John Furrier41
Q5: When have you seen scale-out architectures reach their limits?
Chris Harney
Right after your maintenance contract is due
Jeff Hughes
Stood up a 400 node Isilon system. "It can be done"
jeff dinisco
when the back end interconnect has no port capacity
Andrew Miller
without being to product specific, there was a "scale-out" architecture (RAIN) that maxed at 16 nodes. Just a hard stop...
Chris Dwan
I'm a consultant, they don't call me to fix the scale out system when it's working right.
jeff dinisco
@andriven hahaha, always thought that architecture was a bit of joke
Chris Dwan
It's usually when the backplane gets saturated, so the system can't stay in sync anymore.
Jeff Hughes
@jhughes Just to be clear, it didn't "scale", but it stood up
Chris Dwan
@jhughes And you backed away very slowly, both hands in clear view.
Jeff Hughes
Network almost always defines the limits. There's a reason Google brags about their CLOS setup
Chris Dwan
We ran out of API calls per second on GCS - that was fun.
Chris Dwan
That moment when you realize that east-coast-2 is actually a finite resource.
Andrew Miller
@dinisco would usually see this with Infiniband before 10 GigE based architectures. Of course Infiniband could usually drive lower latency more predictably.
Kenneth Hui @rubrikInc HQ
Any architecture that requires a locking mechanism or uses a one of more master nodes will have scaling limits.
Andrew Miller
@kenhuiny Excellent point - scalable locking mechanisms as well as masterless architecture are huge.
Chris Dwan
Something about split-brain.