It has been clear for quite some time that the network has become the lifeblood of nearly all enterprises. This is not just true for obvious network-centric enterprises such as retail sites or content distributors, but also for enterprises that utilize distributed applications such as SAP, Oracle, or any number of other network-centric applications.
Over the past few years, a number of new enterprise technologies have emerged that are critically reliant on network performance, but where this reliance is not necessarily obvious. These technologies include enterprise-class Voice over IP (VOIP) telephony solutions, Virtual Desktop Infrastructure (VDI) solutions, and enterprise collaboration tools.
While the vulnerability of "classical" distributed applications to network performance issues are well-understood, it is quite a different matter for a VDI session to momentarily "freeze", or for the CEO's VOIP call to get disrupted due to network issues. In short, these issues are far more visible than those associated with classical distributed applications, and as technologies such as software-defined networks (SDN) and hybrid private-public enterprise clouds become more prevalent, these issues are likely to become more rather than less pronounced.
So what do IT departments need to ensure that they can provide the levels of performance from these new technologies that users expect?
The most obvious answer is that they need to know what is really going on in their networks. While this sounds trite, it is far more difficult than one might expect. Causative factors such as microbursts, timeouts, and protocol errors can be difficult to detect with conventional application performance tools, and tying these causative events to the specific "new technology" outages can be even harder.
Given that many of these causative factors can be intermittent in nature certainly doesn't help. This is one of the primary reasons that many enterprises have introduced dedicated "network visibility fabrics" that provide instrumentation at key points in the network, exposing the full set of network packets and flow data that underlie these causative issues. While network visibility fabrics do not prevent these issues from occurring, they do speed the ability to resolve issues, which helps to avoid "outages" of network-centric technologies such as VDI, VoIP, network collaboration tools, and SDN frameworks.
As with most human endeavors, one of the best practices for making good decisions is to have the right data. Even the best decision-making processes can lead to wrong decisions by not having the right data. As networks (and the applications that depend on them) become more complex and carry more types of data, it becomes imperative to have the right data to avoid making guesses as to what is causing network issues. Look to see more enterprises implementing network visibility fabrics as dense 10Gb Ethernet networks become more prevalent, and more enterprises start to deploy these new technologies.
Thanks to APM Digest for the article.