Internet accessed applications and even the “Internet of Things” has more network administrators considering their current strategy for cloud service monitoring. They are asking questions like: What do we need to monitor? How can we monitor the service? What do we do with the metrics?
What are we trying to Monitor?
When setting up a cloud service monitoring system, there is a bit of homework that should be done up front. First of all, what are the business applications that are being hosted in the cloud? Critical or not, they should all be included in the list. The next step is to prioritize their importance. We then need to find out if and how the critical cloud applications are being prioritized through the company’s network infrastructure. Are the applications prioritized in the routers/switches via port ranges, IP addresses or possibly NBAR2 recognized applications? You really need to find out.
What metrics do we need?
The next step is to determine what degradation in the application directly relates to poor end user experience. Most end users will say “slowness”. No surprise there but, how can we monitor for slowness? Is the slowness caused by jitter, packet loss, retransmits? TCP window size issues or round trip time (RTT)? How is RTT measured? Is it a metric representing the time to the server, the client or the response of the application? Some or all of these metrics can be very helpful when trying to ascertain whether a sluggish connection is caused by the Cloud Service provider, the Internet provider, the local network or the end user. Lets take a closer look at how we can gather these metrics.
Cloud Service Monitoring
If your network is Cisco powered, you are in luck because nearly all Cisco ISR and ASR routers can export Application Visibility and Control (AVC) details about the cloud applications cruising over your network. Take a look at this example report:
By filtering to include all of the business applications in the same report, we can quickly become aware of which applications are servicing the most end users, whether or not those applications are receiving priority and the average round trip time experienced per application. By digging deeper and filtering on an application like “Google-Services” we can narrow in on specific issues and examine individual connections from hosts which depend on the cloud service.
If you don’t have the luxury of a Cisco router that you can enable AVC on, you can obtain many of the same metrics by deploying a NetFlow probe. Not only do these appliances allow you to pin point where you want to sleuth the information, they also allow administrators to offload the potential overhead that can be introduced when enabling AVC on an already heavily taxed router.
However you go about setting up your cloud service monitoring solution, keep NetFlow or IPFIX in mind. These are the best technologies available today for staying on top of your dependency on cloud services.