Cloudera Search Overview
Cloudera Search provides easy, natural language access to data stored in or ingested into Hadoop, HBase, or cloud storage. End users and other web services can use full-text queries and faceted drill-down to explore text, semi-structured, and structured data as well as quickly filter and aggregate it to gain business insight without requiring SQL or programming skills.
Cloudera Search is Apache Solr fully integrated in the Cloudera platform, taking advantage of the flexible, scalable, and robust storage system and data processing frameworks included in CDH. This eliminates the need to move large data sets across infrastructures to perform business tasks. It further enables a streamlined data pipeline, where search and text matching is part of a larger workflow.
Cloudera Search incorporates Apache Solr, which includes Apache Lucene, SolrCloud, Apache Tika, and Solr Cell. Cloudera Search is included with CDH 5 and higher.
Using Cloudera Search with the CDH infrastructure provides:
- Simplified infrastructure
- Better production visibility and control
- Quicker insights across various data types
- Quicker problem resolution
- Simplified interaction and platform access for more users and use cases beyond SQL
- Scalability, flexibility, and reliability of search services on the same platform used to run other types of workloads on the same data
- A unified security model across all processes with access to your data
- Flexibility and scale in ingest and pre-processing options
|Unified management and monitoring with Cloudera Manager||Cloudera Manager provides unified and centralized management and monitoring for CDH and Cloudera Search. Cloudera Manager simplifies deployment, configuration, and monitoring of your search services. Many existing search solutions lack management and monitoring capabilities and fail to provide deep insight into utilization, system health, trending, and other supportability aspects.|
|Index storage in HDFS||
Cloudera Search is integrated with HDFS for robust, scalable, and self-healing index storage. Indexes created by Solr/Lucene are directly written in HDFS with the data, instead of to local disk, thereby providing fault tolerance and redundancy.
Cloudera Search is optimized for fast read and write of indexes in HDFS while indexes are served and queried through standard Solr mechanisms. Because data and indexes are co-located, data processing does not require transport or separately managed storage.
|Batch index creation through MapReduce||To facilitate index creation for large data sets, Cloudera Search has built-in MapReduce jobs for indexing data stored in HDFS or HBase. As a result, the linear scalability of MapReduce is applied to the indexing pipeline, off-loading Solr index serving resources.|
|Real-time and scalable indexing at data ingest||
Cloudera Search provides integration with Flume to support near real-time indexing. As new events pass through a Flume hierarchy and are written to HDFS, those events can be written directly to Cloudera Search indexers.
In addition, Flume supports routing events, filtering, and annotation of data passed to CDH. These features work with Cloudera Search for improved index sharding, index separation, and document-level access control.
|Easy interaction and data exploration through Hue||A Cloudera Search GUI is provided as a Hue plug-in, enabling users to interactively query data, view result files, and do faceted exploration. Hue can also schedule standing queries and explore index files. This GUI uses the Cloudera Search API, which is based on the standard Solr API. The drag-and-drop dashboard interface makes it easy for anyone to create a Search dashboard.|
|Simplified data processing for Search workloads||Cloudera Search can use Apache Tika for parsing and preparation of many of the standard file formats for indexing.
Additionally, Cloudera Search supports Avro, Hadoop Sequence, and Snappy file format mappings, as well as Log file formats, JSON, XML, and HTML.
Cloudera Search also provides Morphlines, an easy-to-use, pre-built library of common data preprocessing functions. Morphlines simplifies data preparation for indexing over a variety of file formats. Users can easily implement Morphlines for Flume, Kafka, and HBase, or re-use the same Morphlines for other applications, such as MapReduce or Spark jobs.
|HBase search||Cloudera Search integrates with HBase, enabling full-text search of HBase data without affecting HBase performance or duplicating data storage. A listener monitors the replication event stream from HBase RegionServers and captures each write or update-replicated event, enabling extraction and mapping (for example, using Morphlines). The event is then sent directly to Solr for indexing and storage in HDFS, using the same process as for other indexing workloads of Cloudera Search. The indexes can be served immediately, enabling free-text searching of HBase data.|
How Cloudera Search Works
In near real-time indexing use cases, such as log or event stream analytics, Cloudera Search indexes events that are streamed through Apache Flume, Apache Kafka, Spark Streaming, or HBase. Fields and events are mapped to standard Solr indexable schemas. Lucene indexes the incoming events and the index is written and stored in standard Lucene index files in HDFS. Regular Flume event routing and storage of data in partitions in HDFS can also be applied. Events can be routed and streamed through multiple Flume agents and written to separate Lucene indexers that can write into separate index shards, for better scale when indexing and quicker responses when searching.
The indexes are loaded from HDFS to Solr cores, exactly like Solr would have read from local disk. The difference in the design of Cloudera Search is the robust, distributed, and scalable storage layer of HDFS, which helps eliminate costly downtime and allows for flexibility across workloads without having to move data. Search queries can then be submitted to Solr through either the standard Solr API, or through a simple search GUI application, included in Cloudera Search, which can be deployed in Hue.
Cloudera Search batch-oriented indexing capabilities can address needs for searching across batch uploaded files or large data sets that are less frequently updated and less in need of near-real-time indexing. It can also be conveniently used for re-indexing (a common pain point in stand-alone Solr) or ad-hoc indexing for on-demand data exploration. Often, batch indexing is done on a regular basis (hourly, daily, weekly, and so on) as part of a larger workflow.
For such cases, Cloudera Search includes a highly scalable indexing workflow based on MapReduce or Spark. A MapReduce or Spark workflow is launched for specified files or folders in HDFS, or tables in HBase, and the field extraction and Solr schema mapping occurs during the mapping phase. Reducers use embedded Lucene to write the data as a single index or as index shards, depending on your configuration and preferences. After the indexes are stored in HDFS, they can be queried by using standard Solr mechanisms, as previously described above for the near-real-time indexing use case. You can also configure these batch indexing options to post newly indexed data directly into live, active indexes, served by Solr. This GoLive option enables a streamlined data pipeline without interrupting service to process regularly incoming batch updates.
The Lily HBase Indexer Service is a flexible, scalable, fault tolerant, transactional, near real-time oriented system for processing a continuous stream of HBase cell updates into live search indexes. The Lily HBase Indexer uses Solr to index data stored in HBase. As HBase applies inserts, updates, and deletes to HBase table cells, the indexer keeps Solr consistent with the HBase table contents, using standard HBase replication features. The indexer supports flexible custom application-specific rules to extract, transform, and load HBase data into Solr. Solr search results can contain columnFamily:qualifier links back to the data stored in HBase. This way applications can use the Search result set to directly access matching raw HBase cells. Indexing and searching do not affect operational stability or write throughput of HBase because the indexing and searching processes are separate and asynchronous to HBase.