It’s been a challenging few days of “acceleration” in Munich. We’ve had a large, highly motivated crowd, which has definitely been a highlight.
On the lowlight side, during the last 24 hours we’ve had our first major outage of the Forge platform. A core service, ACM – our Access Control Manager – went down early yesterday morning (for us here in Europe). This impacted both the Data Management and Model Derivative APIs, without which you can’t really use the Forge viewer, for example, as well as various other cloud-connected products such as Fusion 360 and Collaboration for Revit. Thankfully the service has now been restored, which means the Accelerator attendees making use of the affected services can continue their work.
One comment I’ve heard is that people didn’t have a good sense of when things were going to be back online. Part of the problem with this particular outage was that this information wasn’t easily predictable, but here are some pointers on getting a better idea of the current status. The first is the health dashboard for our cloud products. The second is the equivalent page for our web-service APIs. The last one – which I used regularly over the last 24 hours – was the A360 login: this had a “best estimate” countdown for when this particular service was due to come back online. For sure the goalposts kept shifting, but it did reflect the most up-to-date information that was available.
This outage is something we clearly need to learn from. It’s completely unacceptable that a service failure such as this can lead to Forge and dependent products going down for an extended period. I have no doubt there are many frustrated developers and customers out there, right now, questioning the wisdom of using the cloud for mission-critical activities.
We need to understand what went wrong and take concrete steps to make sure it never happens again: that much is clear. I don’t know what those steps are – they will presumably depend on the results of the post-mortem – but I’m confident we have both the mission and the organisation in place to do what’s needed. It’s completely natural for people to have lost some trust in Autodesk, at this point: it’s now our job to regain that trust and deliver on our commitment to provide a reliable infrastructure for cloud-based products and services.
I used yesterday’s downtime as an opportunity to set up the collaborative HoloLens demo where multiple devices control the same model.
Here’s a quick video I recorded of one Accelerator attendee taking it for a spin:
I’m heading home from Munich this afternoon. It’s been an interesting week for a number of reasons – Forge Accelerators are certainly never dull – but it’s time for me to re-engage on my primary responsibilities: Dasher 360 is coming along really well, and I’m impatient to get cracking on testing various scenarios for integrating it with VR.