My experience has been that Simulation Job Manager is not very robust or efficient in downloading. Do other users share my experience?
I have attached a screenshot which show my usual state of affairs ... jobs downloading at a glacial pace. I know that there are 4-6+ Gigabytes of data just waiting on the server for completed jobs, but my local Simulation CFD job window shows only ~50% of the results downloaded. As also shown by the network monitor, Simulation Job Manager is consistently using <10% of the bandwidth available.
Regarding my 'final mile', I have a dedicated line (no other users) with 5 Mbps from my ISP. My ISP says that this link is a low latency link created for teleconferencing and offers the best QoS they have.
I am puzzled by one architectural aspect of the Sim 360 cloud. Why do they use Akamai? I understand that this is a 'geocaching' service (at least I have used it as such).
Per my understanding, the Akamai network is great for cached common public data that is accessed by many users from different geographical locations. For example, copies of a web page are replicated across geographically dispersed 'geocache' servers for fast access from different points on the internet.
However, for Simulation Job Manager I'm not sure what benefit is achieved by sending my individual data to a server network for 'geocaching' rather than directly to the end user. Perhaps the idea is that Akamai is a 'job' cache that unloads the simulation server (not a bad idea)? However, does Akamai have the 'intelligence' to cache my simulation results on the Akamai server in Denver (I know they have a close 'node' there only 2 hops away from my ISP)? Or is it being parked in Seattle, which would require that Akamai then has to sync the job to the Denver 'geocache' server before being sent to me (this would create more network hops and might also move on Akamai's more congested links).
I don't move around that much, so it would make sense for the Akamai network to 'park'/'geocache' the data close to me, but is that happening?
Edit ... I've attached a 'Ping' and 'Traceroute' screenshot. As can be seen, there seems to be good performance between me and the Akamai server that Simulation Job Manager is downloading from.
a72-246-167-18.deploy.akamaitechnologies.com resolves to 188.8.131.52
Pings are ~70 ms
Traceroute shows my ISP is 3 hops away from the Akamai server.
So, the bottleneck must be in the Akamai network or in the link from the Amazon server cloud (I believe Autodesk is using them) to the Akamai network? My guess is that it's the Akamai network.
Solved! Go to Solution.
The Akamai technology used by Simulation Job Manager is for routing rather than caching, it provides much faster routing across the Internet.