Lifeboat Exercise IT

How to Pave the Way for Fast Applications

Conversation around the network’s affect on application performance is getting very interesting. Recently, a key software analyst relayed that a handful of software vendors and customers provoked her with statements that "application performance improvement can't be addressed by networking vendors." But this sentiment needs to be examined more deeply, and it greatly depends on the perspective of the networking vendor. The fact is, applications need to traverse the network, and the path information takes can have as much to do with successful delivery as the actual vehicle.

If a networking vendor takes the time to meet with customers, and to understand an organization’s specific requirements, this trip can indeed be streamlined. What networking vendors need to understand is how user experience, ownership of applications, and ERP systems are being modernized by a combination of customer apps, IT, Oracle solutions, and networking solutions. Through a comprehensive understanding of these factors, the network can indeed deliver much more value than simply routing applications.

From a “time and processing” perspective, application performance is determined by the sum of time it takes for many disparate system components and their intra- and inter-service communications to complete a single user task or transaction. A long list of variables contributing to the “time and processing” equation includes but is not limited to: communications between the database, application drivers, application logic, application code execution, application presentation, application servers and web servers, network switches and routers, the client/browser/OS/PC processor, and human interface and interaction components, just to name a few…whew! As is, analysts, architects, and DBAs can get too caught up with the data, code and process design, information management, and messaging-side of the problem rather than truly considering the role the network and infrastructure can play in addressing user and system pain points.

Consider the auto industry and the Department of Transportation (DOT). Automakers do not tend to think that DOT efforts add any appreciable value to their products, but the DOT can have a huge impact on driver experience. DOT programs provide performance conditions, safety, and availability of roadways to all street-legal vehicles. While automakers might address performance issues by including Mercedes comforts, Lexus wiring, or Ferrari power, they’re still subject to the rules of the road. The problem is that user pain is caused by the combined deficiencies of these two intertwined industries. If the road is bumpy, congested, or in need of repair, it does not matter how much you invest in fine-tuning the machines—the ultimate experience of the driver or passenger depends on the quality and conditions of the roadway. Roads and road-enhancement programs exist to deliver the autos and their drivers to a destination, just as networks exist to deliver applications and to their users.

So how do you get your HOV and Super HOV lanes up and running for your users? One option is to turn to costly tools or even SI teams to do enterprise architecture performance and modernization evaluations. Instead, I suggest some grassroots team building by “lifeboating” the IT silos around individual application problems. This means throwing the key (and often finger-pointing) network, application, security, and regulatory teams into the deep-end of a problem-solving strategy session. Ask them to row home together with a single, comprehensive application modernization solution. And make them understand that they’re all in it together, and that their survival (as well as that of the business) depends on their ability to collaborate. My guess is if they make it back alive from this reality-show scenario, you’ll find they’ve been able to creatively solve the problem. At a minimum, they’ll have a better understanding each others’ silos and requirements, and will likely already be re-architecting their plans and systems to apply technology solutions that address everyone's needs. From a technology management standpoint, it’s also unlikely that they’ll return with plans to deploy tactical solutions from a variety of vendors, but instead will depend on a strategic solution from a best-of-breed vendor.

So perhaps in one sense, the aforementioned contacts bending the ear of the software analyst were right—network vendors might not be able to sufficiently address application performance issues, but a single vendor with a comprehensive solution very well could. Then the original claim that application performance can’t be addressed by networking vendors really comes down to perspective and semantics. In reality, application performance will continue to be affected by both applications and the network. While advanced networking may not fix application performance issues from a core intra-system code perspective, modernized applications and application delivery design can streamline the performance on the inter-system and user optimization-side of the equation. In my own experience at F5, I’ve found an increasingly positive correlation between successful application implementations and collaborative IT groups that look at tuning the network infrastructure and not just their “hotrod” application and database systems.

Proper architectures for scaling applications—while at the same time maintaining cost controls—are vital to IT data centers and the business customers they serve. And because packaged and custom applications are being deployed and consuming network resources in new ways, implementing an effective, flexible architecture is of increasing importance to CIOs and chief architects. As such, designing and deploying via “lifeboating” and laying a strong network foundation are solid best practices in today’s sink or swim/drive or crash financial and IT environments.