A Detailed Look At cassandra.yaml (Edward Capriolo, The Last Pickle) …

A Detailed Look At cassandra.yaml (Edward Capriolo, The Last Pickle) … SlideShare Explore You Successfully reported this slideshow.A Detailed Look At cassandra.yaml (Edward Capriolo, The Last Pickle) | Cassandra Summit 2016Upcoming SlideShareLoading in …5× 0 Comments 3 Likes Statistics Notes zheng jinyuan , ccc at Asiacell at Asiacell Vijayakumar Ramdoss , Platform Architect at Dell Anand Rao , Sr. SOA Architect at Cengage Learning No DownloadsNo notes for slide 1. Introduction toCassandra.yaml& friends 2. Hi, I’m Edward Capriolo.@edwardcapriolohttps://www.linkedin.com/in/edwardcapriolohttp://www.slideshare.net/edwardcaprioloConsultantThe Last PickleCassandra user since (v 0.6.5)White Plains, NY USA 3. We help people deliver and improve ApacheCassandra based solutions.With staff in 5 countries and over 50 yearscombined experience. 4. This talk is the ʻgatewayʼ talk…Many ʻpicklersʼ (TLP staff) are coveringsome points I will quickly cover over indepth in other talks. 5. 1. Key configuration settings2. Configuration outside of the yaml3. Multi-system configuration settings4. Advanced settingsSection Overview5. Exotic settings 6. 1. $ wget <apache-cassandra*.tar.gz>2. $ tar -xf <apache-cassandra*.tar.gz>3. $ apache-cassandra*/bin/cassandraBasic setupResult:Web scale distributed storageDrop Mic. 7. Well almost…We have to do a bit of configuration. 8. cqlsh> CREATE KEYSPACE test WITH replication = 
{'class': ‘SimpleStrategy', 'replication_factor' : 1};
cqlsh> USE test;
cqlsh:test> CREATE COLUMNFAMILY trip (src varchar,... dest varchar, PRIMARY KEY (src,dest));Before we dive into configcqlsh:test> SELECT * FROM trip;src | dest-----+------ny | cacqlsh:test> INSERT INTO trip (src, dest) VALUES ('ny', 'ca');cqlsh:test> INSERT INTO trip (src, dest) VALUES ('fl', 'ca');cqlsh:test> SELECT * FROM trip;src | dest-----+------fl | cany | ca 9. Single Data Center 10. Multiple Data Center 11. data_file_directories:- /var/lib/cassandra/data1. User data is stored in all listed directories2. Do: fast seekʼing storage (SSD)Where does the data go?3. Do: ample free space (30% overhead)4. Donʼt: Store on a SAN 12. commitlog_directory:- /var/lib/cassandra/commitlog2. Donʼt: Assume these are log4j type logs3. Do: use a dedicated disk if possibleCommit log storage4. Do: provide at least 10GB (write velocity)1. Stores unflushed mutations (write/deletes) 13. Ok we now where(most of) the data goes…How do clients connect? 14. Default port binding1. Cassandra does not bind to 0.0.0.02. 127.0.0.1 not web scale3. 7000 is the “Storage Port” inter node traffic4. 9042 is the “Native Port” client traffic 15. start_native_transport: true (default)native_transport_port: 9042 (default)listen_address: localhost1. Change listen_address to a client-reachable addressNative transport2. Do: consider transport security3. Do: consider network routing performance4. Donʼt: put nodes on a public network. EVAR 16. Outside the yaml file… 17. cassandra-env.sh (& friends)1. JVM and startup params defined outside the YAML2. Newer version of c* use jvm.options 18. #MAX_HEAP_SIZE="1G"#HEAP_NEWSIZE="100M"1. max(min(1/2 ram, 1024MB), 
min(1/4 ram, 8GB))2. Do: set lower when experimenting with workstationMemory usage3. Do: leave ample free memory for disk cache 19. JMX1. bin/nodetool uses JMX to administer Cassandra2. All management tools require password if setCheck out Nate’s talk on Securing Cassandra to learn more 20. Multi-node configurations 21. # phi_convict_threshold: 81. Threshold for failure detector3. Do: Raise for flaky WAN networks 10 - 12Phi convict threshold2. False positives make nodes appear down to peers 22. # endpoint_snitch: SimpleSnitch2. Do: use SimpleSnitch for single switch/LAN3. Consider: Multi DC to startDefining network topology1. Snitch with config data determines topology 23. 3. Rack has impact on replication copies4. Donʼt: Change rack unless you understand the impactGossiping Property File Snitch2. DC may not be physical but is a replication unitconf/cassandra-rackdc.properties 
dc=dc1rack=rack11. Information is propagated around the cluster 24. internode_compression: all | dc | noneinter_dc_tcp_nodelay: falseInternode communications1. WAN can benefit from reduced sizeserver_encryption_options:internode_encryption: noneinternode_authenticator:o.a.c.auth.AllowAllInternodeAuthenticator2. Settings which server nodes use to communicate 25. broadcast_address: 1.2.3.4listen_on_broadcast_address: falsebroadcast_rpc_address: 1.2.3.4Broadcast address1. Gossip a specific address (not bind address)2. Useful in NAT and cloud environments 26. Broadcast address 27. Advanced settings 28. Write pathhttp://www.toadworld.com/platforms/nosql/w/wiki/11621.an-introduction-to-apache-cassandra 29. #memtable_flush_writers: 11. Default One per data directoryMemtables# memtable_cleanup_thresholddefaults to 1 /(memtable_flush_writers +#memtable_cleanup_threshold: 0.112. 1 / (1 + 1) = .5 30. #If omitted, both set to 1/4 the heap#memtable_heap_space_in_mb: 2048
#memtable_offheap_space_in_mb: 20481. Depending on the next setting dictates how muchof each memory type is used.5 of what you ask?#heap_buffers: on heap nio buffers
#offheap_buffers: off heap nio buffers
#offheap_objects: off heap objects
#memtable_allocation_type: heap_buffers2. Based on column value buffers vs objects may bebetter 31. trickle_fsync: falsetrickle_fsync_interval_in_kb: 10240Trickle fsync1. Optimization to periodically f-sync large files2. Designed to prevent latency spikes in read path 32. Include image of compaction hereCompactionhttps://www.instaclustr.com/blog/2016/01/27/apache-cassandra-compaction/ 33. concurrent_compactors: 1compaction_throughput_mb_per_sec: 161. Control resources used by compactionCompaction2. Compaction throughput can be changed at runtime3. Generally concurrent_compactors < 8 and > 1 34. disk_failure_policy: stopcommit_failure_policy: stopDisk Failure settings1. stop_paranoid: shut down gossip and clienttransports even for single-sstable errors, kill the JVMfor errors during startup2. die: shut down gossip and Thrift and kill the JVM,so the node can be replaced 35. hinted_handoff_enabled: truemax_hint_window_in_ms: 10800000hinted_handoff_throttle_in_kb: 1024max_hints_delivery_threads: 2hints_directory: /var/lib/cassandra/hintshints_flush_period_in_ms: 10000max_hints_file_size_in_mb: 128
hints_compression: LZ4CompressorHints1. Hints recently redesigned, again again2. Donʼt: tune high and overwhelming recovering node3. Donʼt: tune low and have out of sync data 36. #disk_optimization_strategy: ssd1. Tip for those with rotational disksDisk optimization strategy 37. Exotic settings 38. auto_bootstrap : true(hidden variable)Auto bootstrap1. “Bootstrapping” here means: Should the nodejoining attempt to acquire data from other nodes orstartup empty2. Can be used when bringing on new datacenter3. Can be used when streaming/ join issues 39. incremental_backups: false
snapshot_before_compaction: false
auto_snapshot: trueBackup*Ish options1. Enable with external backup like tools2. Creates hard link files operator must clean up3. Enabling and not cleaning will cause disk fill up4. Truncate/drop makes snapshot 40. read_request_timeout_in_ms: 5000
write_request_timeout_in_ms: 2000
request_timeout_in_ms: 10000Per operation default timeouts1. Each operation type has different timeout2. Applied on the coordinator not the client3. Previously was only global rpc_timeout 41. commitlog_sync: periodiccommitlog_sync_period_in_ms: 10000
commitlog_segment_size_in_mb: 32Commit Log sync1. Alternative batch mode blocks ack to clients2. Commit logs persist until Memtableʼs flush 42. Thanks!@edwardcapriolo Recommended Learning SchoologyOnline Course - LinkedIn Learning Information LiteracyOnline Course - LinkedIn Learning Learning to Write a SyllabusOnline Course - LinkedIn Learning Cgroups, namespaces, and beyond: what are containers made from? (DockerCon Eu...Jérôme Petazzoni Webinar: How Active Everywhere Database Architecture Accelerates Hybrid Cloud...DataStax Webinar | Aligning GDPR Requirements with Today's Hybrid Cloud RealitiesDataStax Designing a Distributed Cloud Database for DummiesDataStax How to Power Innovation with Geo-Distributed Data Management in Hybrid CloudDataStax How to Evaluate Cloud Databases for eCommerceDataStax Webinar: DataStax Enterprise 6: 10 Ways to Multiply the Power of Apache Cassa...DataStax About Blog Terms Privacy Copyright LinkedIn Corporation © 2019 Public clipboards featuring this slideNo public clipboards found for this slideSelect another clipboard ×Looks like you’ve clipped this slide to already.Create a clipboardYou just clipped your first slide! Clipping is a handy way to collect important slides you want to go back to later. Now customize the name of a clipboard to store your clips. Description Visibility Others can see my Clipboard