3 ms·
Interesting, but needs a summary to help conceptualize the details which it slogs through.
by Roboprog 13y ago
Interesting, but needs a summary to help conceptualize the details which it slogs through.
- aphyr 13y agoTL;DR: designing distributed systems is hard, so think carefully about your problem space, required invariants, and risk tolerance; then test your systems accordingly.
- snaky 13y agoTL;DR: first rule of engineering reliable distributed systems: "do not distribute"
- lmm 13y agoThe original (non-Infoqd) article's conclusion page http://aphyr.com/posts/286-call-me-maybe-final-thoughts http://aphyr.com/posts/286-call-me-maybe-final-thoughts was quite nice.
- wheaties 13y agoAntirez responded to the original article citing that Sentinel, which was used for this test, was never designed to accomplish what was tested. So the Redis portion of that article is rather misleading: http://antirez.com/news/55 http://antirez.com/news/55
- aphyr 13y agoI am confident from theoretical argument [1] and experiment [2] that any system for failover between asynchronously replicated Redis nodes (in the absence of a strong external coordinator for requests) is subject to this kind of data loss and inconsistency. If you read the post carefully, and Antirez's followup [3], I think you'll find Salvatore's claim is not that Redis is robust to partition, but that he doesn't think the risk is significant enough to address. I disagree: partitions are a real challenge for distributed systems at all scales. [4] [1] http://aphyr.com/posts/287-asynchronous-replication-with-failover http://aphyr.com/posts/287-asynchronous-replication-with-fai... [2] http://aphyr.com/posts/283-call-me-maybe-redis http://aphyr.com/posts/283-call-me-maybe-redis [3] http://antirez.com/news/57 http://antirez.com/news/57 [4] http://aphyr.com/posts/288-the-network-is-reliable http://aphyr.com/posts/288-the-network-is-reliable