topic Re: Performance File-based Versus Server-based Worksharing in Revit Cloud Worksharing Forum

Performance File-based Versus Server-based Worksharing

EKey — Wed, 09 Nov 2016 16:18:21 GMT

We design our projects with file-based Revit models (central files sitting on network drive).

We noticed as many collaborators join the project as slow is Revit for every user. For example, a user can move/modify/delete elements faster if only two people work with the project file at once. If there are ten people working simultaneously Revit is much slower.

Is the observation correct and Revit has performance degradation on joining new collaborators?

A huge project is coming for us, and I am trying to find a solution which would help to run the Revit model faster.

How about using the Revit Server? Should it be faster in comparison with a central file model sitting on a network drive?

Re: Performance File-based Versus Server-based Worksharing

chubbard — Tue, 15 Nov 2016 17:44:35 GMT

You are generally correct that as more users access the central file your overall speed will slow.

In our experience the Revit server does help with this but not by much. Here is why;

Every time you run a command select an item, open a view, etc. Revit goes back to the central file to check for access of that item. It can only do that one item at a time so you get in line behind the rest of the users that clicked first. If any of these users are on a slow connection (VPN WAN) you will wait till the data travels back and forth to that user and then your request will go through. Revit server will improve that because the server does that across the WAN during downtime if there is any. BUT if 50 people are working the downtime is little so you still have some WAN issues.

If you are all on LAN, have good switching and throughput to your server, 1000t cards, you can help yourself by using worksets. If a users is working on L1 interior they check out L1 interior and the view workset they need. That reduces the checks for Revit on the central because it knows that user has access to ALL those items. Only when they cross boundaries does the check occur. So if you organize properly and enforce worksets etc. you will improve your access time. If using any slow connections look at Revit server to help but not eliminate wait times for access.

STC and reload will still be the same as always, and you will wait till any current transaction has completed before it starts.

Re: Performance File-based Versus Server-based Worksharing

EKey — Tue, 15 Nov 2016 20:05:37 GMT

Here is a response from the Autodesk. It also clarifies the question.

Hi Evgeny,

This is Martin with Autodesk Technical Support. Thank you for contacting Autodesk Technical Support. I will be assisting you with this technical support request case. My understanding of your case is that your preparing for an upcoming large project and considering using Revit Server over File-based workshared project for better performance. You asked if it should be faster in comparison with a central file model sitting on a network drive?

That's a great question. And in general, the answer is yes. Of course, when it comes to Revit performance, there are many factors and variables at play: File size, complexity, workstation configuration, server configuration (for file-based and RS), physical location, network conditions, etc.

As a general ballpark guideline, you may want to consider 5-6 the upper end of amount of users for a File-based project. More than that, it may be good to consider either Revit Server or better yet, Collaboration for Revit (C4R). So yes, In general, client-server worksharing is more robust and can handle more concurrency more effectively than file-based worksharing.

Using Workshare Monitor and each user avoiding SWC on top of each other is a best practice. In other words, each user should self-schedule their SWC activity to avoid syncing when they know another user is already doing so.

And Revit performance is of course also impacted by each teams’ or office’s workflows, standard practices, best practices, skill and efficiency of each user, file maintenance, etc. etc. So your millage may vary.

I hope that helps. Please let me know if that answers your inquiry of if further assistance is needed.

Best regards,
Martin Stewart
Autodesk Support Team

Re: Performance File-based Versus Server-based Worksharing

chubbard — Tue, 15 Nov 2016 20:12:32 GMT

While I agree with that, they are being somewhat vague.

The 5-6 is correct, you will experience some slowness with those. What they are not saying is that with Revit server (a single server) with more than 6 will have similar slowdowns. You really need to break up between multiple servers to really see the performance. Yes we have tested this numerous ways and with different server configurations.

C4R will eliminate the stacking effect of the multiple users but will introduce a potential bottleneck with your internet connection. All users will have local files and the accelerator on their workstation which mitigates the central file issue. BUT SCT and reloads will take much of your bandwidth so you have some issues there.

I still recommend (insist??) that you use the worksets correctly and enforce workset usage management etc. It is really the only way to truly manage a file with more than 6 people locally or WAN, server or file, C4R etc.

Re: Performance File-based Versus Server-based Worksharing

Anonymous — Thu, 22 Nov 2018 10:03:52 GMT

Hi Martin,

I picked up this old thread but kind of relates to the question we have.

We store our files on a NAS/ Synology drive and use file-based worksharing. Do you have recommended system requirements for this? If not, do you have system requirements for a windows based server? You do have recommneded requirements for Revit Server, but I don't think this is quite the same thing.

Thank you.