Central Model performance

[b_weinreder]

Currently we're working in a central model with 3 people. We all made our own non-editable workset, meaning that every view and family also populates its own workset. We're encountering some performance issues as the file is approaching 2 GB (we're working on a gigabit network). The majority of the model still exists in Workset 1, and everything created in the model after that point belongs to the individual worksets.

What would be the performance benefit, if any, of making the individual worksets editable? Would it delegate more families and views to fewer worksets, making it easier for Revit to keep track of these? I'm genuinely curious, since we're not using the worksets like almost everyone else would, it's been working well for us so far, but any suggestion would be welcome.

[b_weinreder]

Forgot to mention that this workflow works for us if we sync occasionally (say every hour on average), or risk getting a message that the element we want to edit is out of sync with the central model.

[hmunsell]

I'd start by looking at your File Size. "I" would consider 2Gig a massively oversized model. BUT, depending on what it is, it may be in line with that type of content. For a residential House, that would be huge; for a 100-story high-rise, maybe not. 

it sounds like you may need a good old model cleanup; start by making a copy of the model so you don't lose anything, then...

• clean up your Model & Detail Groups
  • I can't tell you how many times I see a model with 100 or more model/detail groups, and most are only used once. if it's only used once, drop it and delete the group. 
• get rid of any unneeded views and sheets
• Purge Unused families
• Compact Central Model
• Do an Audit

Are you working with Scope Boxes? using scope boxes can help speed things up, too. 

[Simon_Weel]

Where's the local file stored? 

[Tom_Kunsman]

First, everyone should be syncing with central at least every 30 minutes, if not more often. 

I am not sure how having non-editable workset would create worksets for each view and family. How many worksets does your model have right now?

Everyone can work within the same workset, it is just important to know who is going to work on what item, have people communicate, and not have everyone work in the same area. Worksets are more "containers" for the information, and not necessarily for each person to have their own.

Making the workset editable is NOT recommended. This only locks everyone else out, causes more things for revit to track. It is the worst possibly named dialog box in the whole program. 

Ideally, having a workset for each linked model, linked cad file, one for grids and levels, and one for the main discipline is all you need. 

[b_weinreder]

We use scope boxes.

This model contains our catalog of residential floor plans, currently running up to about 10 models that we can use as a jump-off point, some of them with a few spin-offs. We've also started developing 2 projects from the same model, to pilot a few family- and data changes, one of them plays a huge part in communicating with a configurator so that clients can select options from a catalog in a webbased environment, which can then be automatically applied to the Revit model, Each option (and typically a dozen combinations with other options) is also stored in this central model, which according to my last count now contains about 1700 distinct model groups.

I can hear you think "well, there's your problem", and you would be right. But there are no good alternatives to keeping the 2 pilot projects up to date with changes in our families and floor plans. So there's that. The real performance load comes from actually working with these models, which more and more frequently has some lag, especially since 2 maintainers have been added to it and another one in the foreseeable future.

So, the main question remains whether a different approach to workset management can reduce the overhead a little. Let's say we put everything pertaining to the catalog in its own workset, and everything pertaining to the 2 different pilot projects. If we make them editable then they will be checked out by their respective users, meaning they check out the Entire workset, to which nobody else will have access. Will this make any difference?

[b_weinreder]

I hear ya.

If you go to the Worksets dialog and check the boxes for Families and Views, then you'll see what I mean.

Basically, you're saying that fewer (non-editable) worksets is better, and we already have our optimal solution.

[b_weinreder]

In my case, locally on an SSD. But I'll ask a coworker who's more commonly complaining.

[hmunsell]

WOW... I mean, W O W...  I'm not saying that in a critical way, more out of sheer amazement at the workflow. 

It looks like you have your hands tied to a large extent. Turning off unneeded worksets should help. A while back, we had a project in our office with 5 buildings and around 300 3D CAD links throughout the buildings. Due to the client's requirements/demands, there was no way around it. We heavily used worksets and scope boxes to manage the files. 

when you open a file, I assume you open it with only specific worksets available? we found this helped us with the file performance. 

[SteveKStafford]

The only workset type we specifically control are user created worksets. The other three View, Family and Project Standards are created and managed by Revit. They are created when we make things like views or add a family to the project. We borrow from them when we edit the properties of a view, family, dimensions, text etc..

You would likely benefit from more user created worksets and assign all model elements for a specific design to an individual workset. This way you (and others) can open the project and specify which design(s) to open or close. The data is still part of the model environment but Revit doesn't spend any time displaying those elements. It also doesn't spend any time worrying about them unless you do something that affects the elements assigned to the workset that is closed. Elements that are assigned to different worksets and are in the same place will generate warnings so it is not a strategy for managing different designs that have to coexist in the same location.

The more users the more user data associated with the activities of the team the model will accumulate. Using Compact Central will help reduce internal database inefficiency (think of it like defragging a hard drive, moving bits closer together for faster access).

Look at the accumulated warnings in the model and resolve as many as you can, the closer to zero you are the better. If you have less than one hundred you're doing pretty will given the size of what you describe. My priority for resolving warnings is those that affect:

• Affect calculations, rooms/areas & wall/boundary errors
• Duplication (multiple elements in the same place)
• Affect documentation - Type Marks/Marks/tagging accuracy

Long term I'd be looking closely at a way to break this single project up into logical separate projects. You can pass common families and other design elements between multiple projects so you're using the same library of elements. As this project file grows it isn't going to see better performance...if you're lucky it will just stay manageable and not get worse.

[casandraarmeanu]

Hi @b_weinreder the best way to improve the perfromance is to open only the worksets you need and keep everything else on a closed workset, worksets are usually split by discipline, volumes/group of levels, or type of linked files. If you keep the workset close when opening the model, Revit will not process the data the workset contains, and will consume less RAM and disk space. 

Keep in mind that if you close a workset after the model is opened, the data will still be kept in memory, which can slow down Revit, only when closing Revit the memory is cleared. 

If you model is several years old, you can also try to detach from the central model and re-save as a new central model (better with a different name). This deletes the save history (which can get very large for models spanning on multiple years) and generates only a single partition in the model, which is processed faster by Revit.

[TripleM-Dev.net]

2GB file is massive, even with a Gigabit network.

As long as the locals are stored on the local hard drive (ssd I read) and the changes aren't continues large the sync's should be ok, and the saving before and after the sync to the local disk should also be fast.

Just wondering, what's the performance in time for the 3 steps, save-sync-save?

Some considerations:

- When making a new local, what's it size in relation to the central.

- Any complex families? (try to export all families, and see if any stick out in size)

- Families in place? (even in groups, they can really blow up a model)

- Revit warnings (can cause delays if it's many, try to solve them)

- Constraints in project file (calculation time during editing / sync)

- Model distance from internal origin?

- Any dwg's in views visible (can cause navigating views, not related to opening/sync times)

- A lot of dwg in model can increase size (linked or imported)

- ModelText can also blow up the filesize if used a lot.

- etc...

As far as worksets considered, opening less worksets can help depending the project and how it's setup.

I would make worksets by building not by who's modelling, less editing workset afterwards and all relevant elements are then present.

Split the building in each won Workset, Building_A, Building_B.

If it's possible and no direct connection is to it, it can be split further by exterior/interior, and if there are more disciplines in a model split it by those also.

- Michel