|3221 Built to Fail – Testing and Acceptance|
Today’s virtualized Data Centers are not stand-alone environments. The availability and performance of these environments depends on complex inter-dependencies of networking, server, virtualization and storage vendor components. Behaviors of which can vary by vendor or even firmware version.
Not only are end to end configuration guides that encompass every element in these infrastructures rare, but there are so many elements that one minor mis-configuration may result in your virtualized datacenter not performing deterministically under all conditions. The same rigorous testing that is required for Tier 1 systems and networks in the physical work must now be mapped into the virtual infrastructure.
In this session we will explore testing tools and methodologies to validate a virtualized environments ability to meet SLA’s, and also track performance changes over time as new hardware, firmware, and software is introduced.
**Special needs of Tier 1 applications**
Applications have a small window that you can lose packets before the application breaks. Is your convergence time under that? Voice apps require even stricter controls then this. Requirements include minimums among overall latency as well as the variation of latency as well as a maximum percentage of packet loss that can be sustained.
Quality Of Service –
Voice and video servers are starting to make their way into your virtualized infrastructures. There are specific prioritizations that need to occur at many physical and logical points in your infrastructure. Luckily these can be tested before you get a call about call quality issues.
Will that new app bring your cluster to its knees? Good thing to test first.
**Areas of increasing complexity in virtualized DC infrastructures**
What new components are showing up in our virtual infrastructure that can affect our SLA’s
- Networking Components
A single mis-configuration of bug in firmware of any one of these items can stop you from meeting your SLA’s. We will highlight common components that we need to test in Tier 1 virtualized infrastructures.
**QA and Acceptance testing methodology**
There is no need to reinvent the wheel when it comes to QA and Acceptance testing. We will discuss common tests such as SRLG (Single Resource Link Group) failure tests along with common testing and report frameworks.
**Test tools and procedures **
It doesn’t take a million dollar test set to validate your infrastructure. We will review some great open source tools out there that use can use to validate Availability, Quality of Service, and Throughput as well as show a couple of my favorites in action.