS or MOS of the flows).
Once a problem has been identified through an alarm or an amber/red
indication, the tool offers the ability to drill-down to an
“oscilloscope view” of a given flow which graphs all the basic
metrics of the flow and is updated every 10 seconds.
Figure 9: Globally troubleshoot application performance
globally
4. Allocate responsibilities between the WAN and IT domains
When troubleshooting application performance, it is important to be
able to identify the individual impact of each segment of the
application delivery chain.
Using its comprehensive set of metrics, Ipanema is able to isolate
the contribution to the performance of applications of the
following:
The WAN cloud as delivered by the service provider – these
indicators can be used to check the WAN’s readiness for new
demanding applications such as VoIP.
The WAN including the impact of WAN optimization measured from LAN
to LAN.
1. Ipanema delivers user-centric application SLAs
Traffic management should be aimed not only at maximizing, but also
guaranteeing the performance of applications running over the
network. Traffic management systems face the challenge of
delivering an appropriate service level to the different
applications and their users across the network. The Quality of
Service being delivered needs to be accurately monitored in order
to comply with IT governance initiatives or simply to report
network service levels across the organization.
With other WAN optimization solutions, the performance of the
service is based on low level technical metrics that do not provide
an efficient way to report the compliance of application
performance to defined standards over time. Furthermore, there is
no coordination or automation across devices to provide a level of
determinism in the delivery of WAN application performance.
Ipanema's objective based optimization and advanced visibility
functions are complemented by a complete application SLA framework.
Using Ipanema, enterprises can define the requirements of all
applications in terms of application performance. Baselines can be
defined using the meaningful, high level MOS and AQS indicators.
The system is able to continuously check performance of
applications against the baselines. The solution includes a number
of dedicated application SLA reports that can be
used to regularly report about SLA compliance and others that
enable SLA troubleshooting. Finally, the specifications of the
application SLAs are used as a direct input to the WAN optimization
parameters.
From the SLA specifications, the Ipanema System dynamically
computes the optimal parameters to achieve the SLAs in the varying
WAN usage conditions.
2. Ipanema enables “What if” application performance impact
analysis
SLAs set on WAN application performance can only be delivered when
enough bandwidth is provisional support the users of traditional
bandwidth sizing tools are unable to provide a sizing
recommendation that takes into account the actual performance
delivered to applications. The mechanisms used only take into
account the overall network usage against the link capacity. Such
usage based sizing is not effective as it can result in the
recommendation of bandwidth upgrades due solely to high usage from
non-critical application traffic. Sizing recommendations based on
usage are not optimal and result in a waste of network
resources.
Ipanema's patented bandwidth rightsizing mechanism delivers much
more accurate information. It analyzes, what the demand is coming
from each user for each application individually in real-time. It
is therefore able to differentiate peak demand on the line that is
generated by a critical application from
peaks that are generated by non-critical applications. Combined
with optimization, rightsizing enables the delivery of optimal
application performance with an optimal bandwidth sizing.
Rightsizing is the natural companion to application SLAs as it
closes the loop between the delivery of SLAs and the capacity of
the infrastructure.
3. Ipanema enables users to obtain smart network cost
calculations
Traditional visibility of traffic usage is a poor basis to charge
back network costs, and so has rarely been employed. Usage is not a
useful metric on its own. The network resources are normally
shared, and even with traffic management enabled, service levels
can often be degraded for some periods of time. A smart chargeback
model must take into account not just usage information, but also
delivered quality information.
Ipanema's application SLA framework contains a dedicated smart
chargeback module that aggregates information about usage,
delivered quality and application criticality. Business units,
users or application managers can now be charged back in a way that
everybody understands. This module is customizable and can be
integrated with any existing billing systems.
CONCLUSION
Most enterprises, especially large
ones, require WAN optimization benefits to be delivered globally to
consistently meet the quality of experience requirements of their
whole distributed workforce. However traditional WOC solutions are
faced with four challenges that prevent the delivery of WAN
optimization on a global scale: scalability, efficiency, management
costs and investment costs. Ipanema's next generation solution, the
Ipanema System, has been designed from the ground up for global WAN
optimization deployments. It provides unprecedented scalability,
enabling it to deliver WAN
optimization to thousands of sites. Ipanema’s objective based
management model makes WAN optimization management workload nearly
independent from the number of sites it is deployed over and
requires much less effort than do traditional approaches. The
inherent complexity of large and meshed
networks is fully handled by the system architecture. Finally, the
system supports innovative deployment options enabling the
deployment of devices only in provably required locations.
Visibility is fundamental to a strategic WAN optimization
deployment. Enterprises want to understand what is at stake first
before investing in a global deployment. They also need to
understand the impact of the technology on application and business
performance. Ipanema's Visibility provides a unique insight on
application performance to achieve these goals.
Ultimately, enterprises need to gain confidence in the behavior of
WAN optimization to focus on the business benefits this new
technology can provide to them. WAN optimization has great value on
its own for enterprise customers as this market matures, it is
becoming central to the efficient management of the IT
infrastructure in a distributed enterprise. Ipanema Network
Application Governance features provide - for the first time - a
direct ability to turn WAN Optimization into a strategic tool
within the enterprise.
