Foglight Application Performance Monitoring
Since 2002 Foglight has featured a synthetic end user monitoring capability as part of the end user portfolio (FTR and FTR-A) but this capability must be updated to address more of today's problems. I've been visiting Foglight customers and prospects around the world to discuss their APM needs and also learn where their focus lies over the next year. The same 4 things come up over and over again: AJAX, cloud applications, third party providers, and mobile. All of these newer technologies impact the way applications and end users of those applications are monitored.
AJAX has challenges in a few areas. First, if left unattended, it will skew monitored page response times. This happens because traditional monitors think it’s an element of the parent page and they don't understand that it’s a separate interaction that caused the async event to fire. We have solved this in our end user solution by providing a means to differentiate the async bits from the sync bits. The second part of the problem is in understanding the action that caused the request to be fired. Fortunately a good synthetic solution will show this. A good synthetic solution clearly shows the user where the end user clicked and what the corresponding response time was.
Cloud Applications are a cause for concern because access to monitoring information is limited. IaaS(infrastructure as a service) isn't bad; you might just lose visibility to the metrics from the OS metrics and the network layer. PaaS(Platform as a service) is a bit worse because you lose visibility to the network layer, application layer, and maybe even the web layer. SaaS(Software as a Service) is the worse because you don't have access to the delivery systems at all. Fortunately synthetic transaction monitoring transcends all of these architectures and gives us the same look regardless of what deployment architecture the customer selects.
Third party providers just can't be monitored well with traditional APM solutions. Some vendors, including Dell, are looking to solve this with instrumentation but that still has a way to go. Synthetic transactions, because they are monitored right in the end user session, have complete visibility to the third party calls and can even tell you how long each one of them took. It’s advantageous to be able to tell when it’s a partner that's dragging your performance down. Some of the more mature organizations that we spoke with even suggested that they are responsible for performance of their content delivery and that of their partners. I'm not sure about dell.com but the quest.com site has at least 5 major partners that are serving up the bulk of the quest.com content.
Mobile is again a new issue. First, native apps can't be monitored well and second the web apps are all miniaturized. On the real user monitoring front this is a challenge but on the synthetic front this can be addressed in a few ways, from real device emulation to device type spoofing.
In the next month Dell will be announcing a synthetic monitoring capability that will address many of these new challenges. We expect that this new cartridge and the accompanying capabilities will help our customers to realize the full benefits of synthetic monitoring and help them to address many of their concerns with regards to these newer technologies.