Employees of an enterprise are the direct customers of the IT department. With any customer/vendor relationship, the ultimate goal is to solve their client’s problem and make them happy. This conversation, between the end users and the IT department within the enterprise, happens around application performance. If there are issues with application performance management, the IT manager quickly hears this from his clients and is tasked to correct whatever the issue might be. With most business-critical applications rapidly moving to private and public clouds, the application performance management quickly moves to WAN performance management, as the WAN links are becoming the lifeblood of application performance management.
It is, however, easy to misdiagnose the real problem and therefore prescribe some incorrect medicine to the patient. Let’s go over some of the common traps that you may fall into as an IT person, and then look a little deeper to make sure the diagnosis is correct:
“We don’t have enough bandwidth, let’s upgrade our WAN networks” –
This is a very common antibiotics approach when some of the application performance suffers. Even though, antibiotics may provide you with potential quick relief, if you haven’t cured the illness, it is bound to come back and you may not have strong enough antibiotics to cure the illness next time around. What we are looking at here is a classic misconception that applications starving for bandwidth can only be solved by adding more of the same bandwidth. However, a deeper look at the problem may provide some insight into what may really be going on inside the patient. It is not uncommon that there might be some unwanted, or unimportant traffic, that is bringing the WAN network to its knees. If this is the case, introducing more bandwidth will let the unwanted traffic grab even more bandwidth and, thereby squeezing everything else that results in the application performance ending up right back where it was.
“We need to have an ISP that can provide us tighter SLAs” –
Another common misdiagnosis is to raise the stakes in the blame game. SLAs (Service Level Agreements) are essentially the penalty amount that the service provider agrees to pay in case of not meeting the promised performance. Raising the SLA level of your WAN network may very well make sense, however, frequently the SLA is confused with being the “cure-all medicine” for application performance problems. If the underlying WAN transport is not cured, adding more penalty when the blame game starts will not inherently solve any of the technical issues. Given that the penalty of not meeting the SLAs are minuscule compared to the cost of a downtime or poor performance window, it really is a painkiller, rather than a cure.
“Let’s move our network to an expensive private telco network such as MPLS” –
Although related to the SLA misconception, MPLS (Multi Protocol Label Switching) has the misconception of being more secure because it avoids the public Internet and only runs through the telco-managed layer2 infrastructure. Given recent NSA practices, I am not sure being in a major telco infrastructure can be claimed to be better protection than VPN encryption. Beyond the performance-related claims, MPLS networks are usually extremely expensive per mega bit delivered. It also creates a single-carrier failure point, whereby if the telco’s specific switching network has problems, your business-critical applications are not simply suffering from performance issues but are completely down.
To avoid these common misdiagnoses and associated mis-treatments, it is a good idea to investigate, measure, and track the traffic flows in your network with associated network metrics including average and peak throughput and metrics on latency and jitter. Once you have more detailed data on your network and the traffic flowing through it, advanced inbound QoS and advanced outbound QoS techniques will help tremendously in prioritizing, filtering and traffic shaping your WAN flows.
You can also look into investing in Broadband Bonding (next generation Load Balancing) and WAN Virtualization technologies and designs, as these will enable you to build a WAN architecture that is based on more than a single Telco transport. This will significantly increase the reliability and the uptime of your network.
It is crucial to make the right diagnosis before jumping to a prescription – knowledge is power.
Cahit Akin, CEO, Mushroom Networks, Inc.
Mushroom Networks is the provider of SD-WAN (Software Defined WAN) and NFV solutions capable of Broadband Bonding that enables self-healing WAN networks that route around network problems such as latency, jitter and packet loss.