BANDWIDTH: That Dimming, Thick Lump

Scenario 1

A world-renowned white goods manufacturer required more than 99 percent uptime during working hours as well as very good performance to meet users’ expectations. To achieve this, the company put in place a heterogeneous and fallback network using VSATs, leased circuits and ISDNs (with STD dialing). Still, it faced problems regarding: 

• Heavy dependence on the VSAT network to maintain high network uptime
• Bandwidth clogging
• Huge investments on VSAT bandwidth and manpower at different locations
• No provision for voice over the network due to high costs 
• Existing leased line network being used as secondary medium of communication because of poor reliability and performance 

Even after investing a huge amount in its network infrastructure, users at the company always had complaints of the poor response time on the network.

Scenario 2

A year later, the same company was able to ensure 40 percent reduction in its connectivity costs while providing excellent response time to its users. It was able to achieve the following:

• Very high uptime running their mission critical applications 
• Improved performance of terrestrial network by fine-tuning networking parameters and keeping the errors under control
• Network intelligence available on traffic
• Ability to do intelligent capacity planning
• Better than committed uptime
• Significant bandwidth savings–due to better utilization of existing bandwidth
• Cost savings on VSAT bandwidth by moving away from their predominant VSAT network

The company was able to do intelligent bandwidth optimization!

The above example, taken from a case study, depicts that it takes more than just investments to ensure uptime. It takes business intelligence! Today, a CIO is most worried about: 

• Unhappy customers due to slow app response time 
• Mission-critical applications taking a backseat to give space to non-critical data like e-mail and Internet surfing
• Employees hogging bandwidth for surfing while the customer-centric applications are suffering

With virtually no network visibility, it becomes difficult to analyze the root cause for slow application response time. The easiest way out seems to be to add on more bandwidth, a solution that might bring temporary gains in business application performance but also augment your customer access costs. And then starts the spiral of poor-performance, increase-in-bandwidth, higher costs, adding an incremental cost to the CIO budget every time, as indicated in the example given above. 

The right approach in this case is to optimize bandwidth by analyzing the cause of the choke. Today, customized tools are available that provide complete visibility into the network, indicating where the choke is happening, and which application or user is using most of the bandwidth and for what purposes. Through the drill-down analysis of bandwidth utilization patterns, a CIO can become aware of the traffic patterns generated by various applications on the network at various locations. This helps in understanding the real cause of performance degradation and user behaviors during the day. 

After knowing the usage patterns, bandwidth can be re-allocated between processes to arrive at an optimal distribution. There may be excess bandwidth allocated to a particular user group or application, whereas another group or application may be starved of bandwidth for critical operations. By deploying policies on the applications and prioritizing mission-critical applications on the network, an organization can enhance its network performance. 

The CIO should know, at a click of a mouse:

• Where is the choke happening–Status of WAN/LAN links and ports, traffic volumes being generated in volume links, capacity utilization of various links, percentage of uptime for each link, proactive alerts to be sent through SMS, pager, and e-mail as soon as the link goes down
• Why is the choke happening–Is it because the number of speakers has gone up, or because newer applications are generating more loads on the network, or because bandwidth is being utilized in an unproductive manner
• Who is using how much bandwidth–Application wise, customer wise, division wise, and even user wise drill-down analysis of bandwidth utilization patterns enabling capacity planning
• When is the bandwidth being utilized the most–Hourly, weekly, monthly utilization patterns detailing the time of maximum and minimum usage
• Which applications are getting bandwidth priority–Reports on outbound queue lengths detailing which applications are getting bandwidth priority

It is through answers to the above that long-term bandwidth optimization can be achieved. 

CRD Prasad, V-P (professional services) HCL Comnet