|By: Amichai Lesser|
Businesses today are highly dependent on distributedapplications to support every aspect of operations. If theseapplications under-perform for remote users or fail, losses ofproductivity, revenue and opportunity inevitably result. It isthus critical to ensure the consistent performance ofapplications across the network.
One of the gating factors controlling application performance isbandwidth. As more applications and services are activated onthe network, they contend for the finite available bandwidth.Bandwidth can be an especially critical factor for companieswith small or overseas locations that may not have high-capacitynetwork connections to the data center.
Typically, IT organizations approach this critical relationshipbetween application performance and bandwidth by managingsupply. This supply-side management approach is characterized byadding more bandwidth or implementing technologies thatprioritize use of the bandwidth that's currently available.
But IT organizations can no longer depend on supply-sidebandwidth management alone. Demand -- driven by moreapplications, higher volumes of data and increasing intensity ofuse -- is just growing too fast. Funding for technologyinfrastructure is growing too slowly. And the consequences ofservice interruptions are too great. In fact, supply-sidemanagement alone fails to address a variety of issues. Someapplications aren't very well designed for deployment on thenetwork, so they won't perform well, regardless of how muchbandwidth you throw at them. Some applications will perform abit better with more bandwidth, but those incrementalperformance gains aren't worth the cost of the additionalinfrastructure. In some cases, management needs to considerretiring an application altogether. In other cases, steps mustbe taken to reduce end-user demand.
Simply put, network managers have to do more than just managebandwidth supply. They have to apply best governance practicesto the consumption of bandwidth, so that utilization of networkresources is closely aligned with business drivers. Only byexercising this kind of governance can IT use its infrastructuredollars in the most effective possible way.
The Governance Lifecycle
Good bandwidth governance actually begins well before anapplication is deployed on the network. With the righttechnologies, developers can start assessing the behavior oftheir applications over the network early in the design anddevelopment stages. That way, they can resolve excessivebandwidth consumption or poor performance issues as soon as theyarise, rather than later in the game, when such problems can bevery costly to fix.
This kind of testing should continue right up to deployment, sothat there are no surprises when the application is rolled outonto the production network. It should also be done every timethe application is upgraded or modified, because subtle changesin code often have unexpected impact on the behavior ofapplications on the network. IT can apply these bandwidthgovernance best practices to applications that are already inproduction, too. For example, before throwing bandwidth at anapplication performance problem, network managers should firstmodel potential solutions to find out if the additionalbandwidth will, in fact, deliver expected improvements. What-ifscenarios should also be run to answer key governance questionssuch as "Will current bandwidth levels support the addition of20 users in our Atlanta office?" and "How will night shift usersbe affected if we start backing up remote servers over thenetwork at 2:00 AM?"
Only by answering these kinds of questions in advance cannetwork managers ensure that bandwidth is being used for thebest possible business purposes.
Bandwidth Governance Best Practices
To achieve best practices bandwidth governance, IT organizationsrequire technology capable of replicating the production networkenvironment as it exists today and as it might look tomorrow.This "virtual enterprise" should be capable of assimilating allthe factors that impact application performance in the realworld: live applications, the data center that supports them,the topology and bandwidth constraints of the network, thenumber of distribution of end users, etc.
By leveraging this virtual environment, everyone involved withbandwidth governance -- from application designers and QA staffto network managers and architects -- can more effectivelycontrol bandwidth utilization and preempt potential consumptionand performance problems. They can also verify the effectivenessof any planned supply-side measures, such as QoS and bandwidthgrooming, they plan to implement in production.
Unfortunately, most IT organizations rely only on developmentLANs (which don't reflect conditions on real-world enterprisenetworks) or mathematical simulations to assess the behavior ofapplications. These resources are useful, but don't provide theprecision or flexibility necessary for the kind of truebandwidth governance IT will have to implement if it is going tomaximize returns on development and infrastructure investments.
That's why it's essential that IT organizations re-evaluatetheir bandwidth management strategies and their technologyportfolios. Those that continue to manage application networkperformance in one silo and application development in anotherwon't be able to govern bandwidth effectively across theapplication lifecycle. Only with an accurate, flexible andproactive approach can IT bridge the gap between development andproduction, and thereby meet its goals of reliable performance,cost-efficient service delivery, and tight alignment ofexpenditures with business priorities.
To learn more, visit www.shunra.com. Shunra empowers enterpriseorganizations and technology vendors to eliminate the risksassociated with rolling out complex, distributed, applicationsand services. The Shunra Virtual Enterprise (Shunra VE) solutionprovides accurate, highly granular insight into how networkedapplications will function, perform and scale for remoteend-users. It creates an exact replica of the production networkenvironment, allowing users to safely develop, test andexperiment with applications and infrastructure in a labenvironment before deployment in production.