Johannesburg, 03 Aug 2005
With networks converging onto single enterprise entities, the importance of assigning enterprise applications as highest-priority traffic and delivering robust connections to these applications, is becoming an ever more troubling task.
To ensure data is kept as rapidly accessible as possible, protocols such as TCPIP with quality of service (QoS) must be deployed within the enterprise, says Andy Brauer, chief technology officer in Business Connexion`s Networks Competency.
"Actual traffic must be properly measured according to priority before deploying solutions that guarantee QoS," notes Brauer. "Vendors such as Cisco and Nortel offer solutions with AutoQoS features which provide enterprise traffic quality in line with set priorities."
What organisations need to do is to add better security and QoS with careful threshold and capacity analysis and then combine this with the ability to auto-escalate before congestion. Preventative steps can be taken from a network perspective to ensure applications are always available in a complex converged network. These steps include looking at the servers which provide the applications and ensuring there are mechanisms for failover to ensure uninterrupted application delivery, comments Brauer.
He explains that once a capable network, on which high-end applications will run, is deployed, organisations can explore a few approaches to tweak the infrastructure to deliver the maximum benefits from these business software solutions.
"For example, Business Connexion uses tools it has developed in-house which deliver measurements of how well the QoS mechanisms are performing - such as identifying business class applications that are dropping packets."
According to Brauer, this gives an indication of performance allowing for network tuning. Engineers can use sniffers, extract Netflow information from switches and routers, use simple network management protocol (SNMP) with Cisco network-based application recognition software probes or even remote monitoring and switch monitoring (RMON/SMON).
Brauer stresses that in order for this to be done correctly, an enterprise-wide full layer of correlated traffic flow analysis must be performed to make sense of the total dynamics of how a specific business operates.
"Since strange results often come out of the woodwork, perception is not sufficient and a bit of science is required to identify those applications and services that must be prioritised."
AutoQos aims to do this, however, there are still some issues which require human intervention and interpretation at present, he adds.
The main drive behind businesses deploying complex solutions in their business processes is to reduce labour and operational costs and increase productivity levels. The deployment of processors within the network (intelligent networks), application-specific integrated circuits (ASICs) and carrier-grade non-disruptive equipment is a trend aimed at achieving this goal for organisations.
Other trends include work being done on network-aware applications, neural intelligence, self-healing networks and next-generation routing. Because of the complex applications that run on converged networks it can be difficult at times to retrieve data from the network.
"IT executives are encouraged to review and balance network complexity, costs and performance requirements in the service level agreements (SLA) they enter into with service providers," concludes Brauer.
Share