Scaling Massive Elasticsearch Clusters

Scaling Massive ElasticSearch Clusters

Rafał Kuć – Sematext International@kucrafal @sematext sematext.com

Who Am I

• „Solr 3.1 Cookbook” author• Sematext software engineer• Solr.pl co-founder• Father and husband :-)

What Will I Talk About ?

• ElasticSearch scaling• Indexing thousands of documents per second• Performing queries in tens of milliseconds• Controling shard and replica placement• Handling multilingual content• Performance testing• Cluster monitoring

The Challenge

• More than 50 millions of documents a day• Real time search• Less than 200ms average query latency• Throughput of at least 1000 QPS• Multilingual indexing• Multilingual querying

Why ElasticSearch ?

• Written with NRT and cloud support in mind• Uses Lucene and all its goodness• Distributed indexing with document

distribution control out of the box• Easy index, shard and replicas creation on live

cluster

Index Design

• Several indices (at least one index for each day of data)

• Indices divided into multiple shards• Multiple replicas of a single shard• Real-time, synchronous replication• Near-real-time index refresh (1 to 30 seconds)

Shard Deployment Problems

• Multiple shards per node• Replicas on the same nodes as shards• Not evenly distributed shards and replicas• Some nodes being hot, while others are cold

Default Shard Deployment

ElasticSearch Cluster

Node 1 Node 2

Node 3

Shard 1 Shard 2 Shard 3 Replica 1

Replica 2

Replica 3

What Can We Do With Shards Then ?

• Contol shard placement with node tags:– index.routing.allocation.include.tag– index.routing.allocation.exclude.tag

• Control shard placement with nodes IP addresses:– cluster.routing.allocation.include._ip– cluster.routing.allocation.exclude._ip

• Specified on index or cluster level• Can be changed on live cluster !

Shard Allocation Examples

• Cluster level:curl -XPUT localhost:9200/_cluster/settings -d '{ "persistent" : { "cluster.routing.allocation.exclude._ip" : "192.168.2.1" }}'

• Index level:curl -XPUT localhost:9200/sematext/ -d '{ "index.routing.allocation.include.tag" : "nodeOne,nodeTwo"}'

Number of Shards Per Node

• Allows one to specify number of shards per node

• Specified on index level• Can be changed on live indices• Example:curl -XPUT localhost:9200/sematext -d '{ "index.routing.allocation.total_shards_per_node" : 2}'

Controlled Shard Deployment

Node 1 Node 2

Node 3

Shard 1

Shard 2

Shard 3 Replica 1Replica 2

Replica 3

Does Routing Matters ?

• Controls target shard for each document• Defaults to hash of a document identifier• Can be specified explicitly (routing parameter) or as

a field value (a bit less performant)• Can take any value• Example:

curl -XPUT localhost:9200/sematext/test/1?routing=1234 -d '{ "title" : "Test routing document"}'

Indexing the Data

Node 1 Node 2

Node 3

Shard 1

Shard 2

Shard 3

Replica 1

Replica 2

Replica 3

Indexing application

How We Indexed Data

Node 1 Node 2

Shard 1 Shard 2

Node 3

Indexing application

Nodes Without Data

• Nodes used only to route data and queries to other nodes in the cluster

• Such nodes don’t suffer from I/O waits (of course Data Nodes don’t suffer from I/O waits all the time)

• Not default ElasticSearch behavior• Setup by setting node.data to false

Multilingual Indexing

• Detection of document's language before sending it for indexing

• With, e.g. Sematext LangID or Apache Tika• Set known language analyzers in configuration

or mappings• Set analyzer during indexing (_analyzer field)

Multilingual Indexing Example

curl -XPUT localhost:9200/sematext/test/10 -d '{ "title" : "Test document", "langId" : "english"}'

{ "test" : { "_analyzer" : { "path" : "langId" }, "properties" : { "id" : { "type" : "long", "store" : "yes", "precision_step" : "0" }, "title" : { "type" : "string", "store" : "yes", "index" : "analyzed" }, "langId" : { "type" : "string", "store" : "yes", "index" : "not_analyzed" } } }}

Multilingual Queries

• Identify language of query before its execution (can be problematic)

• Query analyzer can be specified per query (analyzer parameter):

curl -XGET localhost:9200/sematext/_search?q=let+AND+me&analyzer=english

Query Performance Factors – Lucene level

• Refresh interval– Defaults to 1 second– Can be specified on cluster or index level– curl -XPUT localhost:9200/_settings -d '{ "index" :

{ "refresh_interval" : "600s" } }'• Merge factor– Defaults to 10– Can be specified on cluster or index level– curl -XPUT localhost:9200/_settings -d '{ "index" :

{ "merge.policy.merge_factor" : 30 } }'

Let’s Talk About Routing Once Again

• Routes a query to a particular shard• Speeds up queries depending on number of

shards for a given index• Have to be specified manualy with routing

parameter during query• routing parameter can take any value:

curl -XGET 'localhost:9200/sematext/_search?q=test&routing=2012-02-16'

Querying ElasticSearch – No Routing

Shard 1 Shard 2 Shard 3 Shard 4

ElasticSearch Index

Application

ElasticSearch Index

Application

Querying ElasticSearch – With Routing

Performance Numbers

Queries without routing (200 shards, 1 replica)#threads Avg response time Throughput 90% line Median CPU Utilization

1 3169ms 19,0/min 5214ms 2692ms 95 – 99%

Queries with routing (200 shards, 1 replica)#threads Avg response time Throughput 90% line Median CPU Utilization

10 196ms 50,6/sec 642ms 29ms 25 – 40%

20 218ms 91,2/sec 718ms 11ms 10 – 15%

Scaling Query Throughput – What Else ?

• Increasing the number of shards for data distribution

• Increasing the number of replicas• Using routing• Avoid always hitting the same node and

hotspotting it

FieldCache and OutOfMemory

• ElasticSearch default setup doesn’t limit field data cache size

FieldCache – What We Can do With It ?

• Keep its default type and set:– Maximum size (index.cache.field.max_size)– Expiration time (index.cache.field.expire)

• Change its type:– soft (index.cache.field.type)

• Change your data:– Make your fields less precise (ie: dates)– If you sort or facet on fields think if you can reduce

fields granularity• Buy more servers :-)

FieldCache After Changes

Additional Problems We Encountered

• Rebalancing after full cluster restarts– cluster.routing.allocation.disable_allocation– cluster.routing.allocation.disable_replica_allocation

• Long startup and initialization• Faceting with strings vs faceting on numbers on

high cardinality fields

JVM Optimization

• Remember to leave enough memory to OS for cache

• Make GC frequent ans short vs. rare and long– -XX:+UseParNewGC – -XX:+UseConcMarkSweepGC – -XX:+CMSParallelRemarkEnabled

• -XX:+AlwaysPreTouch (for short performance tests)

Performance Testing

• Data– How much data do I need ?– Choosing the right queries

• Make changes– One change at a time– Understand the impact of the change

• Monitor your cluster (jstat, dstat/vmstat, SPM)• Analyze your results

ElasticSearch Cluster Monitoring

• Cluster health• Indexing statistics• Query rate• JVM memory and garbage collector work• Cache usage• Node memory and CPU usage

Cluster Health

Node restart

Indexing Statistics

Query Rate

JVM Memory and GC

Cache Usage

CPU and Memory

Summary

• Controlling shard and replica placement• Indexing and querying multilingual data• How to use sharding and routing and not to

tear your hair out• How to test your cluster performance to find

bottle-necks• How to monitor your cluster and find

problems right away

We Are Hiring !

• Dig Search ?• Dig Analytics ?• Dig Big Data ?• Dig Performance ?• Dig working with and in open – source ?• We’re hiring world – wide !

http://sematext.com/about/jobs.html

How to Reach Us

• Rafał Kuć– Twitter: @kucrafal – E-mail: rafal.kuc@sematext.com

• Sematext– Twitter: @sematext– Website: http://sematext.com

• Graphs used in the presentation are from:– SPM for ElasticSearch (http://sematext.com/spm)

Thank You For Your Attention

Scaling Massive Elasticsearch Clusters

Technology

Transcript of Scaling Massive Elasticsearch Clusters

High-resolution imaging of embedded massive star clusters ...ao4elt6.copl.ulaval.ca/proceedings/401-9snP-251.pdf · The formation and evolution of young massive star clusters is a

Elasticsearch Terminology

Running High Performance and Fault Tolerant Elasticsearch Clusters on Docker

It’s A Match! - GrafanaCon...Infrastructure Monitoring Component Method Elasticsearch Clusters Prometheus elasticsearch exporter Kafka Clusters Prometheus kafka exporter, Prometheus

Bimodal regime in young massive clusters leading to ... · Introduction Motivation: Cooling winds !multiple populations young massive clusters have winds stellar winds, SNe !collisions

Formation of Very Young Massive Clusters and implications ...sambaran/docs/VYMCformation...Formation of Very Young Massive Clusters and implications for globular clusters 3 1-3Myr)stellarassembly.Theoveralltwo-bodyrelaxationtime,thatdeterminesthe

Elasticsearch Documentation€¦ · Elasticsearch Documentation, Release 1.4.0 Ofﬁcial low-level client for Elasticsearch. Its goal is to provide common ground for all Elasticsearch-related

THE COEVOLUTION OF NUCLEAR STAR CLUSTERS, MASSIVE …orca.cf.ac.uk/129843/1/Antonini_2015_ApJ_812_72.pdf · THE COEVOLUTION OF NUCLEAR STAR CLUSTERS, MASSIVE BLACK HOLES, AND THEIR

Massive clusters in_the_inner_regions_of_ngc 1365

New Candidate Massive Clusters from 2MASSNew Candidate Massive Clusters from 2MASS Dirk Froebrich Centre for Astrophysics and Planetary Science, University of Kent, Canterbury, UK

Neo4j Integration with ElasticSearch - ElasticSearch Meetup

Clusters Massive Cluster Gigabit Ethernet System Architecture for Extreme Devices David Culler culler U.C. Berkeley DARPA Meeting.

Winds Driven by Massive Star Clusters

young massive proto-globular clusters in the local Universeconference.astro.ufl.edu/STARSTOMASSIVE/eproceedings/talks/davies... · young, massive, proto-globular clusters in the local

Elasticsearch - nosqlroadshow.comnosqlroadshow.com/.../elasticsearch_Alexander_Reelsen.pdf · Elasticsearch - The Company • Founded in 2012 • By the people behind the Elasticsearch

Massive open star clusters using the VVV survey · Massive open star clusters using the VVV survey⋆ I. Presentation of the data and description of the approach ... VVV-SkZpipeline

MASGOMAS project: Two new obscured, massive and young … · 2017-01-17 · MASGOMAS project: Two new obscured, massive and young Galactic clusters S.A. Ram rez Alegr a1;2, A. Herrero1;2,

Recovered File 1 - mpia.de · Starbursts(and(Super(Star(Clusters((• luminous,106 L • massive,(10410 6 M • compact,(2B5(pc(• protoBglobular(clusters?(• found(in(all(starbursts

Massive Star Clusters in Non-Interacting Galaxies Dynamical Mass Estimates and the (I)MF

The role of massive stars in the turbulent infancy of Galactic ...The role of massive stars in the turbulent infancy of Galactic globular clusters: Nucleosynthesis, superbubble dynamics