Bug 🐛: Cloud credits are not integrated well. Reconsidering the Removal of Local Simulation in Fusion 360.

[asheepcalledgeorge]

Cloud credits are not integrated into Fusion well enough yet. Our productivity is suffering as a result, possibly to a cost higher than the credits themselves.

Our use-case is commercial and small-scale, but many others in the forum have their own considerations, such as to teach and more.

Reference: https://www.autodesk.com/support/technical/article/caas/sfdcarticles/sfdcarticles/Updates-to-the-Fus...

I have emailed the team separately to discuss, and they were very professional and helpful.  However, today's experience is reported as a bug since the workflow is crippling us.  I am a big promoter of Fusion in our business, and the Solidworks crew are starting to make dents in our arguments despite their pricing.

What Happened Today:

My team and I are evaluating design ideas for a part that needs to be changed on a deadline... A common use-case across the industry 🙂  This is our experience:

1. Wait 20 minutes to find out that a very static stress simulation (that would take 2 minutes locally) failed due to a meshing error.
2. Fix meshing error (1 min) then resubmit.  Wait another 10 minutes.  Get results. Refine.
3. Repeat iteration around 3 more times.  Wait 10-20 minutes. Get results. Refine.
4. Decide we need a modal frequency analysis.  Configure the study.  Submit.
5. Cannot submit because cloud credits needed.  Hm ok, but fair enough.
6. Spend another 15 minutes trying to work how/where to buy cloud credits.
7. After Googling - turns out Cloud Credits are actually now called Tokens.  No help available on Autodesk website that we could see.
8. There is a small button in the app called "Buy Tokens" inside a Preferences dialog (????).  I would expect this to be a button in the simulation workspace, but ok, whatever.
9. The link to "Buy Tokens" takes me to an Autodesk webpage.  I am not in the US, so click the button to take me to our local webpage.  This navigates me away from the page I need to buy the tokens.  I explore the local site for another 2 minutes before getting lost. Cannot find a way to buy the tokens on the UK site.
10. Eventually decide to stay on the US site, get accidentally sent back to the UK site a couple of times.  Eventually manage to see the price list.
11. Fall off chair in shock at token pricing model. Would love to have been in the pricing meeting to hear rationale.  Microtransactions > chunks of team budget.  Minimum spend is $300 USD to do a single study?  Do they expire?  Can they be shared between our org?  We have no idea.  Still no actual confirmation Tokens are Cloud Credits from the Autodesk website.
12. Walk to our admin department. Make case that we need more budget for the Tokens. 
13. Purchase made.  Phew.  Go back to Fusion.  Deadline looming.
14. Fusion is still not showing that we have available cloud credits / Tokens.  Purchase has not updated in the app.  Panic that Cloud Credits are not Tokens?  
15. Wait 10 minutes.  No luck.
16. Save all our work. Restart Fusion. No luck.
17. Wait 1 hour, still not showing as updated.
18. A Fusion server outage later that afternoon meant that my team lost more money waiting than the cost of the cloud credits.  This is frustrating.

We were unable to complete our work today.  Furthermore, the static stress studies we were able to use, were significantly slower.  I expect that on "light use" CAD days, we will loose 30 minutes per person per day.

Impact / Reflection:

• Hardware Investment: My team and I have heavily invested in powerful hardware to optimize our simulation results. By eliminating the local simulation option, you are essentially rendering our investment futile.  Competitors are not.
• Increased Wait Time: Cloud simulations increase wait times. With local simulations, we can complete simulations in a timely manner. This directly impact our productivity.
• Frequent Iterations: Design engineers require multiple iterations to refine and optimise a product. The current cloud-based approach slows down this iterative process.
• Cost Efficiency: Our time costs money, and the increase in waiting times leads to a loss of productivity and wasted expenses.  This is to say nothing of the time wasted unsuccessfully trying to actually buy and use cloud credits.
• Reliability Concerns: Local simulations offer more reliability and control over the process, whereas cloud-based simulations may fail due to unforeseen issues such as server crashes or connectivity problems.  Especially for simple static stress analysis - often the "first port of call" when evaluating a new model.
• Poor Integration leads to Frustration: The cloud credits alternative is poorly integrated into Fusion 360.  A core workflow component should not be buried inside a preferences dialog.  The user journey is more of an adventure than a positive pathway; forcing engineers to context switch between designing / evaluating / and begging the admin department for irregular blocks of money (rather than microtransactions, or a credit line).

Proposed Solution:

1. Integrate cloud credits into the Fusion 360 product more effectively.  Spend time on the workflow, and UI/UX issues that stem from the app-to-web user journey.  It's a wrenching context switch. This is NOT our preferred solution, but in our view, is essential in either case.  Specific focuses are: usability, in-app purchases, credit sharing, and server failures.  Please consider pay-as-you go, or a line of credit.
   
   
2. Reconsider the removal of local simulation, especially for static stress studies.  In addition to the impact listed above, static stress analysis (local, for free) is a gateway drug for most people into the other more advanced studies.  By removing this option, Fusion is a weaker product than it needs to be, and seems to do more commercial harm than good by driving customers, such as us, toward other products.

Fusion sells because it is has some features that are "too good".  We like that.  Keep a few of the gems around. They blow us away and turn us from users into champions that give you the userbase that you have today.

p.s. none of the above is meant as criticism, but only as sincere constructive feedback.  We love Fusion and are trying to help you make it a better product 🙂 

[asheepcalledgeorge]

p.s. @keqingsong, please can I have the money back for the cloud credits?

[leganzish]

Very well said!  I couldn't agree more!  I am sorry for your frustration with cloud credits and hope you get a resolution.  

Your Point #2 of your Proposed Solutions is critical to me.  Their decision to force that basic form of FEA to the cloud is a bad business move.  They won a lot of customers by including/integrating FEA with the subscription.  It's  understandable to upcharge for more advanced studies.  But like you said, linear static is a gateway drug.  Removing local seems like a bad business decision driving customers away.

Like you, the outages and reduced speed of cloud solves had a massive negative impact on my productivity the past 2 days, not only costing the company more, but losing us a massive business opportunity.  Simply unacceptable for a business tool.  I'm evaluating if a change in CAD systems is right for us  because of this.  I'm sure a chunk of their user base is doing the same.  

I hope someone from AD will answer and provide some insight to this:  Why are cloud solves so much slower than local solves?  I would imagine the resources are optimized more than my work laptop.  Of course data transfer up+down adds some time, but I noticed the actual solve step took close to 5 minutes per iteration today on a simple study I estimate would have taken 30 seconds or less on my laptop.  I was hoping to iterate quickly and as you described, I had to sit and wait or switch gears into another task while it solved.

This was my first use of cloud solve, and I expected it to be faster or the same.  Between the slowed productivity plus lost productivity of the outages, it was not a good first impression!  If I do stick with Fusion in the future I suspect I will only run studies when absolutely necessary.  They've really turned me off from it.  In the past I've loved how quickly you can setup and solve a simple study to validate a design or even do a sanity check early in the design process before refining and fully modelling.  It was an effective workflow.  I guess I'll go back my pre-fusion workflows.  

[asheepcalledgeorge]

Thanks 🙏 I'm sorry to hear about your experience too.

I was reflecting on this a little more:

• Recent outages: I think these are likely the exception rather than the rule, but it does highlight weaknesses in the SaaS model.  Why have an offline mode at all? (👀 looks at all the people without rapid internet connections).
  
  
• Token model:  I think most of us are on board with the principle, but the pricing strategy seems misaligned with the needs / demographics of the Fusion userbase.  If a microtransaction/pay-per-use option is too hard to manage/implement, I would advocate a pay-as-you-go approach that riffs off the £10, £20, £50, $100 top-ups with early mobile phone contracts. Most of us changed to a mobile contract when the deal became better and our spend was higher!
  
  As it stands today, buying big "blocks" of tokens has no "real life" analogy.  Please lower the barrier to entry.  Tokens are not something we intrinsically know how to value. People fear what they don't know. 
  Alternatively, considering Tokens as Time might make more sense.  For example, I rent 10 hours of time on the Autodesk cloud is easy to understand.  I can spend that however I like; fast little static stress sims, or complex harmonics, rendering, etc.  For millennials like me, you could make the argument that you have CPU/GPU optimisations that make it more environmentally friendly than on my own computer 🌿🌍!
  
  
• Integration and user journey:  I think this is essential to get right, and at the moment it is bad.  If tokens become a "virtual currency" with the Autodesk ecosystem, then we as users need to know what is in our wallet.  It needs to be easy to buy them and spend them. We need to know the value of the token, and be able to attribute it to value in the product (e.g. time, simulations, etc).
  
  
• Cloud simulation performance: We all know we still have free linear static on the cloud, but it's not fast enough. Quick design iteration with the simulation is non-negotiable for a lot of us.  I would PAY for linear static stress if I could simulate the entire model and get the results faster.  Please don't make me though.  It seems faster, cheaper, less bandwidth-y, and more environmentally friendly to give me the choice - at least for now.
  
• Fusion is a SaaS platform and we want to invest in scale rather than your OS / hardware issues.  Understandable.  However, if we are honest with each other , the architecture you have today cannot compete with local for linear static stress, and customers are suffering.  Please can you share a timeline on when you think cloud will outperform local? 
  
  From what I understand of solvers like Nastran/ALGOR the linear static stress is performant enough on standard x86 stuff anyway so would make the case for its reinstatement in the meantime.  I would hope it is also relatively simple to deploy and maintain, since you already have the implementation.  And hey, you can always limit your hardware targets if it's really such a big deal.

Have a good weekend!

[leganzish]

• Recent outages: I think these are likely the exception rather than the rule, but it does highlight weaknesses in the SaaS model.  Why have an offline mode at all? (👀 looks at all the people without rapid internet connections).

---

That is the part that kills me!  I'm sure they have 98% or greater up time.  Unfortunately without any backup option, that 2% is enough to cost a lot of business a lot of money.  I just had bad timing and bad luck, but it was eye opening and now I must protect my business.  I can no longer trust another business to do it.

[asheepcalledgeorge]

I just wanted to bump this so the conversation is not lost/unaddressed.

[FrodoLoggins]

Yes they do expire lol. Only good for one year and then poof I guess you shouldn't have paid us so much money now you owe us more 🙂

[asheepcalledgeorge]

Hello again, I was just hoping to get an acknowledgement or response to the thoughts in this thread.  Thank you!

[gerrit.agel]

I can just approve of this post. Th decision to remove local simulations is costing us so much time and causing immense frustration! I already wrote an Email to the Autodesk Team, but no response. If you want to keep your loyal customers, this is not the way to do it! Bad enough, all the valuable data has to be in a cloud.

[asheepcalledgeorge]

Unfortunately another outage means work is paused today.

I would really appreciate a response from Autodesk on the points raised above regarding:

* Integration of cloud credits

* Slow local simulation workflow

[doug3XXFT]

Hi all,

Adding another bump and 'same problem here'. Exact same process and outcome on this end where I was the sole Fusion 360 user and had been for years until I finally won the rest of the team over when they saw the great benefits of the overall package, especially Mac support and rapid integrated simulation (we also have Comsol, which has become nearly obsolete since full adoption of Fusion as the ability for anyone on the team to run FEA simulations quickly, and with far less of a learning curve directly in Fusion was an absolute game changer). Roll on 6 months and the switch to cloud credit only simulation became crippling. The cost is almost the same as keeping Comsol running (and that's including the electricity bill to feed the processors!!). 

The decision to remove local simulation is a joke. I have also emailed support on the topic, asking for clarification on various points, to ask them to explain their rationale and to receive some communication on the future roadmap of the cloud based system.... surprisingly, absolutely no response, not even an acknowledgement of the issue. 

With todays broken update (Mac track-pad issues) and unbearable token system crippling our workflow, Fusion is fast becoming an unviable solution for us. We will have to seriously re-evaluate our options if nothing changes.

Please, support/devs/anyone, if you're seeing these multiple threads, you are costing us valuable time, money and business, re-instate local simulation. This is not acceptable. 

Thanks, 

Doug

[asheepcalledgeorge]

Just wanted to bump this again, as I would really appreciate some feedback from Autodesk.

[asheepcalledgeorge]

Bump again.  I would really appreciate some recognition of this post from Autodesk.