Add a comment
In a previous post we wrote about network engineers and their ability to go days without sleep. There was no one transformative event that gave network engineers this super power. Network engineers simply adapted over time to shrinking budgets and smaller staffs, plus the big trends we discussed previously--data center consolidation, the increased number of remote users and the rise of VoIP and video traffic.
Here are three more things that keep engineers up at night:
1. Legacy Applications
"Chatty" applications designed to run over local area networks often do not perform well when deployed over the WAN. This has caused many companies to shy away from change, which isn't generally the best way to approach business. Check out "Jeremy Miller's "Death March" horror story about the challenges of rewriting a very large legacy application:
"One of the VB6 client applications in the larger system was a target for a tremendous amount of user and production support venom. I had all of the code available to me, so I decided to see if I could spot what the problem. I looked at all the forms and didn't find anything but some innocuous looking code. Then I accidentally opened the code for the splash screen(!) and out popped 10,000+ lines of VB6 code."
2. Software as a Service
More organizations are choosing applications, such as Salesforce.com, that are delivered over the Internet as a service, in place of internally-hosted, client-server applications. These services both require increased reliability and performance from the network. At the same time they increase the traffic as the application’s front-end is essentially downloaded each and every time the end-user connects to it.
Application services continue to find the way into new sectors, and according to Zoli Erdos, Program Chair of the Silicon Valley Association of Startup Entrepreneurs, they will continue to alter the way businesses look at IT for years to come:
"SaaS has already gained traction in number of application areas--such as payroll, human capital management, CRM, conferencing, procurement, logistics, information services, and e-commerce)--and should make gains across a much broader cross-section of applications over the next 3 years. Out of 34 application areas we have examined, only nine are unlikely to see some SaaS adoption over through 2008" [Zoli's Blog]
3. More Complex Applications, Including Service-Oriented Architectures (SOA)
Distributed architectures and the use of Web Services as a means to develop reusable software for rapid application delivery and easier maintenance introduce tons of network traffic between the various application tiers and infrastructure components. (Or, in other words, the server that used to do many things now does one thing and then passes the data on to the next server in the chain - like the evolution of increasingly specialized assembly line workers in Henry Ford's Model T factory.) Not only does this increase traffic, it increases complexity. Enterprise applications may have ten or more networked tiers with each tier representing a possible failure point or bottleneck:
"By encouraging widespread reuse of scattered software components, SOA threatens to transform the enterprise network into a complex, sprawling unmanageable mess. Left ungoverned, SOA could allow anyone anywhere to deploy a new service at any time, and invoke and orchestrate that service -- and thousands of others -- into ever more convoluted messaging patterns.
"In such an environment, coordinated application planning and optimization become fiendishly difficult. In addition, rogue services could spring up everywhere, passing themselves off as legitimate nodes and wreaking havoc on the delicate trust that underlies production SOA." [CIO Today]
Applications are the life-blood of any business, and the network and server infrastructure exists to ensure applications are delivered efficiently to meet the business’ needs. Too often, organizations use utilization as the basis to make infrastructure upgrade decisions. However, high utilization is only a problem if delay impacts application performance. The best indication of how applications are performing for the end user is to measure response times monitoring real traffic.
End-to-end response time measurements will also help clearly identify whether the problem is the application, server, or network. While it is true that, if the problem resides in the network, then utilization (and protocol traffic analysis) becomes critical. However, utilization should not be the basis of any network performance solution, because it is not a proxy for performance.
 |
