1. Direct attached storage is much cheaper than SAN storage.
2. It can be purchased incrementally as the VDI environment grows, instead of being an upfront cost.
Combine this with 40Gb or 100Gb Ethernet-or even teaming dual port or quad port 10Gb Ethernet NICs-and you can easily achieve 2x to 6x the bandwidth compared to a SAN with a 16 Gb FC connection.
Once storage and networking considerations have been taken care of, there are still some key considerations to keep in mind when implementing the VDI server environment.
- Get plenty of RAM. RAM is relatively cheap and can have a huge impact on response time and customer satisfaction.
- Design servers to scale horizontally, similar to how you would design a Web farm, so that additional servers can be added without a service disruption.
- Ensure servers are clustered for load balancing and availability.
- Separate application, configuration, and executable data to allow stateless operation and quick re-provisioning in the event of a VM failure.
Many systems administrators would rather purchase and play with shiny new hardware than invest in monitoring systems. This can be a huge mistake both in terms of end-user satisfaction and cost. A robust monitoring system lets the administrator see problems early before users are affected, and make adjustments in time to prevent problems.
VDI monitoring systems can also be a mechanism for reducing capital costs. With real-time visibility into resource performance and utilisation, the VDI infrastructure can safely operate at much higher utilisation rates. Higher levels of hardware utilisation use existing resources more efficiently and delay additional costly purchases. Critical areas to monitor in a VDI implementation include:
- Data centre virtual infrastructure performance, availability & capacity
- End-user experience
- Network performance & bandwidth usage
- VM sprawl management
VDI Best Practices & Procedures
Finally, don't forget to set up communications approaches and procedures at the start of a new VDI implementation. Regular communications with end-user departments can be the easiest way to avoid problems and maintain end-user satisfaction. Knowing when major workload events will occur allows the administrator time to adapt and adjust even better than the best monitoring system can.
Additionally, setting resource limits (how much storage is needed, how long are files kept, etc.) at the start of a project is easier than retroactively implementing them later.
The good news is that new technologies are allowing SMBs with limited budgets to deploy first-class VDI implementations without breaking their bank. Following a few best practices will keep all the stakeholders happy, including the finance officer.
Don Thomas Jacob is Head Geek at SolarWinds
Sign up for CIO Asia eNewsletters.