You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 26 Next »

HiveServer2 (HS2) is a service that enables clients to execute queries against Hive. HiveServer2 is success to HiveServer1 which has been deprecated. HS2 supports multi-client concurrency and authentication. It is designed to provide better support for open API clients like JDBC and ODBC.

HS2 is a single process running as a composite service, which includes the Thrift based Hive service (TCP or HTTP) and a Jetty web server for web UI. 

HS2 Architecture

The Thrift based Hive service is the core of HS2 and responsible for servicing the Hive queries (e.g. from Beeline). Thrift is a RPC framework for building cross-platform services. Its stack consists of 4 layers: Server, Transport, Protocol and Processor. You can find more details about the layers at: https://thrift.apache.org/docs/concepts .  

The usage of those layers from HS2 implementation is described as below.

Server

HS2 uses a TThreadPoolServer (from Thrift) for TCP mode, or a Jetty server for the HTTP mode. 

Regarding the TThreadPoolServer, it allocates one worker thread per TCP connection. Each thread is always associated with a connection even if the connection is idle. So there is a potential performance issue resulting from a large number of threads due to a large number of concurrent connections. In future we may think about switching to another server types for TCP mode, for example TThreadedSelectorServer. Here is a article about a performance comparison between different Thrift Java servers.  

Transport

HTTP mode is required when a proxy is needed between the client and server (for example, for load balancing or security reasons). That's why it's supported as well as TCP mode. You can specify the transport mode of the Thrift service through the hive config: "hive.server2.transport.mode".

Protocol

The Protocol implementation is responsible for serialization/deserialization. We are currently using TBinaryProtocol as our thrift protocol for serialization. In future we may think about other protocols such as TCompactProtocol based on more performance evaluation.

Processor

Process implementation is the application logic to handle requests. For example, ThriftCLIService.ExecuteStatement() method implements the logic to compile and execute a Hive query.

Dependencies of HS2

  • Metastore
    Metastore can be configured as embedded (in the same process as HS2) or a remote server (which is a Thrift based service as well). HS2 talks to Metastore for the metadata required for query compilation.  
  • Hadoop cluster
    HS2 prepares physical execution plans for various execution engines (MapReduce/Tez/Spark) and submits jobs to the Hadoop cluster for execution.

You can find a diagram of the interactions between HS2 and its dependencies here.

JDBC Client

JDBC driver is recommended for the client side to interact with HS2. Note that there are some use cases (e.g. Hadoop Hue) where Thrift client is used directly and JDBC is bypassed.

Here is a sequence of API calls involved to make the first query:

  • The JDBC client (e.g. beeline) creates a HiveConnection by initiating a transport connection (e.g. TCP connection) followed by a OpenSession API call to get a SessionHandle. The session is created from server side.
  • The HiveStatement is executed (following JDBC standards) and ExecuteStatement API call is made from Thrift client. In the API call, SessionHandle information is passed to server along with the query information.
  • HS2 server receives the request and asks the driver (which is a CommandProcessor) for query parsing and compilation. The driver kicks off a background job that will talk to Hadoop and then immediately returns response to client. This is an asynchronous design of the ExecuteStatement API. The response contains a OperationHandle created from server side.
  • Client uses the OperationHandle to talk to HS2 to poll the status of the query execution.

Source Code Description

Hopefully this is kind of a manual for a new starter to locate some basic components from source code.

Server Side

  • Thrift IDL file for TCLIService: <BaseDir>/service-rpc/if/TCLIService.thrift (<BaseDir> is typically: <YourWorkspace>/hive/) 
  • TCLIService.Iface implemented by: org.apache.hive.service.cli.thrift.ThriftCLIService class 
  • ThriftCLIService subclassed by: org.apache.hive.service.cli.thrift.ThriftBinaryCLIService and org.apache.hive.service.cli.thrift.ThriftHttpCLIService for TCP mode and HTTP mode respectively.
  • org.apache.hive.service.cli.thrift.EmbeddedThriftBinaryCLIService class: embedded mode for HS2.

  • org.apache.hive.service.cli.session.HiveSessionImpl class: Instance of this class are created on server side and managed by org.apache.accumulo.tserver.TabletServer.SessionManager instance.

  • org.apache.hive.service.cli.operation.Operation class: defines an operation (e.g. a query). Instances of this class are created on server and managed by org.apache.hive.service.cli.operation.OperationManager instance.
  • org.apache.hive.service.auth.HiveAuthFactory class: a helper used by both HTTP and TCP mode for authentication. Refer to Setting Up HiveServer2 for various auth options.

Client Side

  • org.apache.hive.jdbc.HiveConnection class: implements java.sql.Connection interface (part of JDBC). An instance of this class holds a reference to SessionHandle instance which is retrieved when making Thrift API calls to server.
  • org.apache.hive.jdbc.HiveStatement class: implements java.sql.Statement interface (part of JDBC). The client (e.g. beeline) calls the HiveStatement.execute() method for the query. Inside the execute() method, Thrift client is used to make API calls.
  • org.apache.hive.jdbc.HiveDriver class: implements java.sql.Driver interface (part of JDBC). Core method is connect() which is used by JDBC client to initiate a SQL connection.

Interaction between client and server

  • org.apache.hive.service.cli.SessionHandle class: session identifier. Instances of this class are returned from server and used by client for as input for Thrift API calls.
  • org.apache.hive.service.cli.OperationHandle class: operation identifier. Instances of this class are returned from server and used by client to poll the execution status of an operation. 

 

Other resources

How to set up HS2: Setting Up HiveServer2

HS2 clients: HiveServer2 Clients

Cloudera blog on HS2: http://blog.cloudera.com/blog/2013/07/how-hiveserver2-brings-security-and-concurrency-to-apache-hive/

  • No labels