Gartner 2018 数据库（OPDBMS）魔力象限：阿里云、Actian、MongoDB 上榜

作者：Merv Adrian、Donald Feinberg 和 Nick Heudecker

在竞争日益激烈的市场，评估OPDBMS方案的数据和分析主管必须兼顾当前需求和未来需求。非关系型和基于云数据库的供应商为全球企业提供了新的机会，企业可以利用这份魔力象限在更新改造方面做出更明智的选择。

战略性规划假设

• 到2019年，为云DBMS架构设计的存储资源和计算资源实现分离将成为主导性的数据库平台即服务（dbPaaS）模式，也会开始出现在本地环境（on-premises）。

• 到2020年，基于开源的DBMS产品将占DBMS收入的20%以上，这将加大其对主流买家的吸引力。

• 到2020年，关系数据库技术将继续用于至少70%的新应用和新项目。

• 到2023年，所有数据库中75%将放在云平台上，这个动向将显著改变DBMS供应商的格局。

市场定义/描述

操作型数据库管理系统（OPDBMS）市场的主角是适合用来支持业务流程的传统事务的关系和非关系数据库管理产品。其中包括一系列广泛的企业级应用软件，既指外购的业务应用软件，比如ERP和CRM应用软件，也指定制的事务系统。我们对这个市场的定义还包括支持面向物联网的交互和事件处理（传输中数据）的DBMS产品。

按照Gartner的定义，OPDMBS管理的工作负载包括如下：

• 批量加载

• 实时或持续的数据加载

• 并发的在线和基于Web的新颖/更新事务

• 操作型报告

• 管理外部分布式流程，比如“后备”（look-aside）查询

OPDBMS产品必须提供对这些工作负载划定优先级的功能，以便它们在并发运行时，满足服务级别协议（SLA）。

Gartner对DBMS的定义是用于定义、创建、更新、管理和查询数据库的一套完整软件系统。这里的术语“数据库”是指有组织的数据集，这些数据可能有多种格式，存储在某种形式的存储介质中（存储介质包括传统硬盘、闪存、固态硬盘和DRAM）。数据库必须包括这种机制：隔离和管理工作负载需求，并在数据的托管实例里面控制最终用户访问的各个参数。此外，DBMS应提供与独立程序和工具联系的接口，并允许和管理多种类型的并发工作负载的高效运行。不存在这个前提：DBMS非得支持关系模式或如今使用的可能全部类型的数据。此外，我们的定义并未规定DBMS必须是闭源产品，还包括商业支持的开源DBMS产品。

OPDBMS可以支持多种不同的交付模式，比如独立的DBMS软件、云（公共云和私有云）映像或容器化版本、符合认证的配置以及数据库设备。分析每一家供应商时，一起讨论和评估这些模式。

在这份魔力象限中，我们将供应商的所有OPDBMS产品视为一套。如果某供应商销售可用作OPDBMS的多款DBMS产品，我们会在介绍这家供应商的部分中加以描述，但我们将该供应商的所有产品作为一个整体加以评估。如果任何“优势”和“注意事项”涉及一款或多款特定的产品，会在介绍供应商的部分中予以注明。由于选择范围越来越广，买家更常注重单项最佳策略，企业组织评估不同供应商的产品很重要。

Gartner 2018 年操作型数据库管理系统魔力象限：

Gartner 2017 年操作型数据库管理系统魔力象限：

Gartner 2016 年操作型数据库管理系统魔力象限：

Gartner 2015 年操作型数据库管理系统魔力象限：

新增和跌出的厂商

由于市场不断变化，我们相应审查并调整了魔力象限的入围标准。由于这番调整，入围任何魔力象限的供应商组合可能会随时间而变化。一家供应商出现在某一年的魔力象限，但未出现在下一年的魔力象限，这未必表明我们改变了对这家供应商的看法。这可能表明了市场发生变化、因而评估标准发生变化，或者这家供应商关注的重心发生变化。

新增

• Actian

• 阿里云

• MongoDB

跌出：一家都没有

云头条摘选了比较受关注的三家供应商的优势和注意事项翻译发布，供各位读者参考，其他公司内容请参考英文版：

阿里云

阿里云是一家全球云计算公司，总部在中国杭州，国际总部在新加坡。它提供众多的服务，比如：支持MySQL（基于阿里云AliSQL）/SQL Server/PostgreSQL的ApsaraDB for RDS（关系数据库服务）、ApsaraDB for Redis、POLARDB、HybridDB for MySQL及PostgreSQL，以及Elastic MapReduce for Hadoop。此外，Apsara Stack提供本地私有云实施。

优势

• 广泛的产品组合：阿里云拥有的DBMS服务品种是这份魔力象限中云服务提供商（CSP）中最广的。虽然它缺少宽列非关系模式（比如用于Apache Cassandra），但除此之外的模式都可供选择；而且在许多情况下，有多种选择（比如PostgreSQL版和MySQL版）。阿里云拥有为数不多的针对时间序列的云端产品之一，即HiTSDB。它还有面向EnterpriseDB Postgres Plus和MariaDB的托管服务。

• 市场占有率：据Gartner的DBMS市场数据显示，阿里云在2017年攀升至第12位，较2016年增长88%，当时还排在第20位。阿里云是中国最大的云提供商，但也支持中国境外的10个数据中心区域，2个在美国。与亚马逊相似，阿里云以零售行业起家，这让它有望作为全球四大或五大CSP中的一员，继续增长。

• 云和混合模式的潜力：阿里云推销的Apsara Stack是一款面向本地部署的完整私有云。这与AWS和谷歌云平台相比具有竞争优势。它还让阿里云有可能跨所有服务，在云环境和本地环境之间共享数据――其他大多数CSP做不到这点。此外，它为所有的Apsara DBMS产品提供了“平移”（ lift and shift）灵活性。

注意事项

• 重心在中国：阿里云是中国的一家CSP，虽然国际总部在新加坡。这种安排存在两个问题。首先，想与美国的各大CSP竞争，阿里云就需要大幅增加其在美国境外的区域数据中心的数量（出于政治原因，它无法依赖美国境内的大幅增长势头）。其次，阿里云的DBMS产品组合中大部分只在中国境内可用。这让产品无法被世界上的其他地区使用――几个调查对象强调了这个限制因素。

• 功能、成本和支持：虽然阿里云的大多数调查得分都是平均值，但自动数据分发和高速事务处理的得分却远低于平均值。这些功能对于物联网和高端生产系统来说很重要。此外，几个调查对象提到高成本和不一致的支持是两大问题。此外，鉴于阿里云实现的高增长率，支持是老大难问题。

• Serverless模式和产品集成：阿里云的大多数DBMS服务是独立的，最近才凭借POLARDB增添了Serverless模式。虽然阿里云几乎拥有客户所需的每一种DBMS模式，但选择正确的模式并将其集成到应用软件中却很困难。此外，Serverless模式现在必不可少，以便在集成来自多个服务的数据时降低成本并增强灵活性。

AWS

AWS是亚马逊的全资子公司，总部位于美国华盛顿州西雅图。AWS提供Amazon DynamoDB（非关系文档和键值DBMS）、Amazon ElastiCache（提供Redis和Memcached）、Amazon Neptune（图形DBMS），以及Amazon Elastic MapReduce（EMR）Hadoop发行版。它还销售亚马逊关系数据库服务（Amazon RDS），其关系数据库引擎支持MariaDB、Microsoft SQL Server、MySQL和Oracle，以及支持MySQL和PostgreSQL的Amazon Aurora。

优势

• 市场发展势头：2017年，作为基础设施即服务（IaaS）和平台即服务（PaaS）两大产品市场的绝对CSP领导者，AWS的DBMS收入增长了一倍以上（连续第二年），超过SAP，夺得第四位。产品组合继续扩大，增加了RDS产品，比如Aurora for PostgreSQL和Aurora for MySQL、面向图形使用场景的Amazon Neptune，以及在开发和数据移动方面日益扮演重要角色的AWS Glue。AWS为其数据库迁移服务制定了积极大胆的路线图，并实现了既定目标，这加快了迁移到云、愿意考虑替代DBMS的客户采用新产品的步伐。亚马逊声称，截至2018年5月，70000个客户迁移了数据库。

• 快速交付：AWS经常添加新的功能、区域和相关产品，以挑战传统产品的领导地位。比如说，DynamoDB新增了“生存时间”特性、用DynamoDB加速器（DAX）进行加速、自动扩展、静态加密、按需备份和时间点恢复等功能。AWS的调查分数反映了它对于大多数工作负载而言有多强的竞争力。虽然它并非在任何类别都居于领先，但只在专业服务、云/混合部署和可调优的一致性方面低于平均值。AWS积极寻求国际认证，在此过程中打破了前几年阻碍客户采用云的众多障碍。

• 加快合作和垂直市场活动：AWS合作伙伴网络在2017年新增了10000个成员，声称其60%的合作伙伴在美国境外。AWS Marketplace现在提供4200多个软件列表。AWS新增了在支持基于vSphere的私有云的Amazon RDS方面与VMware集成的功能，这为混合部署使用场景提供了契机。AWS已派出了涉及16个职能领域的营销团队、销售团队和咨询团队，覆盖垂直市场、迁移和AP支持等使用场景，以及存储和DevOps等技术专业领域。

注意事项

• 有限的本地功能：AWS仅在云端提供其服务。虽然一些AWS产品基于本地产品，也有强大的迁移服务，但缺少本地版本对一些企业组织来说却是限制因素。接受调查的客户给AWS所打的分接近云/混合部署的最低值。AWS最近宣布与VMware建立合作伙伴关系，还宣布了与相应的本地DBMS联系的连接件，以便在混合环境中起到帮助。

• 来自零售竞争对手的阻力：虽然AWS有许多零售客户，但仍然有一种看法：与亚马逊竞争的组织（比如电子商务和零售公司）不应该使用AWS，因为这么做只会便宜了竞争对手（即AWS）。一些为谷歌云平台提供参考客户的大型零售商已公开承认：在证明了谷歌云的价值，并确信谷歌的能力之后，从AWS迁移到了谷歌。

• 糟糕的专业服务结果：在参考客户调查中，AWS的专业服务得分最低，不止一个标准偏差（STD）低于平均值。一些客户认为不同的设计假设是个挑战，自己不知道该如何应对；另一些客户提到了AWS的迁移工具相对不成熟。AWS最近努力加强与领先服务提供商的合作伙伴关系，有望改善情形。

谷歌

Google总部位于美国加利福尼亚州芒廷维尤，是Alphabet控股公司的全资子公司。谷歌云平台（GCP）旗下的谷歌dbPaaS产品包括：Cloud Spanner关系DBMS（RDBMS）、Cloud Bigtable、支持非关系DBMS使用场景的Cloud Datastore、支持内存数据存储的Cloud Memorystore for Redis、Firebase实时数据库，以及支持移动应用的Cloud Firestore（测试版）。为了支持其他平台中的数据，谷歌提供了支持托管版MySQL和PostgreSQL的Cloud SQL。谷歌还与许多数据库供应商建立了合作伙伴关系，以便在虚拟机上轻松创建和管理数据库映像。

优势

• 托管服务交付：接受调查的参考客户盛赞GCP的OPDBMS管理功能，因此该公司在操作易用性方面得分最高，在总体满意度方面得分第二。但是，一些参考客户表示希望更深入地了解DBMS操作和调优。这表明公司企业要转变文化，才能充分利用GCP的托管服务。

• 不断扩大的合作伙伴生态系统：去年，谷歌与思科、NetApp、Salesforce和SAP等大型企业软件供应商建立了合作伙伴关系。谷歌还在深耕针对SaaS合作伙伴的计划，并承诺所有的销售、专业服务和营销活动都有100%的合作伙伴参与率。

• 客户可享用的技术渠道：四分之三的谷歌参考客户称自己愿意采用比较新、风险比较高的技术。这对谷歌来说是件好事，因为它经常在产品生命周期的早期阶段为客户提供新颖的和更新后的产品，以便获得即时反馈。参考客户很欣赏这种可享用性，因为这让它们能够更早地规划采用，并保持领先竞争对手。

注意事项

• 功能不足：谷歌的参考客户中有三分之一表示，GCP OPDBMS套件功能差或缺少功能。参考客户具体提到了与以下几方面有关的不足：更丰富的客户端库、更精细的身份及访问管理功能、查询分析、操作工具以及预算管理方面的可见性。此外，谷歌在数据库活动监控方面从参考客户处获得的分数是倒数第二。

• 支持方面的挑战：参考客户提到了一级支持方面的问题，在这方面给谷歌所打的分接近所有供应商中的最低值。它们提到了对请求和操作问题的响应速度低于预期，除非问题上报到专门的产品团队，否则不太可能得到解决。

• 滞后的市场意识：该魔力象限调查的450多个参考客户中有15%评估了GCP，但最终没有选择。与去年的6%相比，这个数字已有了大幅提高，但评估率仍然落后于GCP最直接的竞争对手，而且差距明显。虽然谷歌在提高GCP能力的知名度方面还有大量工作要做，但咨询Gartner的用户却对GCP产品表示了越来越浓厚的兴趣。

Actian

Actian, which is headquartered in Palo Alto, California, U.S., offers Actian X Hybrid Database for combined operational and analytical processing, NoSQL Object Database and Zen Embedded Database by subscription-based and perpetual licensing. Actian X began shipping in April 2017 as a free upgrade to Ingres, which is nearly four decades old. Half of the Ingres customer base is European. A managed service is available, but no database platform as a service (dbPaaS).

Strengths

• Loyal customers: Most of the Actianreference customers we surveyed had been using Ingres for over 10 years. Its low maintenance requirement was singled out as a key reason for this. Ingres is the standard DBMS in half of Actian’s reference customers.

• Renewed product investment: Actian X Hybrid Database combines Ingres with the X100 engine of Actian’s Vector offering to create a real-time updatable column store with features such as single instruction, multiple data (SIMD) vector processing, chip cache exploitation, and automatic storage indexes for reducing I/O. In addition, Actian partners with Esri, Safe Software and others to enhance its spatial tools.

• Suitability of Zen for edge and IoT uses: Actian Zen Embedded Database is a purpose-built self-tuning, zero database administrator (DBA) offering with multiple licensing models. In addition to Windows and Linux, it supports Android, Raspbian, Windows 10 IoT Core and Nano Server. Instances can share data without extraction, transformation and loading (ETL), and DataConnect data integration ties Zen devices and Zen gateways to Vector and Actian X to create end-to-end data flows.

Cautions

• Lateness of key enterprise features and multimodel support: Online patching is not possible, nor are Java stored procedures. There is no continuous statistics collection or continuous tuning, and Actian has not yet begun to create internal machine learning (ML)-based optimization. Synchronization of the in-memory column store requires the creation of triggers and stored procedures. As yet, there is no document support via JavaScript Object Notation (JSON) or XML, graph or linked data types. This will make it difficult for Actian to capture the new workloads that are behind much new business.

• Customer dissatisfaction: Actian received several survey sentiment scores greater than one standard deviation (STD) below the mean, including for value, negotiation, overall product capabilities, integration and deployment, and service and support. Respondents singled out cloud/hybrid deployment, high availability/disaster recovery (HA/DR), automated data distribution, and gaps in security, such as the absence of data masking and poor database activity monitoring.

• Poor customer upgrade momentum: Actian’s customer loyalty is a double-edged sword, as three-quarters of its customers are running a version three or more releases behind the latest. The company still offers no dbPaaS version that could help with upgrade paths and new trials.

DataStax

DataStax, which is based in Santa Clara, California, U.S., provides DataStax Enterprise (DSE), a nonrelational multimodel DBMS in an integrated platform. DSE is aimed at mixed workloads and built on the Apache Cassandra DBMS, with wide-column, key-value and document/JSON support, plus a graph store. The product is available in two subscription package levels: Basic and Max. There are two add-on options: DSE Analytics Solo and DSE Graph. DSE is available on-premises, through multiple cloud providers and for hybrid cloud deployment. DataStax also offers a managed dbPaaS, DataStax Managed Cloud, in multiple public cloud environments.

Strengths

• Cloud offerings and data distribution: DataStax Managed Cloud, introduced in 2017, has proven a solid offering, resulting in a survey score for cloud/hybrid deployment greater than one STD above the mean. The score for automated data distribution was also up (from 2017) to one STD above the mean, aided by the availability of additional management capabilities for shards and nodes.

• Service, support and professional services: Many survey respondents identified customer support as a strength, stating that they were happy they chose to pay for support. This demonstrates that DataStax has increased the quality of these services over the past year. Survey scores for these areas rose from one STD below the mean in 2017 to the average score for all vendors in 2018.

• Go-to-market and sales strategy: DataStax has revamped its go-to-market strategy to target specific prospects, and this, coupled with vertical-segment sales support, resulted in a 50% increase in growth from 2016 to 2017. DataStax must continue to pursue this strategy and change the market’s perception that it is leaving behind open-source Cassandra.

Cautions

• Negotiations and pricing: Reference customers for DataStax identified difficulties with negotiating contracts and gave it a survey score for negotiations well below the mean. These difficulties are also mentioned by Gartner clients during interactions with analysts. This is a consistent theme with open-source vendors as they attempt to balance open-source pricing with creation of revenue opportunities.

• Ease of operations and programming: Survey scores for both these aspects of DataStax’s offering were well below the mean, and many respondents mentioned the need for professional services and education to use the product effectively. Respondents claimed that more upfront education and planning were required, although they described the education that was provided as excellent.

• Open-source perception: Gartner’s inquiry service continues to receive questions about DataStax’s apparent withdrawal from the Apache Software Foundation. Although we believe DataStax remains committed to the Apache Cassandra project, DataStax must continue to assure existing and prospective customers of its continued commitment. DataStax remains the primary contributor of bug fixes and new functionality.

EnterpriseDB

EnterpriseDBis aprivately held vendor based in Boston, Massachusetts, U.S. It sells the EDB Postgres Platform, based on the PostgreSQL open-source DBMS. It offers Developer, Standard and Enterprise subscriptions. There is also a private cloud built on OpenStack, and the Postgres Plus Cloud Database (PPCD) dbPaaS service. EDB Postgres Ark is a framework for provisioning databases in multiple cloud platforms. EDB Postgres is an operational DBMS standard at over half the company’s surveyed reference customers, where it has typically been in production for four years or more.

Strengths

• Growth and community leverage: EnterpriseDB’s revenue grew by over 30% in 2017. It participates in the sizable Postgres community; releases typically follow closely upon the open-source version. Its website offers free introductory training on demand.

• Improving functional richness and hybrid deployment:EDB Postgres Ark integrates with AWS, Azure and OpenStack to provide more hands-on operational control. EDB Replication Server supports deployments spanning clouds and on-premises environments, with Data Adapters for Oracle, Microsoft SQL Server and SAP ASE. The addition of TimescaleDB, the Postgres array type and AgensGraph (Cypher-based) enhances EnterpriseDB’s relatively full SQL capabilities, which include arrays and windowing, declarative table partitioning, publish-subscribe and quorum commits.

• Steady customer satisfaction: Surveyed reference customers highlighted EnterpriseDB’s value. Its improvement in last year’s survey in terms of overall customer satisfaction was continued in this year’s survey, with solid middle-of-the-pack scores. A new customer success program is in place to build on these advances.

Cautions

• Performance scores and feature expectations: Surveyed reference customers scored EDB Postgres more than one STD below the mean for automated data distribution capabilities, high-speed data ingestion and high-speed transaction processing. Comments highlighted the absence of zero-downtime upgrades, autotuning and an in-memory column store. Security also scored below average. Although EnterpriseDB offers a pioneering Postgres Security Evaluation Service, customers noted the absence of data masking and activity monitoring. In addition, they deemed the auditing capability limited.

• Competitive environment: Cloud platform vendors now all offer some type of PostgreSQL dbPaaS. Their opportunity to optimize for their own managed storage represents a significant threat to EnterpriseDB. On-premises, additional Postgres-based vendors are gaining some visibility. For Oracle replacement opportunities, MariaDB now competes with its implementation of Oracle’s Procedural Language/Structured Query Language (PL/SQL) in its MySQL-based offering.

• Ecosystem building: Adoption by third-party software vendors remains a challenge for EnterpriseDB (except for its existing relationship with Infor), especially in the application space. EnterpriseDB relies on resellers in southern Europe, Japan and Latin America for sales and Tier 1 and 2 support. The company claims to have products in development with major system integrators and expanded partnerships with Alibaba, Hewlett Packard Enterprise, IBM and Red Hat, and its work with Quest and Pivotal is now in market. This is, therefore, an area in which EnterpriseDB is improving.

IBM

IBM, which is based in Armonk, New York, U.S., offers Db2 for Linux, UNIX and Windows (LUW), Db2 for z/OS, Db2 Hosted, Db2 on Cloud, Db2 Event Store, the Db2 Analytics Accelerator appliance, IBM Graph, Information Management System (IMS) and Informix. There is also IBM Compose for several open-source managed DBMSs (including non-IBM ones) in the cloud, and IBM Cloudant (proprietary but based on Apache CouchDB).

Strengths

• Rich features and open-source support: Db2 LUW and Db2 for z/OS have rich feature sets. IBM also has fully compatible managed cloud (Db2 on Cloud) and Db2 Hosted cloud offerings, in addition to its Db2 on-premises offerings. Although many of IBM’s surveyed reference customers mentioned that IBM is slow to deliver new features, they also praised its strong feature set. Further, many praised IBM’s support for, and delivery of, open-source components integrated into its commercial products.

• Availability and stability: IBM Db2 is well known for its high availability and stability as a DBMS, especially in the case of Db2 for z/OS. IBM also offers Db2 pureScale, which adds capabilities similar to those of Db2 LUW. The majority of IBM’s reference customers mentioned these features and gave IBM high scores for HA/DR.

• Service and support: IBM has traditionally offered great service and support through a global organization that includes a partner program with global reach. This has not changed. This year, 90% of IBM’s surveyed reference customers made at least one positive comment about the quality and timeliness of service and support from IBM’s customer support and field support teams.

Cautions

• Sales execution: IBM continues to struggle with its sales and marketing model, as evidenced by a continued decline in its DBMS revenue, which fell again in 2017, according to Gartner’s information. Several reference customers remarked that IBM’s sales model has identified too many sales silos and that it is difficult to navigate. Our conversations with Gartner clients indicate that Db2 is seldom shortlisted, even when the client already has it. IBM’s new digital marketing effort will help by simplifying the sales model for data management products. IBM must fix its sales and marketing model if it is to reverse the downward trend.

• Cloud and hybrid deployment: For the second year, IBM received an overall score one STD below the mean for cloud and hybrid deployment in the reference customer survey. Although IBM has a wealth of cloud DBMS products, it appears that customers struggle to integrate on-premises and cloud systems.

• Pricing: IBM has long struggled with complex pricing models, and this year, its surveyed reference customers identified this as a major problem. For both suitability of pricing method and satisfaction with value for price, IBM scored the lowest of all the vendors in the survey, at greater than one STD below the mean. In addition, respondents identified IBM’s pricing as the thing they most disliked about its products. However, we believe IBM’s new digital marketing effort and the pricing and packaging changes made more than a year ago (although slow to be adopted by customers) will improve its pricing.

InterSystems

InterSystems, which is based in Cambridge, Massachusetts, U.S., was founded in 1978. Building on the success of Caché, a hybrid, multimodel DBMS supporting relational and nonrelational access, InterSystems introduced the IRIS Data Platform in January 2018 to increase its focus on scalability heterogeneous data and fast data. InterSystems maintains its position in the top eight DBMS vendors, in Gartner’s estimation, although it is lagging slightly behind the overall market’s growth as it focuses on adding the new offering to its portfolio.

Strengths

• Functionality: InterSystems’ multimodel DBMSs support SQL across object and nonrelational models. In addition, a Spark connector and Predictive Model Markup Language (PMML) support expanded heterogeneous data enablement. Surveyed reference customers identified flexibility and functionality as key advantages, and gave good scores for security features. Multitenant capabilities and the InterSystems Cloud Manager, a tool that provides, deploys and manages the IRIS Data Platform on the cloud of choice, offer a path to hybrid and cloud deployment for partners — a key part of the company’s strategy.

• Loyal customer base: Reference customers continue to indicate they will increase their usage for InterSystems’ products. InterSystems’ high score for perceived value is helped by its support being included in the licensing model, although the licensing model itself received some negative comments. InterSystems is introducing new cloud-based licensing models and no-cost development licenses. Furthermore, it is expanding beyond its strong healthcare base by gaining customers in the finance and manufacturing sectors.

• Stability and support: In general, InterSystems again received excellent scores from reference customers for minimal downtime and for overall service and support. However, its geographic expansion has exposed a need to improve its service and support in Asia/Pacific.

Cautions

• Market awareness: InterSystems is correctly perceived as being primarily active in the healthcare sector, which still represents 80% of its customer base. Continued strong marketing in other industry segments, such as finance and manufacturing, based on growing success, is imperative. In addition, InterSystems is still heavily focused on North America. It also lacks its own dbPaaS, which may prevent it uncovering more opportunities.

• Skills and functionality: A recurring theme in previous Magic Quadrant assessments of InterSystems has been that relevant skills are hard to find — even though reference customers describe implementation as easy. Additionally, there remain gaps in terms of mobile features and heterogeneous replication. InterSystems has not yet begun to use internal ML for operational management, or to ease its essentially “do it yourself” approach to geospatial and time series use cases.

• Documentation: InterSystems has typically scored below the average for its documentation. This year there were fewer complaints, but documentation remains a concern, especially for a product that still requires different skills that are in short supply.

MapR

MapR, which is based in Santa Clara, California, U.S., provides the MapR Data Platform, which includes MapR-DB, MapR-XD for file and container support, and MapR Streams. Optional Hadoop, Spark and ML components are also available. MapR-DB is a multimodel OPDBMS compatible with the Apache HBase API. MapR also offers a small-footprint version of its Data Platform, called MapR Edge, that is suitable for IoT-style deployments. The Data Platform is available on-premises and through various cloud providers.

Strengths

• Support and professional services: Reference customers gave MapR the second-highest overall score for its support and professional services organization. They were particularly impressed by the advantages of its Quick Start program, which is used to accelerate deployment and delivery of business outcomes

• Overall satisfaction: MapR tied for the highest overall score from reference customers for organizational satisfaction. It received the second-highest score from reference customers for its HA/DR facilities.

• Diverse platform vision: MapR continues to close functional gaps and build features comparable to those of much larger companies. Its operational and analytical convergence, coupled with a platform spanning on-premises, cloud and edge deployments, makes it unique among competitors of similar size.

Cautions

• Usability challenges: Surveyed reference customers gave MapR’s product the second-lowest overall score for ease of programming. They also identified usability challenges, with administrative tooling, for example, being considered merely adequate and therefore in need of user interface and user experience improvements. MapR has introduced a developer portal to help developers become productive more quickly.

• Open-source compatibility and support: Reference customers routinely criticize MapR for its lagging support for the most current versions of popular open-source projects, and for its lack of support for projects used in the broader Hadoop ecosystem. In addition, development around Apache Drill, promoted as MapR’s primary analytical engine, appears to have slowed during the past year.

• Market visibility and direction: MapR has always struggled to attain the same level of market visibility as its competitors. It continues to move toward a comprehensive developer-targeted proposition centered on containers and analytics, which could prove challenging to sell in a market that is increasingly buying fit-for-purpose platforms, particularly in the cloud.

MarkLogic

MarkLogic, which is based in San Carlos, California, U.S., offers a nonrelational multimodel DBMS, which it describes as “operational and transactional.” The product is available in two editions: Essential Enterprise and a free developer edition. Essential Enterprise can be deployed in on-premises environments and in clouds (both as an image and as a PaaS, the MarkLogic Query Service), including those offered by AWS, Microsoft (Azure) and Google (GCP). MarkLogic also supports containers, natively.

Strengths

• Execution: MarkLogic has focused on execution over the past year, with great results — it maintains its position as the top Challenger. With its focus on global growth, it now derives 30% of its revenue from outside North America. Its expansion into new industries (with vertical solutions) and a growing partner ecosystem have yielded strong results. MarkLogic’s implementation of a subscription license model (although not open-source), has been well received, with a survey score one STD above the mean for value for money.

• Functionality and security: Reference customers scored MarkLogic one STD above the mean for multimodel, cloud and hybrid deployment and security. Security has become a focus of the DBMS market, and MarkLogic scored above the mean in all security areas, its highest score being for activity monitoring. Many survey respondents praised the flexibility of MarkLogic’s multimodel capabilities (especially for graphing). They also praised its search capabilities, an original feature of the product.

• Service and support: MarkLogic’s reference customer scores for service and support were above the mean. Additionally, respondents praised the quality of its support and the level of involvement of its service organization. These comments help to explain the product’s reliability, which is demonstrated by the lowest outage number of any vendor in this Magic Quadrant.

Cautions

• Developer skills and training: MarkLogic scored far above the mean for documentation and training, which was far better than in the previous survey, but many respondents again identified difficulty finding developer skills. Many also expressed surprise at the level of training necessary to use the product effectively. In addition, they expressed a desire to have spent more time in the planning phases of projects, as they believe this would have reduced the difficulties they encountered using the product.

• Adoption challenges: MarkLogic’s survey scores for integration with other DBMS environments and deployment were far below the mean. Many respondents commented on difficulties in these areas, stating that they required professional services from MarkLogic or partners to solve problems, and that they needed assistance with implementation. We believe this is caused more by MarkLogic’s DBMS being nonrelational than by a deficiency in the MarkLogic DBMS.

• Competitive landscape: Competition to MarkLogic is growing in two areas. First, its vision of a data platform focused on unifying data silos, which was innovative but early, is now being pursued by many vendors, including incumbent vendors and Hadoop distributors offering data platforms. Second, almost all vendors are moving to multimodel DBMSs and supporting a document-style model with JSON support. MarkLogic must continue to focus on its data platform and define its next innovative step.

Microsoft

Microsoft, based in Redmond, Washington, U.S., markets its SQL Server DBMS and Azure SQL Database (a DBMS PaaS based on SQL Server) as flagship products for the OPDBMS market. It also markets Azure Cosmos DB, a nonrelational, globally distributed document DBMS PaaS solution that is compatible with SQL, Azure Tables, MongoDB, Cassandra and Gremlin graph APIs.

Strengths

• Market-leading execution: Gartner’s 2017 data for the DBMS market shows that Microsoft’s revenue has grown above the market rate for the past four years, with the cloud accounting for an increasingly large part of its revenue. After introducing multiple cloud-based services in the past few years, Microsoft is now focused on building a cohesive experience across its portfolio.

• Breadth of portfolio and capabilities: Microsoft has created a far-reaching portfolio with both its on-premises and cloud-based offerings. Its portfolio, once strongly defined by relational products, has recently been expanded to include a nonrelational offering with the addition of Azure Cosmos DB.

• Overall experience and value: Reference customers gave Microsoft the highest score for overall experience with a vendor, as well as for value for money. Additionally, Microsoft scored above the average for end-user training and administration and management capabilities.

Cautions

• Completeness of product suite: Thirty percentof Microsoft’s surveyed reference customers identified absent or weak functionality in its products. They pointed, for example, to weak tools for migrating on-premises workloads to Azure, shortcomings in the administration of distributed SQL Server instances, and lagging security features. Another issue they identified is the lack of parity between Microsoft’s on-premises tools, which are viewed as more robust, and the versions in Azure. Microsoft’s Azure SQL Database Managed Instance, which launched on 1 October 2018, should help address this disparity.

• Pace of feature delivery: The pace of Microsoft’s feature delivery was a concern for many reference customers. Multiple reference customers stated that Microsoft is too focused on developing new features, rather than maturing existing capabilities. Others found it challenging to understand how disparate features were weaved into a cohesive story.

• Challenges communicating hybrid vision: Reference customers repeatedly described Microsoft’s new features as ad hoc and fragmented, rather than as parts of an overall, holistic vision of a hybrid data management environment spanning on-premises and the cloud. Although Microsoft continues to build on its hybrid capabilities with features like transactional replication to SQL Server on Azure and the Azure Database Migration Service, it needs to do more to communicate a cohesive vision for hybrid deployment.

MongoDB

MongoDB, which is based in New York City, U.S., offers MongoDB Enterprise Advanced; MongoDB Enterprise for OEM; MongoDB Stitch, a managed “back end as a service”; MongoDB Atlas, a cloud-based dbPaaS offering; and MongoDB Mobile. Additionally, the company offers MongoDB Charts, a data visualization product currently in beta testing, as well as management tools and various connectors for business intelligence and analytics.

MongoDB did not respond to requests to participate in this research, to provide details of reference customers or to supply supplementary information. Therefore, Gartner’s analysis of MongoDB in this Magic Quadrant draws on credible public sources.

Strengths

• Expanding product portfolio: MongoDB’s growing portfolio of offerings is well-positioned to satisfy its core market of developers. Components like Charts and its aggregation pipeline builder expand the usefulness of data stored in MongoDB, which previously had limited utility.

• Growth in dbPaaS: Since its introduction in 2017, MongoDB claims Atlas has won over 4,400 customers (as of April 2018). Atlas includes a free tier and is offered in AWS, GCP and Microsoft Azure.

• Developer mind share: MongoDB’s focused marketing efforts continue to attract significant mind share among developers and general interest in its product portfolio.

Cautions

• Pricing and contract negotiation: Users of Gartner’s inquiry service regularly complain of price increases and inflexibility when negotiating contracts with MongoDB. It is common for customers to look for alternatives or to use the unsupported MongoDB Community Edition as a way to control costs.

• Migration challenges: MongoDB positions its products as replacements for any DBMS product, but primarily those of incumbent RDBMS vendors and especially Oracle. Conversations with Gartner clients that have tried migrating complex applications indicate that they routinely encounter challenges to completing the migration.

• Increasing competition: Since MongoDB helped define the NoSQL (now “nonrelational”) DBMS segment, dozens of competitors have emerged, ranging from large incumbents to fellow “insurgents.” MongoDB may have difficulty maintaining its competitive position as more established companies compete with the same marketing message of developer agility, supported by stronger technical foundations and better integration across portfolio components.

Oracle

Oracle, which is based in Redwood Shores, California, U.S., markets a complete set of DBMS products for operational systems. These include Oracle Database, Oracle TimesTen, Oracle Berkeley DB, Oracle NoSQL Database and MySQL. In addition to stand-alone software and cloud versions, several of Oracle’s DBMSs are available in engineered systems (appliances). Oracle recently released its Autonomous Transaction Processing (ATP) dbPaaS.

Strengths

• Cloud innovation: Oracle has continued to innovate in the cloud by releasing its Autonomous Database. This is a major effort to incorporate ML into the management of not only Oracle Database but also other products, such as Oracle NoSQL, at all levels. By using ML techniques and underlying hardware, Oracle is able to tune databases automatically, update and patch the DBMS without downtime, and provide stronger DBMS security. This effort will reduce the routine tasks of a database administrator, while improving performance.

• Customer loyalty, performance and functionality: Oracle has many longtime large customers, as is demonstrated by the survey of reference customers, with three-quarters having more than 1,000 licenses and well over half having used Oracle for more than eight to 10 years. Oracle received the highest score of any vendor in this Magic Quadrant for product capabilities, and 89% of Oracle’s surveyed reference customers chose it for its product functionality and performance.

• Product satisfaction: Reference customers for Oracle scored it above average for both database security activity monitoring and HA/DR. In both categories, Oracle was the only vendor one STD above the mean.

Cautions

• Cloud competition: With the notable exception of the Application Express (APEX) service, Oracle was late to the market with managed services (dbPaaS). It has now released a true dbPaaS in the form of ATP. However, adoption of Oracle Cloud has been predominantly by existing Oracle customers, whereas Oracle’s cloud competitors have customers from all DBMS vendors. For Oracle to compete in the public cloud sector, it will need to add managed services (dbPaaS) from its competitors. It must also end the practice of not certifying or changing the licensing metrics of the Oracle Database on competitors’ cloud platforms.

• Negotiations and license complexity: Business practices remain an issue for Oracle, with contract negotiations scoring one STD below the mean. Clients who contact Gartner often accuse Oracle of having draconian licensing practices. Surveyed reference customers complained about Oracle’s cost, auditing practices and virtualization policies as major issues on-premises. Oracle continues to require double the number of licenses for using its cloud competitors, while reducing the available functionality. However, Oracle has made its cloud licensing more flexible by, for example, including bring your own licensing (BYOL) options to encourage customers to move to the cloud.

• Support and patching challenges: Oracle received the lowest survey score of all vendors for service and support. This score is underlined by comments made by clients during Gartner interactions, which identify difficulty with patching, poor responses and the need for escalations. This has been an issue for Oracle for several years as the size of its customer base has grown. However, Gartner has seen service and support issues reduced as customers move to the cloud, where patching is automated or performed by Oracle directly.

SAP

SAP, which is based in Walldorf, Germany, offers several OPDBMS products: SAP Adaptive Server Enterprise (ASE), SAP SQL Anywhere and SAP HANA. SAP HANA is available as an appliance or as software only (as SAP HANA Tailored Datacenter Integration). Both SAP ASE and SAP HANA are available as cloud offerings, including SAP HANA as a Service.

Strengths

• Continuing market leadership: SAP remains among the leaders in Gartner’s market share statistics for 2017, although its growth slowed to below the market rate. SAP HANA claims more than 25,000 customers as of July 2018 — over half of the SAP application installed base. SAP is stepping up its pursuit of general-purpose DBMS use cases with simplified pricing and changes of sales force focus, and expanding partner and ISV programs.

• Market vision: SAP continued to pursue an aggressive vision in 2017 by enhancing its multimodel capabilities with document support, including text, geographical and spatial analytics. Gartner expects to see OrientDB technology (from newly acquired Callidus Software) reflected in future releases. SAP bundles numerous capabilities that are available from other vendors only at extra cost. Thus SAP HANA Enterprise Edition includes predictive and streaming analytics and innovative privacy capabilities. SAP’s aggressively communicated roadmap encompasses expanded deployment options, microservices, tool enhancements, data tiering and life cycle support.

• Performance and integration enhancements: Surveyed SAP reference customers again praised SAP HANA’s performance, speed, and ability to combine transactions and analytics in the same database (hybrid transaction/analytical processing [HTAP]). There were also many appreciative comments about HANA’s seamless and well-supported migration capabilities — which are crucial, as most customers still migrate to HANA from existing DBMSs. SAP’s portfolio expanded in 2018 by enhancing SAP HANA with a Data Management Suite that includes data discovery, cleansing, governance and connections to third-party data.

Cautions

• Perceived value for money: Despite recent aggressive pricing, bundling and packaging optimizations, surveyed SAP HANA reference customers scored SAP one STD below the median for both value and pricing method. Overcoming this perception will require continued work with existing and prospective customers to highlight SAP HANA’s pricing and packaging model and total cost of ownership, its superior data compression and minimal redundancy, included value-add features, and flexible packaging options.

• Challenging usage experience: SAP’s HANA reference customer scores point to issues with ease of programming, quality of training and professional services. All of these are areas that have received significant investment (and a refreshed Enterprise Architecture Designer), but SAP has more work to do to communicate the improvements to customers.

• Functionality gaps: SAP’s aggressive addition of SAP HANA features had prioritized support for its own applications. For general-purpose DBMS use, SAP HANA now needs additional capabilities to compete better, including write-capable replicas, Java stored procedures, and increased support for features in SQL 2003 and beyond.

数据库交流群欢迎加入，群主微信：aclood（备注任职单位+职位，否则不予通过）

相关阅读：