Shulou(Shulou.com)05/31 Report--

This article introduces the relevant knowledge of "what are the differences between Ignite and Hazelcast". In the operation of actual cases, many people will encounter such a dilemma. Next, let the editor lead you to learn how to deal with these situations. I hope you can read it carefully and be able to achieve something!

Introduction to 1.Ignite

Apache Ignite in-memory data organization framework is a high-performance, integrated and distributed memory computing and transaction platform, which is used for large-scale data set processing and has higher performance than traditional disk-based or flash-based technology. At the same time, it also provides high-performance, distributed in-memory data organization and management functions for applications and different data sources.

History of 2.Ignite

Ignite comes from GridGain software developed by GridGain Systems, founded by Nikita Ivanov in 2007. Nikita led the development of the leading distributed memory on-chip data processing technology-the leading Java memory on-chip computing platform, which is up and running every 10 seconds worldwide today. He has more than 20 years of experience in software application development, founded HPC and middleware platforms, and contributed to a number of startups and well-known enterprises, including Adaptec, Visa and BEA Systems. Nikita is also a pioneer in server-side application development using Java technology, and he practiced it when he was working on the integration of large European systems in 1996.

In March 2014, GridGain opened up more than 90% of the features and code of the software, retaining only high-end enterprise features such as security, data center replication, advanced management and monitoring in the commercial version. In January 2015, GridGain entered the Apache incubator under the Apache 2.0 license and soon graduated and became a top-level Apache project. Version 1.4.0 was released on September 28th and version 1.5.0 was released in early January 2016. it should be said that the speed of development and iteration is very fast. There are few related materials about this technology, but it is indeed a potential technology to solve several pain points faced by large-scale, large amount of data, high concurrency enterprise or Internet applications.

The relationship between 3.Ignite and Hadoop and Spark

Ignite and Hadoop solve different problems, even though similar underlying technologies may be applied to some extent. Ignite is versatile and related to OLAP/ OLTP's in-memory data structures, while Hadoop is just one of many data sources that Ignite natively supports (and accelerates).

Spark is a project similar to Ignite. But Spark focuses on OLAP, while Ignite performs better in hybrid OLTP/ OLAP scenarios with its powerful transaction processing capabilities. In particular, Hadoop,Ignite will provide plug-and-play acceleration for existing Map/Reduce,Pig or Hive jobs, avoiding the practice of starting all over again, while Spark needs to do data ETL first, which is more suitable for newly written analysis applications.

4. Comparison with similar technologies

Before Ignite, in order to solve the outstanding problems of data caching, reducing database load and improving query performance, many enterprise or Internet applications used Hazelcast or Oracle Coherence or GemFire (such as 12306 website) or Redis, which is more and more widely used. This paper makes a simple comparison of these related technologies. The basic content comes from its official website. It has been translated and sorted out to make it convenient for more people to know him.

4.1.Ignite and Hazelcast

Both Apache Ignite and Hazelcast provide the characteristics of rich data grid and solve the problem of caching and partitioning data in memory in a scalable distributed cluster environment.

There are many differences between Ignite and Hazelcast in caching, supporting both transaction and data queries. The following table lists some major differences that we need to pay special attention to when choosing in-memory data grid products.

Serial number comparison project Apache igniteHazelcast1 focuses on the open source Ignite is also adding and enhancing the open source version of the feature set, the latest major version also added support for C++ and .net / C #, and Node.js integration will come soon, but also add faster deadlock-free transactions. The development of Hazelcast focuses on the paid enterprise version. Features such as out-of-heap storage and continuous query are only available in the paid version. SSL encryption has been removed from the open source version and is only available in the paid version. 2JCache (JSR107) Ignite is fully compatible with JCache (JSR107) cache specification Hazelcast is fully compatible with JCache (JSR107) cache specification 3 out-of-heap storage Ignite according to user configuration support to store data in the heap or out-of-heap Hazelcast provides the function of out-of-heap storage only in the commercial version 4 as long as out-of-heap storage is configured, Ignite will store the index outside the heap (in order not to affect user applications that use in-heap memory.) Does not support 5 continuous query Ignite supports fault-tolerant continuous query, that is, continuous notification of changes in client-side and server-side subscription data is only available in the commercial (paid) version of Hazelcast. 6SSL encryption ignite supports SSL/TLS encryption, which allows all clients and servers to communicate through encrypted SSL or TLS protocols. Hazelcast removes SSL encryption support from the open source version, which is currently only supported in the commercial (paid) version. 7SQL query Ignite supports full SQL (ANSI-99) syntax to query data in memory Hazelcast provides limited support for SQL (only a few keywords) 8 associative query Ignite supports complete SQL associations, including associations across multiple caches, for example: select * from An a, B b where a.b_id = b.idHazelcast does not support any association queries, with or without SQL, if needed Developers need to manually process the results of multiple queries. 9 query consistency Ignite provides complete query consistency, that is, the query is executed in a specific snapshot, and the data update after the query starts does not affect the query results. The Hazelcast query is inconsistent, and it is possible that some of the query results will see some updates, while others will not. 10 query fault-tolerant Ignite query is fault-tolerant, that is, the query results are always consistent and will not be affected by changes in the cluster topology, such as the addition, exit or crash of nodes. Hazelcast queries are fault-tolerant, that is, the query results are inconsistent when the cluster topology changes, and the data is being rebalanced in the background. 11 data consistency Ignite supports atomic and transactional consistency of data in memory, regardless of whether the data is stored in a partition or replication cache. Hazelcast only supports atomic and transactional consistency in partitioned caches, while data stored in replication caches has no guarantee of transactional consistency. 12 deadlock-free transactions Ignite supports deadlock-free optimistic transactions, it does not acquire any locks, and users no longer have to worry about lock order, which also provides better performance. It is often impossible to update data sequentially in Hazelcast to avoid deadlocks, especially in large projects. Transactional EntryProcessorIgnite supports transactional EntryProcessor, which can execute juxtaposed business code in a transaction on the server side. You cannot use EntryProcessor in Hazelcast or execute business code in a transaction. 14Web session clustering Ignite provides caching and clustering support for Web sessions for all known application servers. Hazelcast only provides Web session clustering support in the commercial (paid) version. 15 Computing grid Ignite provides MapReduce, ForkJoin and basic distributed lambda processing in the cluster, including job load balancing, fault tolerance, checkpointing, scheduling tasks, etc. Hazelcast only supports MapReduce and distributed random jobs within the cluster. 16 streaming data Ignite supports in-memory stream processing, including query and maintenance support for sliding windows of streaming data. Hazelcast does not provide any support for convection processing. 17 Service Grid Ignite makes it easy for users to cluster their services, including supporting various cluster singletons. Hazelcast-managed services do not provide singleton clustering capabilities. 18.Net/C# support Ignite provides .NET / C # users with a complete memory organization API, including the execution of C # closures, C # caching, transactions and query API, native C # CacheStore API and other functions. Hazelcast only provides a limited client-side API for .NET / C #. 19Craft + supports Ignite to provide C++ users with a complete memory organization API. Hazelcast only provides a limited client API for C++. 20Node.js support Ignite provides client-side API support for Node.js users. Hazelcast does not support Node.js. 4.2.Ignite and Coherence

Both Apache Ignite and Oracle Coherence provide the characteristics of rich data grid and solve the problem of caching and partitioning data in memory in a scalable distributed cluster environment.

Ignite and Coherence have many differences in caching and transaction methods, while supporting data queries. The following table lists some major differences that we need to pay special attention to when choosing data grid products.

Serial number comparison project Apache IgniteOracle Coherence1 open source and closed source Ignite is an Apache project, and is constantly adding and enhancing the open source version of the feature set, the latest major version provides support for C++, .NET / C #, support for Node.js will come soon. Coherence is a proprietary software for Oracle and is not available in open source or free versions. 2JCache (JSR 107) Ignite is fully compatible with JCache (JSR 107) cache specification Coherence is fully compatible with JCache (JSR 107) cache specification 3 out-of-heap storage Ignite provides limited options for developers to store data in heap or out-of-heap Oracle Coherence based on user configuration support. 4. As long as out-of-heap storage is configured for out-of-heap indexes, Ignite will store indexes outside the heap (in order not to affect user applications that use in-heap memory). 5SQL query Ignite does not support complete SQL (ANSI-99) syntax to query data in memory. 6 associated query Ignite supports complete SQL association, including associations across multiple caches, for example: select * from An a, B b where a.b_id = b.idCoherence does not support any associated queries, regardless of whether SQL is used or not. If necessary, developers need to process the results of multiple queries manually. The 7ACID transaction Ignite provides excellent performance of thousands of transactions per second per server. Coherence does not recommend using transactions because of its low performance. 8 tiered storage Ignite supports a hierarchical storage model. Data can be stored and moved in and out of the heap and in the swap space. The upper layer will provide more storage capacity, and of course latency will increase. 9-stream data Ignite does not support in-memory inflow computing, including maintenance and query of sliding windows that support streaming data. 10 configuration Ignite supports the ability to configure the system through Java Bean and native Spring XML integration, as well as the ability to easily configure the system through code. Coherence is configured through proprietary XML format files and is not supported by code. 4.3.Ignite and Gemfire

Both Apache Ignite and Pivotal Gemfire provide the characteristics of rich data grid and solve the problem of caching and partitioning data in memory in a scalable distributed cluster environment.

Ignite and Gemfire have many differences in caching and transaction methods, while supporting data queries. The following table lists some major differences that we need to pay special attention to when choosing data grid products.

Serial number comparison project Apache IgnitePivotal Gemfire1 open source and closed source Ignite is an Apache project, and is constantly adding and enhancing the open source version of the feature set, the latest major version provides support for C++, .NET / C #, support for Node.js will come soon. Gemfire is proprietary software for Pivotal. 2JCache (JSR107) Ignite data Grid is an implementation of the JCache (JSR107) specification, which provides an easy-to-use but powerful API for data access. Gemfire does not implement JCache and uses proprietary API. 3 in-heap and out-of-heap Ignite supports storing data in and out of the heap according to user configuration. Gemfire does not support storing data out of the heap 4SQL query Ignite supports a complete SQL (ANSI-99) query syntax to query data in memory. Gemfire does not support the standard SQL syntax, but it provides its own object query language called OQL. 5 Association query Ignite supports complete SQL association, including associations across multiple caches. For example, select * from An a, B b where a.b_id = b.idGemfire does not support any cross-region or cross-cache association queries. If necessary, developers need to process the results of multiple queries manually. 6 Cross-partition transactions Ignite supports cross-partition transactions, which can be executed in all partitions cached in the entire cluster. Gemfire does not support transactions that span multiple cache partitions or nodes. 7 tiered storage Ignite supports the tiered storage model. Data can be stored and moved in and out of the heap and in the exchange space. The upper layer will provide more storage capacity, and the latency will also increase. 8-stream data Ignite does not support in-memory inflow computing, including maintenance and query of sliding windows that support streaming data. 9 configuration Ignite supports the ability to configure the system through Java Bean and native Spring XML integration, as well as the ability to easily configure the system through code. Gemfire is configured through proprietary XML format files and is not supported by code. 10 deploying Ignite nodes is peer-to-peer and automatically joins the cluster at startup (no locator servers are required). Gemfire needs to start and maintain a locator server to control the joining and exiting of nodes. 4.4.Ignite and Redis

Both Apache Ignite and Redis provide distributed caching capabilities, but the features provided by each product are very different. Redis is mainly a data structure storage, but Ignite provides many in-memory distributed components, including data grid, computing grid, stream computing and, of course, data structures.

Ignite is an in-memory data organization and provides more functions, so it is impossible to compare a corresponding functional feature, but we can still compare some data grid functions.

The ordinal comparison project Apache IgniteRedis1JCache (JSR107) Ignite is fully compatible with the JCache (JSR107) cache specification does not support 2ACID transactions Ignite fully supports ACID transactions, including optimistic and pessimistic concurrency models and READ_COMMITTED, REPEATABLE_READ and SERIALIZABLE isolation levels. Redis provides limited support for optimistic transactions on the client side. In the case of concurrent updates, the client needs to retry the transaction manually. 3 data partition Ignite supports partition cache, which is similar to a distributed hash. Each node in the cluster stores a part of the data. When the topology changes, Ignite will automatically rebalance the data. Redis does not provide partitioning, but it does provide fragmentation of the copy, which is very rigid, and requires a series of rather complex manual steps whenever the topology changes, both on the client and on the server. 4 full replication Ignite supports cached replication, and each key-value pair of each node in the cluster supports it. Redis does not provide direct support for full replication. 5 Native object Ignite allows users to use their own domain object model and provides native support for any Java/Scala, C++ and .NET / C # data types (objects). Users can easily store any program and domain object in the Ignite cache. Redis does not allow users to use custom data types and only supports predefined sets of basic data structures, such as Set, List, Array, and others. 6 client-side (near) cache Ignite provides direct support for client-side caching of recently accessed data. No support for 7 (server-side) juxtaposition Ignite supports the direct execution of any Java, C++ and .NET / C # code near the data on the server side. Redis usually does not have any juxtaposition ability, the server basically only supports LUA scripting language, the server does not directly support Java, .NET, or C++ code execution. The 8SQL query Ignite supports the full SQL (ANSI-99) syntax to query data in memory. Redis does not support any query language, only the client cache API. 9 continuous query Ignite provides support for client-side and server-side continuous queries, and users can set server-side filters to reduce and reduce the number of events transmitted to the client. Redis provides support for client-side key-based event notifications, but it does not provide server-side filters, resulting in a significant increase in update notification network traffic on both the client and server sides. 10 Database Integration Ignite can automatically integrate external databases-RDBMS, NoSQL, and HDFS. Redis cannot be integrated with an external database. That's all for "what's the difference between Ignite and Hazelcast?" Thank you for reading. If you want to know more about the industry, you can follow the website, the editor will output more high-quality practical articles for you!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.