oracle公司推出21版

自20世纪70年代以来，独立关系数据库公司的数量一直在稳步下降，直到只剩下Oracle。像Sybase、Ingres、Informix、MySQL、SQL Server等熟悉的名字要么已经停业，要么已经被收购。

因此，有理由相信数据库市场已经商品化，一个关系数据库和另一个数据库一样好（正或负），尽管今天Oracle占据了市场的主导份额。但如果商品化在几年前是真的，那么现在肯定不是。

在最近的过去，市场份额的丧失、技术上的停顿以及用户对针对Oracle的定价和相关政策的小小反抗，帮助初创公司和亚马逊等富有的竞争对手推出了新一代数据库。

问题是，有关商品化和otacle市场份额下降的报道被大大夸大了。在过去的十年里，甲骨文已经建立了自己的旗舰产品，不仅是更好的，但远远优于竞争对手。它通过几种方式做到了这一点。

首先，它开发了像Exadata这样的硬件，将大型数据库工作负载转移到内存中。这是一个可预见的举措，但受到大型数据库用户的高度赞赏，因为它消除了相对较慢硬盘的性能瓶颈，从而使数据库的运行速度提高了100万倍。磁盘驱动器的运行速度是毫秒级的，硅酮的运行速度是纳秒级的，这可能会大大加快速度。

接下来，也许同样重要的是，Oracle拥抱了云计算。尽管Oracle从一开始就支持大多数云计算公司，尽管Oracle的开发人员与云计算客户一起研究如何提高性能，但由于Oracle没有提供自己的云应用程序，因此被视为一个落后者。

今天的业务需求

尽管如此，Oracle还是从客户那里学到了很多东西，并将其研究成果重新投入到其核心产品中，而其他人正焦躁不安地寻找替代品。

竞争日趋激烈，如今，Oracle有着优秀而可靠的竞争对手。尽管Oracle和高德纳都有大量的数据记录Oracle卓越的性能。

Oracle利用的一大不同是，它的竞争对手现在提供多个版本的数据库，以适应特定的工作负载，如OLTP或analytics。这是一个好的战略，使公司能够集中精力优化各种功能，为谨慎的市场。

但我们不是这样工作的。

现代商业的需求已经趋同，因为像CRM这样的应用程序继续需要比以往任何时候都更多的计算和数据库访问方式。当关系数据库存在以支持行和列屏幕和报表时，挑战就少了。

然而，如今的CRM应用程序可能需要相同的数据，但它还需要知道下一个最好的服务，即分析和机器学习工作。它还可能需要认真的图形处理，以便用户直观地了解可能的下一步。

简而言之，我们的商业应用程序不能只做一件事，我们的数据库必须跟上。专业化的想法在理论上是有道理的，但在实践中会损失很多——特别是如果给定的专业数据库不如聚合版本那么健壮的话。

聚合数据库

这就是为什么我对Oracle关于其旗舰RDB新21c版本的简报中的一个特别部分如此感兴趣。

根据Oracle和Gartner的报告，Oracle自主数据库在所有四个测试的操作用例中排名第一，根据Gartner的出版物《操作用例的云DBMS关键能力》，在所有四个分析用例中排名第一或第二

那是不应该发生的。专家应该比多面手更擅长于一件事，而不是试图为所有人做所有的事。但不是在这里。在这种情况下，性能领先者仍然是市场上最早的参与者之一。此外，公司可以投入的所有重要资源（即资金）都没有带来显著的优势。

为了强调这一点，Oracle现在将其产品称为“聚合数据库”，以帮助实现差异化，并强调许多企业通常不仅仅关注OLTP或AI，而是他们的业务需要一点一切。你可以在新版本的Oracle数据库21c上搜索详细信息。

毫无疑问，数据库市场在经历了几十年的现状停滞之后，正在经历第二次繁荣，现在有一些不错的选择。一些进步需要等待更快的硬件，而另一些则需要一个商业案例来获取Oracle每年数十亿美元的研发费用。

为了让这一真理正确一次，布丁的证据真的是在吃。

原文题：Oracle Launches Version 21c

原文：Since the 1970s there's been a steady decline in the number of free-standing relational database companies until only Oracle remains. Familiar names like Sybase, Ingres, Informix, MySQL, SQL Server and others are either out of business or have been acquired.

So, it would be reasonable to believe that the database market has commoditized, and that one relational DB is as good as another (plus or minus), though today Oracle has the dominant share of the market. But if commoditization was true a few years ago, it's certainly not now.

In the recent past, lost market share, technology hiccups, and a minor rebellion among users over pricing and related policies aimed at Oracle, helped to launch a new generation of databases from startups and well-heeled competitors like Amazon.

Trouble is, the reports of commoditization and of Oracle's flagging market presence were greatly exaggerated. Over the last decade, Oracle has built up its flagship product to be not simply better, but much better, than the competition. It did this in several ways.

First, it developed hardware like Exadata to move big database workloads into memory. It was a predictable move, but greatly appreciated by large DB consumers because it removed the performance bottleneck of relatively slow hard disks, thus enabling databases to run as much as a million times faster. Disk drives operate at millisecond rates -- and silicone runs by the nanosecond -- potentially enabling that dramatic speed up.

Next, and perhaps just as important, Oracle embraced cloud computing. Even though it had supported the majority of cloud companies from the beginning, and even though Oracle developers lived with cloud customers to see how to boost performance, it was seen as a laggard because the company didn't offer cloud applications of its own.

Today's Business Needs

Nonetheless, Oracle learned a lot from its customers and plowed its findings back into its core product just as others were getting restless and seeking alternatives.

Competition heated up and today there are good and credible competitors for Oracle. Though Oracle, and Gartner, have scads of data documenting Oracle's superior performance.

One big difference that Oracle exploits is that its competitors now offer multiple versions of their databases tuned to specific workloads like OLTP or analytics. It's a good strategy that enables the company to concentrate on optimizing various functions for discreet markets.

But that's not how we work.

Modern business has converged needs because applications like CRM continue to demand more styles of computing and database access than ever before. Back when relational databases existed to support row and column screens and reports, there was less challenge.

Today though, a CRM application might need the same data, but it will also want to know the next best offer to provide, an analytics and machine learning job. It might also require serious graphics processing to give users visual understanding of possible next steps.

In short, our business apps don't do just one thing and our databases have to keep up. The idea of specialization makes sense in theory, but the idea loses a lot in practice -- especially if a given specialty database is not as robust as the converged version.

Converged Database

That's why I was so intrigued by one particular part of Oracle's briefing about the new 21c version of its flagship RDB.

According to Oracle and Gartner, the Oracle Autonomous Database placed first in all four operational use cases tested -- and first or second for all four analytical use cases according to Gartner's publication, "Critical Capabilities for Cloud DBMS for Operational Use Cases."

That's not supposed to happen. Specialists are supposed to be better at their one thing than generalists are at trying to be all things to all people. But not here. In this case, the performance leader continues to be one of the earliest players in the market. Moreover, all of the significant resources that companies can throw at the problem (i.e. money) have not resulted in significant advantages.

Just to highlight the point, Oracle is now calling its product a "converged database" to help with differentiation by highlighting that many businesses don't often only focus on OLTP or AI, but that their businesses require a bit of everything. You can search on the new version of Oracle's DB, 21c for the details.

There's no doubt that the database market is undergoing a second flowering after decades of status quo stasis, and there are some good options out there. Some of the advances had to wait for faster hardware and others needed a business case to capture part of the billions of dollars that Oracle spends on R&D every year.

To get that truism right for once, the proof of the pudding really is in the eating.