Tuning Storage Performance In Virtualized Environments
In our last entry, Fixing Storage Performance In Virtualized Environment, we discussed some of the options to make the environment faster. What if you can't put in a new system? Maybe you just bought the one you have or maybe you have a feeling you can get more out of it. That's a good thought. Most storage infrastructures are underutilized from a performance perspective but you need to know how to mana
In our last entry, Fixing Storage Performance In Virtualized Environment, we discussed some of the options to make the environment faster. What if you can't put in a new system? Maybe you just bought the one you have or maybe you have a feeling you can get more out of it. That's a good thought. Most storage infrastructures are underutilized from a performance perspective but you need to know how to manage peak loads.Tuning storage performance in general means gathering data about your entire storage infrastructure and altering configurations to remove bottlenecks or better balance load. The tools that come with the components of that infrastructure will often provide you with some of the details you need to make those decisions. For example most switches will provide a list of the busiest hosts. Making sure all those hosts are not on the same switch or blade may help performance, making sure all those hosts are not sharing the same physical drives can help performance but it is a manual process.
Reality is that tuning the storage infrastructure with built in tools is hard to do, not because the tools are necessarily bad but because the tools in many cases don't provide an end to end view of how storage interacts with the storage network and the servers that will use it. Correlating the information can be time consuming, error prone and certainly not real time.
As we will discuss in our upcoming webinar "Addressing Storage Performance Challenges in a Virtualized Environment" this manual method is especially difficult in the virtualized server environment. Not only do we have physical servers attaching we also have virtual servers as well. We have to understand not only how those physical servers interact with each other and storage, we also have to understand how those physical servers interact with the virtual servers they play host to.
As we discuss in our article "End-To-End AND Top-To-Bottom Virtualization Management", tools that are focused on the virtual environment should be able to provide views that can correlate all the information that you need from a virtual angle from a physical angle and from a storage angle. You need to be able to see how virtual machines are interacting with each other and how they are using the storage infrastructure. In addition, beyond storage you need to know how they are consuming server memory and CPU resources since a shortage of those can cause performance problems for which storage often gets blamed for.
Tuning at this level means maximizing utilization of the entire storage infrastructure and the connecting hosts. This means eliminating the headroom that IT administrators often count on to handle peak loads. Maximum utilization of the storage network is also going to require a near constant monitoring system which essentially eliminates a manual spreadsheet driven strategy. We'll discuss real time monitoring in our next entry.
2014 Next-Gen WAN SurveyWhile 68% say demand for WAN bandwidth will increase, just 15% are in the process of bringing new services or more capacity online now. For 26%, cost is the problem. Enter vendors from Aryaka to Cisco to Pertino, all looking to use cloud to transform how IT delivers wide-area connectivity.
The UC Infrastructure TrapWorries about subpar networks tanking unified communications programs could be valid: Thirty-one percent of respondents have rolled capabilities out to less than 10% of users vs. 21% delivering UC to 76% or more. Is low uptake a result of strained infrastructures delivering poor performance?