Suuchi - Distributed System Primitives

Talk presented as part of Devday, April 2017 - https://www.meetup.com/devday_chennai/events/238855098/

Video of the talk - https://www.youtube.com/watch?v=0pW6tAM8rIQ

References - https://github.com/ashwanthkumar/suuchi-ds-primitives

1. ashwanth kumar @_ashwanthkumar software engineer suuchi - library of distributed systems primitives
2. blank yep, it’s intentional
3. from simple to unmanageable beasts how we scale systems? an opinionated view
4. Service
5. Service
6. Service Service
7. Service Service Service
8. Service Service Service
9. Service Service Service
10. Service Service Service
11. rise of KV stores distributed, replicated, fault-tolerant (optionally sorted) 2006 BigTable from Google 2007 Dynamo from Amazon 2009 VoldemortDB from LinkedIn Cassandra from facebook 2008
12. Distributed (NoSQL) Datastores Service Service Service
13. Distributed (NoSQL) Datastores Service Service Service Network Latency
14. data locality for low latency / data intensive applications
15. Service Service Service Co-locate data to improve performance
16. Sharded and replicated to improve throughput Service Service Service
17. Service Service Service Deal with complex distributed system problems at the application layer
18. 1.1.1.1 1.1.1.2 1.1.1.3 A stats.service.ix 1.1.1.1 1.1.1.2 1.1.1.3 DNS based Load Balancing Distributed stats aggregation system @indix Count(“a”, 1L) Unique(“a”, 1L) Count(“c”, 1L) Unique(“a”, 1L) Count(“b”, 1L) monoid.plus monoid.plus monoid.plus
19. पांग ப Communication key=”foo” key=”bar” key=”baz” Request Routing Sync / Async Replication Replication Data Sharding Cluster Membership distributed system problems
20. suuchi github.com/ashwanthkumar/suuchi library of distributed systems primitives
21. + HandleOrForward - Broadcast - ScatterGather primitives - communication uses http/2 with streaming
22. + Consistent Hash Ring - Your own sharding technique? primitives - sharding/routing Consistent hashing and random trees: Distributed caching protocols for relieving hot spots on the World Wide Web node 2 node 1 node 3 node 4
23. primitives - membership static dynamic fault tolerance in case of node/process failure scaling up/down needs downtime of the system
24. primitives - replication sync async provides very high availability for write systems at the cost of eventual consistency every request is successful only if all the replicas succeeded
25. + KeyValue + RocksDB - Your Own Abstraction? primitives - storage embedded KV store from FB for server workloads
26. - define a gRPC service using proto2 (or proto3) - generate the stubs in java / scala - implement the services - connect them together using Suuchi - Server abstraction getting started
27. let’s see some code see reference for actual links
28. - HTML Archive System - Handles 1000+ tps - write heavy system - Stores ~120TB of url and timestamp indexed HTML pages - Stats (as Monoids) Storage System* - All we want is approximate aggregates real-time - Real-time scheduler for our crawlers* - Finds out which of the 20 urls to crawl now out of 3+ billion urls - Helps crawler crawl 20+ million urls everyday suuchi @indix
29. idea behind suuchi membership, request routing / sharding 2011 Gizzard from Twitter 2016 Suuchi 2016 Slicer from Google 2015 RingPop from Uber
30. questions? references available at github.com/ashwanthkumar/suuchi-ds-primitives
31. more on consistent hash ring
32. primitives - request routing Consistent hashing and random trees: Distributed caching protocols for relieving hot spots on the World Wide Web node 2 node 1 node 3 node 4
33. primitives - request routing Consistent hashing and random trees: Distributed caching protocols for relieving hot spots on the World Wide Web node 2 node 1 node 3 node 4
34. primitives - request routing Consistent hashing and random trees: Distributed caching protocols for relieving hot spots on the World Wide Web node 2 node 1 node 3 node 4
35. primitives - request routing Consistent hashing and random trees: Distributed caching protocols for relieving hot spots on the World Wide Web node 2 node 3 node 4
36. - Peer to Peer system - no single point of contact - Each node handles or forwards requests transparently - Uses pluggable partitioner scheme - Can be customized as weighted distribution / Rendezvous Hash etc. primitives - request routing