One of the biggest issues that I have run into In the past is not using VMware's "linked clones" or Citrix MCS.
While you want to avoid "over engineering" a solution, you really want to get familiar with the target users of the solution, how they leverage it, what the application profile / footprint for those users are so you can tailor it to their expectations.
Performance is a tricky thing as it is very subjective, you might think performance is good, but that 1 second lag with a screen refresh can be seen as a performance issue with the user. etc) in a proof of concept will surface some issues either with the design, the product(s) that you are using in this design. In my experience, testing the most resource intensive actions (i.e. What are the typical challenges faced in such VDI setups in terms of performance, booting, peaks and failures? A common design is 2 data centers in each region where the majority of the users are located and splitting the VDI instances between the 2 data centers (for resilience / DR) This again would depend on how many users in each region and whether all the users in that region connect to the same infrastructure. Is this the best solution for a large global setup? etc), my suggestion is to do a lot of testing with each of these solutions to see which ones fit your immediate needs and can expand into future needs (i.e. There are a few key players in this space (Nutanix, Simplivity. For us HCI made sense and allowed us to expand, reduce our VDI environment and re-task esx hosts as needed to our Server Virt environment. We used to run traditional VDI (Blades with either SAN or Netapp NFS) and recently moved to HCI. * Do you have budget for expansion / on going support cost for a SAN / Storage infra?
* Do you have the technical staff to support a SAN / Storage infra? * Do you have SAN.etc storage in house that you can leverage for this? The dependencies are pretty straight forward though i.e. This would really depend (yeah I know not the best answer in the world). Is a converged platform from HPE, Dell or Cisco better or is it better to go with the traditional methodology of Blade / Storage infrastructure for a typical setup of around 2000 full VDI desktops?
0 kommentar(er)
