Главная
Study mode:
on
1
Intro
2
Why you should listen to me
3
Quick foundation
4
What makes distributed systems different
5
A subset of failures
6
Clients stuck to an overloaded process
7
Partial failure
8
"It's slow" is the hardest problem you'll ever debug
9
Create partial availability
10
"Who to Follow" in the monorail
11
Knowing what the system has done
12
Percentiles, not averages
13
Tracing
14
On profiling
15
Releases should change a metric
16
Free-form logs are liars
17
Common "problems" are overlogged
18
Uncommon problems
19
Avoid coordination
20
Backpressure
21
Dropping new messages on the floor
22
Returning "overload" error responses
23
Timeouts and exponential back-offs
24
Roll out infrastructure with feature flags
25
if (Decider.available..)
26
Multiple versions are the norm
27
Datacenter schedulers are worth it
28
Collaboration is politics
29
No time-traveling stalkers
30
moral necessity
31
Data minimization is a
Description:
Explore practical strategies for deploying distributed systems in production environments in this GOTO Chicago 2018 conference talk. Delve into the challenges of productionizing distributed systems, including handling failures, debugging performance issues, and implementing effective logging and monitoring. Learn about tactics for managing partial availability, tracing, profiling, and release management. Discover approaches to avoid coordination problems, implement backpressure mechanisms, and utilize datacenter schedulers. Gain insights on collaboration, data minimization, and the importance of considering ethical implications in system design. Acquire valuable knowledge from an experienced distributed systems engineer to improve your ability to build and maintain robust, scalable systems in real-world production environments.

Practicalities of Productionizing Distributed Systems

GOTO Conferences
Add to list