1

I'm about to implement RDS (Remote Desktop Services Server using Windows Server 2016). Users will use an accountability software, no browser or any other app. I'm considering to put it as a remote app to make the user experience less complicated so they will not confuse remote desktop with local desktop. By now the max concurrent users will be just 7, users access from different cities.

I'll put the server on a datacenter so it can benefit for 24/7 accesibility.

I can put the server outsite my country for less money than on a local datacenter (almost half the price). I'm aware that latency is pretty important (even more than bandwith) for this kind of services so the user can have a better/smooth user experience.

The average time from my site to the external datacenter (outsite my country) is about 65ms.

The average time from my site to the local datacenter (same city) is about 10ms.

Of course, local latency is better than external in numbers. But in real life, this will make a huge difference in user experience? 65ms is not that bad in my opinion.

3
  • 2
    IME, 65 ms latency shouldn't impact the performance negatively.
    – joeqwerty
    Jul 12, 2019 at 23:15
  • @joeqwerty thanks for the comment, suppose that for any reason it goes to 100ms due to the location of the user, will it feel ok too? what is the latency that the user will start to experience a bad performance?
    – Yaazkal
    Jul 12, 2019 at 23:36
  • 100 ms is approaching the point where the users will notice performance degradation.
    – joeqwerty
    Jul 12, 2019 at 23:56

1 Answer 1

1

Ideally, you deploy in both sites at once and compare the support requests of "it is slow". While geographically distant sites are great for business continuity, that is a large expense you might not want to spend.

"Fast" is 100 ms, "keeps user attention" is 1000 ms. While 65 ms is within budget, that assumes the application remains snappy.

Take just one possible performance problem, network latency within the app. If an action takes 2 calls to a database also 65 ms away, that immediately feels slow. When all of the application components are close to each other, then you have more latency budget for RDP to the user.

Latency data is more important than system resource utilization for response time issues. Very recently RDS added User Input Delay counters to track input lag. Even better if the application is instrumented for detailed performance metrics. Possibly more work than you care to spend, but really fast applications watch response time and latency closely.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .