Page History

Live Long and And Process (LLAP) functionality was added in Hive 2.0 (HIVE-7926 and associated tasks). HIVE-9850 links documentation, features, and issues for this enhancement.

...

• Asynchronous spindle-aware IO
• Pre-fetching and caching of column chunks
• Multi-threaded JIT-friendly operator pipelines

Also known as Live Long and Process, LLAP provides a hybrid execution model which .  It consists of a long-lived daemon replacing which replaces direct interactions with the HDFS DataNode, and a tightly integrated DAG-based framework.
Functionality such as caching, pre-fetching, some query processing and access control are moved into the daemon. Small Small/short queries are largely processed by this daemon directly, while any heavy lifting will be performed in standard YARN containers.

...

The diagram below shows an example execution with LLAP. Tez AM orchestrates overall execution. The initial stage of the query is pushed into LLAP. In the reduce stage, large shuffle is shuffles are performed in their own separate containers. Multiple queries and applications can access LLAP concurrently.

Persistent

...

Daemon

To facilitate caching and JIT optimization, and to eliminate most of the startup costs, a daemon runs on the worker nodes on the cluster. The daemon handles I/O, caching, and query fragment execution.

...

• The daemons are optional. Hive can work without them and also is able to bypass them even if they are deployed and operational. Feature parity with regard to language features is maintained.
• External orchestration and execution engines. LLAP is not an execution engine (like MapReduce or Tez). Overall execution is scheduled and monitored by an existing Hive execution engine (such as Tez) transparently over both LLAP nodes, as well as regular containers. Obviously, LLAP level of support depends on each individual execution engine (starting with Tez). MapReduce support is not planned, but other engines may be added later. Other frameworks like Pig also have the choice of using LLAP daemons.
• Partial execution. The result of the work performed by an LLAP daemon can either form part of the result of a Hive query, or be passed on to external Hive tasks, depending on the query.

• Anchor
  Resource management Resource management

  Resource Management. YARN remains responsible for the management and allocation of resources. The YARN container delegation model is used to allow the transfer of allocated resources to LLAP. To avoid the limitations of JVM memory settings, cached data is kept off-heap, as well as large buffers for processing (e.g., group by, joins). This way, the daemon can use a small amount of memory, and additional resources (i.e., CPU and memory) will be assigned based on workload.

...

• Eviction policy. The eviction policy is tuned for analytical workloads with frequent (partial) table-scans. Initially, a simple policy like LRFU is used. The policy is pluggable.
• Caching granularity. Column-chunks are the unit of data in the cache. This achieves a compromise between low-overhead processing and storage efficiency. The granularity of the chunks depends on the particular file format and execution engine (Vectorized Row Batch size, ORC stripe, etc.).

A bloom filter is automatically created to provide Dynamic Runtime Filtering.

Workload Management

YARN is used to obtain resources for different workloads. Once resources (CPU, memory, etc.) have been obtained from YARN for a specific workload, the execution engine can choose to delegate these resources to LLAP, or to launch Hive executors in separate processes. Resource enforcement via YARN has the advantage of ensuring that nodes do not get overloaded, either by LLAP or by other containers. The daemons themselves is under YARN’s control.

...

LLAP servers are a natural place to enforce access control at a more fine-grained level than “per file”. Since the daemons know which columns and records are processed, policies on these objects can be enforced. This is not intended to replace the current mechanisms, but rather to enhance and open them up to other applications as well.

Monitoring

Configurations for LLAP monitoring are stored in resources.json, appConfig.json, metainfo.xml which are embedded into templates.py used by Slider. 

LLAP Monitor Daemon runs on YARN container, similar to LLAP Daemon, and listens on the same port.  
The LLAP Metrics Collection Server collects JMX metrics from all LLAP Daemons periodically.  
The list of LLAP Daemons are extracted from the Zookeeper server which launched in the cluster. 

Web Services

HIVE-9814 introduces the following web services:

JSON JMX data - /jmx
JVM Stack Traces of all threads - /stacks
XML Configuration from llap-daemon-site - /conf 

HIVE-13398 introduces the following web services:

LLAP Status - /status
LLAP Peers - /peers 

/status example

curl localhost:15002/status

{
  "status" : "STARTED",
  "uptime" : 139093,
  "build" : "2.1.0-SNAPSHOT from 77474581df4016e3899a986e079513087a945674 by gopal source checksum a9caa5faad5906d5139c33619f1368bb"
}

/peers example 

curl localhost:15002/peers
{
  "dynamic" : true,
  "identity" : "718264f1-722e-40f1-8265-ac25587bf336",
  "peers" : [ 
 {
    "identity" : "940d6838-4dd7-4e85-95cc-5a6a2c537c04",
    "host" : "sandbox121.hortonworks.com",
    "management-port" : 15004,
    "rpc-port" : 15001,
    "shuffle-port" : 15551,
    "resource" : {
      "vcores" : 24,
      "memory" : 128000
    },
    "host" : "sandbox121.hortonworks.com"
  }, 
]
}

SLIDER on YARN Deployment

LLAP can be deployed via Slider, which bypasses node installation and related complexities (HIVE-9883).

LLAP Status

AMBARI-16149 introduces LLAP app status, available with HiveServer2.

Example usage.

/current/hive-server2-hive2/bin/hive --service llapstatus --name {llap_app_name} [-f] [-w] [-i] [-t]

 -f,--findAppTimeout <findAppTimeout>                 Amount of time(s) that the tool will sleep to wait for the YARN application to start. negative values=wait
                                                      forever, 0=Do not wait. default=20s
 -H,--help                                            Print help information
    --hiveconf <property=value>                       Use value for given property. Overridden by explicit parameters
 -i,--refreshInterval <refreshInterval>               Amount of time in seconds to wait until subsequent status checks in watch mode. Valid only for watch mode.
                                                      (Default 1s)
 -n,--name <name>                                     LLAP cluster name
 -o,--outputFile <outputFile>                         File to which output should be written (Default stdout)
 -r,--runningNodesThreshold <runningNodesThreshold>   When watch mode is enabled (-w), wait until the specified threshold of nodes are running (Default 1.0
                                                      which means 100% nodes are running)
 -t,--watchTimeout <watchTimeout>                     Exit watch mode if the desired state is not attained until the specified timeout. (Default 300s)
 -w,--watch                                           Watch mode waits until all LLAP daemons are running or subset of the nodes are running (threshold can be
                                                      specified via -r option) (Default wait until all nodes are running)

Resources

LLAP Design Document

Hive Contributor Meetup PresentationTry Hive LLAP

Save

Save

Save

Save

Save

Save

Save

Save