Trimble Viewpoint One Vista VRL Cloud Network Standards and Testing
(How do I know if my network can handle hosting Vista using VRL in your cloud? What are your network standards?)
Author: Eric Vasbinder
IMPORTANT - STANDARDS AND TESTS ARE FOR VRL ONLY!
This document has network standards and specific tests for Vista Remote Link (VRL) ONLY.
Users of our RDP Published Apps (VEC/RDP) and RDP Desktop (VFC) connections to Vista in our cloud MUST follow the instructions and standards located HERE:
Network Standards for RDP Cloud Connections and Testing Them
CRITICAL REQUIREMENT - TURN OFF CLOUD FIREWALLS FOR VIEWPOINT CLOUD
These tests will NOT be valid if you are using NetSkope, zScaler, or any other cloud access security broker (a.k.a. Cloud Firewall).
These tools route all internet traffic through their external Internet cloud, thus causing all traffic to exit from ONE specific point into the Azure cloud. This will hide any potential latency issues and thus give you incorrect latency figures.
Please TURN OFF these tools temporarily during latency testing.
Vista VRL Cloud Network Standards
The requirements against which to validate are listed in Figure 1.
CRITICAL NOTE: LATENCY IS THE MOST IMPORTANT REQUIREMENT BY FAR.
The overall impacts of each level of network performance are:
* OPTIMAL: With network latency, packet loss, and Bandwidth at these levels, your Vista performance experience should be near to the same as your on-premise performance.
** GOOD: With network levels in this category, your Vista performance should be Good, but WILL be slower than on-premise for data entry, some report generation, and loading forms. NOTE: To increase form loading speed, please ensure you have Paging turned ON: Enabling Vista Paging.
*** ACCEPTABLE: With network levels in this range, your Vista client performance should be OK for most users, but your heavy data entry personnel and backoffice accounting staff may find the majority of client operations (NOT BATCH JOBS) to be measurably slower than on-premise. These users may require the use of Azure Virtual Desktop (AVD) to ensure their Vista client performance is suitable for their needs.
**** UNSUPPORTED: At this network level, all operations will be significantly slower than on-premise. Some users may find the usability improvements of VRL to be worth the speed hit, but many others will find the performance impact to be significant.
Figure 1: VRL Network Standards
^ Test using https://www.azurespeed.com
^^ Test using PSPing (Windows) or TCPing / MTR (Mac OS X and Linux)
^^^ PER USER BANDWIDTH (i.e. if you have 20 concurrent users, you'll need 20 x this amount) / Test bandwidth using https://www.speedtest.net
Why is Vista Over VRL in the Cloud Often Slower than On Premise?
When a customer is running Vista on-premise, the time it takes to get data back and forth between the client and the Vista server is quite small: usually less than 5ms, pointing to a server down the hall. However, when moving the server to the cloud, the physical distance increases significantly, sometimes over a thousand miles. This will slow things down. This is why latencies in anything other than the "Optimal" tier (shown above) will have an experience of slower end user performance as listed.
For more details, including an easy to understand analogy, please see the following Cloud FAQ article: Are there any times when VRL might be slower than RDP? Anything that stands out? Any workarounds?
Steps to Measure Network Performance Level:
NOTE: The measurements must occur from the customer’s network; not a 4G hotspot.
You MUST test network performance from the customer’s network where your heavy data entry users ("The 10 key experts") of Vista are located. For example, if your executive team is located in Philadelphia, but your AP clerks, Payroll clerks, and controller are located in Cincinnati, please ensure your results are for Cincinnati.
If you are on a mobile device, make certain to have a tool loaded to perform this testing (e.g. “Network Analyzer Pro” on Android or iOS, WinMTR and PSPing on Windows PCs, or MTR and TCPing on Mac OS X and Linux). Please ask Viewpoint Cloud Operations for help in this case.
Latency
Go to https://www.azurespeed.com and click on “Latency Test” in the left menu-bar. Make sure all regions are checked if possible.
If your company does NOT have data residency concerns, (E.g. U.S. based company who is comfortable with their data residing in Canada or visa versa) then we can evaluate regions in another country if those would result in significantly improved performance.
Please note that hosting your data outside of your country of operation is possible, it does require your full agreement and possibly the signing of a legal addendum to your contract to that effect.
Monitor the results to see which datacenter is listed as the closest to your location, and what the latency is for that region. Take a screenshot of the list of closest datacenters. See below for an example
NOTE: Latency tests can also be performed directly to your VRL reverse proxy server using PSPing, TCPing, and mtr by using the VRL server's DNS name instead of the names shown in the following section.
Packet Loss
Install PSPing if you are on Windows: https://docs.microsoft.com/en-us/sysinternals/downloads/pstools
If you are on Mac OS X or Linux, install TCPing and MTR using tools like Macports or your local package manager (e.g. apt-get).
Open the Windows command prompt by hitting the Windows-R key combination, then typing "CMD" in to the window that comes up.
NOTE: You may also open the command prompt by clicking on the Cortana search field on the bottom left of the windows task bar.
Then use the following command: >psping -t -4 REGIONTESTDNSNAME:443
Use the following Servers to Test Packet Loss to your preferred Azure Region Datacenter:
Stop the program after about 10 seconds by hitting Control-C
The results should look something like this:
The Packet loss statistic you need is in the second to the last line in the tool output.
NOTE: this tool will also show you the most accurate view of the average latency.
NOTE: Packet loss may be found using other tools than psping. E.g. WinMTR, mtr (mac / linux), Ping Plotter, Open Visual TraceRoute, TCPing, and others. Viewpoint Cloud Operations can help.
(OPTIONAL) The number of network hops can also be an important performance factor. If you can, please perform a traceroute to determine the physical route that packets take through the Internet in order to reach your preferred VRL Cloud datacenter. This usually requires software like WinMTR, mtr, Ping Plotter, or Open Visual TraceRoute. Viewpoint Cloud Operations can help.
Upload and Download
To test these two areas of network performance, please ensure you are using either the web site for https://www.speedtest.net, or the Speedtest app, from the local workstation(s) of actual end users that will be performing your daily, heavy usage Vista tasks (i.e. "The 10 Key Experts").
Summary
Latency is the most critical of the items to test above, as with higher latency, performance of Vista hosted in VRL Cloud will suffer. High packet loss can also result in poor user experience, as the network would need to “restart” requests frequently, resulting in lower performance as well.
If the latency is above 15ms, your performance will be slower than on-premise. As your latency increases to 60ms, it is increasingly likely that the performance in your environment would be inappropriate for our VRL Cloud. If this is the case, please investigate hiring a networking expert / consultant to review your network for performance problem spots. In addition, utilizing other mechanisms to modernize and democratize your workflows, such as Vista Web for time entry, are highly recommended.
Tags: speed test, speedtest, Azure speed test
changelog
Friday, 26 April 2024 at 11:56AM:
Added clarification around possibility that customer data, upon customer request and when customer does not have data residency concerns, can be hosted for performance reasons in another country if that would result in significant performance gains.
Wednesday, 18 October 2023 at 09:30AM:
Additional performance clarification
Tuesday, 12 September 2023 at 06:06PM:
Added East US 2 as an officially supported region; several others added as test points for under consideration.
Friday, 18 August 2023 at 08:29AM:
Added link to "why Vista VRL is slower".
Monday, 14 August 2023 at 10:46AM:
Cleaned up structure of overall page. Moved testing sites underneath table. Added column for near on-premise performance. Called out what users at each network level will likely experience. Added summary section.
Monday, 24 April 2023 at 01:34PM:
Added in requirement to disable CASBs / Cloud Firewalls during latency testing.
Wednesday, 08 June 2022 at 05:15PM:
Updated to show that two new datacenters are now available: West US in Fresno, CA and North Central in Chicago, IL.
Updated: Thursday, 08 July 2021 at 08:24PM