I’ve written a lot on the performance improvements gained by implementing RDP 8 on my RDPSoft corporate blog over the past 2 years. If you migrate to Windows Server 2012, or soon, Windows Server 2016, and set up RDP to use TCP and UDP as its transport protocols, you will solve a significant number of connection issues you had to contend while running RDP 7 on Windows Server 2008.
RDP 8 and Later – Network Quality Thresholds and Limitations
However, it should be noted that RDP 8 is NOT a panacea for ALL types of networks. Here are my general rules of thumb when it comes to the question of “is RDP good enough?”
1.) If the upper limit of latency between your remote desktop clients and servers is typically less than 300ms
2.) If the average packet loss rate between your remote desktop client and servers is at or less than 2%
then RDP 8 on a pure Remote Desktop Services implementation will probably work just fine. And yes, if there’s less latency but slightly more packet loss, RDP 8 may still be able to adapt decently. However, if your remote desktop connections are consistently outside the above thresholds based on your remote worker environments, you will need to consider other protocol technologies leveraged by Citrix. For more on those technologies (e.g Framehawk and ThinWire), please read this wonderful performance “bake off” test conducted by my fellow Microsoft MVP Marius Sandbu that he published on his blog.
Given the near ubiquity of decent quality broadband Internet connections in most major metropolitan areas, you can see why RDP 8 will work just fine for the majority of use cases. And this is why I maintain that for an increasing number of businesses, switching to “Pure RDS” over Citrix can save considerable sums of money in licensing, while not significantly impacting user experience (if implemented correctly). So if you are judging Remote Desktop Services by the connection quality you saw when running session hosts on Windows Server 2008, you need to quickly stand up a copy of Windows Server 2012 or Windows Server 2016 (Technical Preview) and fully test RDP 8. You will be amazed at the difference.
And even if you still have some “edge case problem child clients” – the smart CIOs and CTOs out there can save money by shrinking the size of their Citrix farm to only maintain licensing for the 5% or 10% of the workforce that has poor network connectivity, and deploying Citrix clients selectively to only the worker devices that truly need them.