Saul - I am far, far, far from any sort of tech expert, but here is how I understand Fastly’s tech in a very simple way:
Think about the master, mainframe Netflix server located at one central spot in the world. Now think about the millions of people trying to access that server all over the world, all at different times throughout the day. If all those millions of users had to be routed through this single, master server, from wildly different geographical distances from this server, there would never be enough bandwidth for all of us to enjoy HD or 4K streaming we demand. To solve this, many clusters of servers are placed in different geographic regions between the viewer and the master, mainframe Netflix server, and those clusters of servers are spread across the entire world. So now, Netflix only needs to communicate with say, a dozen cluster servers, instead of a million, individual users. This keeps things running beautifully.
Another benefit here, is that these clusters of servers provide protection to the master, mainframe Netflix server from cyber attacks. These clusters can serve as an end-points, where the virus cannot proceed to the next level (in this case, the main Netflix server). (more on this later)
What I have described above is known as an “edge” network, or Content Delivery Network (CDN).
Now, here is how Fastly comes into play…imagine having the ability to better program those clusters to share resources with other clusters! Lets take the same Netflix example above, and look at it differently:
Through programming, these clusters also have the ability to flex bandwidth where it is necessary to ensure optimum viewing. For example, between the hours of 1am - 6am in America, the need for Netflix is theoretically at its lowest point. This allows for the cluster of servers to send all that unused bandwidth to places all over the world where users are viewing Netflix during primetime hours, say 7pm - 10pm.
Also, lets just say that to watch Netflix in India, you might traditionally have to go through servers that are located in Spain. But suddenly the Spain clusters go down. Programmable clusters recognize the outage, and can flex and re-route themselves to send the data around Spain’s downed servers to another cluster of servers that are working perfectly, and then onward to India.
So clearly these servers are talking to each other .
Now, using the security example above, imagine if one cluster server that was solely focused on delivering Netflix suddenly decided to talk to a Crowdstrike cluster server. Instant Crowdstrike protection for Netflix!
I’ll stop here, but imagine the possibilities this could lead to.
(Note: I learned all this from Muji of course…https://hhhypergrowth.com/what-are-edge-networks/)