Editor: Death of commercial databases: Oracle facing difficulties
The past two decades, the commercial database market is still the most stable IT industry, one of the most sticky areas, Oracle, IBM and Microsoft, the three companies divided up the 80% share. However, we believe that this area of the speed and magnitude of the recession may make many investors by surprise.
It has affected the database market just beginning to show deflationary pressures, including:
- Move to software as a service (SaaS), most products use free open source database;
- Social media, networking and unstructured / semi-structured data, using the occasion to usher in faster growth, these occasions are not suitable for standard SQL, and database oligarchs just depends on the standard;
- A large number of free open-source option increasingly stable, increasingly powerful, most of the options is "Not Only SQL" (NoSQL), and therefore extremely suitable for the above use cases; and
- Because of Moore's Law has brought improvements in processor, memory, solid state storage, and network throughput, NoSQL databases continue to be abnormal significant benefits - These improvements enhance the ability of rapid processing SQL and NoSQL use the occasion to applications, and gradually the pure SQL database become marginalized, as pure SQL database allows mainframe-based database become marginalized in the late 1980s and 1990s. )
Microsoft and IBM will only slightly hurt, because the database software only about each vendor a mere 5% of total revenue.
If you do not take positive action to a substantial increase in non-database revenue, we believe that Oracle can not offset the decline in the forthcoming commercial databases revenue decline quickly enough to maintain its current market value.
Specifically, we believe that Oracle needs to implement more radical and more rapid transformation and cultural transformation of the organization into a "cloud first" corporate incentive sales force and customers to actively move to the cloud. The company also needs to integrate a variety of in-house developed and acquired from the cloud / SaaS products into a single platform in order to obtain future operating leverage. The next few years, these projects will bring pain in several ways, especially financially. In order to partially offset the pain, Oracle should be stripped of the remaining hardware and other non-core businesses.
Topic Description
For those who are more interested in technology get to the bottom of it, from a historical perspective the following more in-depth explanation of the five interrelated trends, these trends intensified, will jointly lead the commercial database market in 2021 to shrink 20-30%.
Companies continue to migrate to SaaS / Cloud
In addition to the early establishment of several SaaS offerings in the Oracle database on the basis of (ie Salesforce.com, NetSuite and Oracle itself), but it is difficult to find any commercial database of SaaS providers. For the founder of the company since 2005, this figure is almost zero.
Today, the vast majority of SaaS providers use open source database, or as SaaS Human Capital Management (HCM) provider Workday as developing its own database. Use on-premise enterprise software applications that many users before, including the five core client / server applications in a category: ERP (enterprise resource planning), CRM (customer relationship management), HCM (human capital management), SCM (supply chain management) and BI (business intelligence) software, after the switch to SaaS model, eliminating the commercial database seats (seat), thus eliminating maintenance originally brought / income support and future upgrades income. Oracle enterprise income seats to create even use Salesforce.com is also much less than originally deployed in the enterprise environment.
Therefore, we believe that companies move to SaaS / Cloud for commercial databases income is very negative trend. In addition, we are still in the early stages from client / server to the SaaS / cloud. According to different research claims that this migration completed only 10-25%, only just beginning to affect more important mission-critical applications (such as ERP), they are more likely in the higher-end SQL database (including Oracle and IBM running on) those databases, and a small part of Microsoft's database.
Commercial SQL databases are not well suited for handling the most attractive emerging usage scenarios
SQL has become the standard in 1987, 20 years in the definition of how to organize it, the status of searching and sorting data than shake. However, by mid-2000, major technology companies in terms of the absolute number of different structures and processes data began to encounter the limitations of SQL databases exist, they would like to retain as needed, browse, analyze data, and provide data to the user / client.
Amazon, Google, LinkedIn and Facebook through the development and implementation of their own database software - a standard SQL database software to break the constraints to solve the problem own extensions.
In addition, shortly after each company released a version of the open source database. Therefore, in 2008 and 2009, they developed many new open source database, has led the author of "next-generation database" Guy Harrison (Guy Harrison) said "some Cambrian explosion" phenomenon. As the succession away from SQL, these databases are part of "NoSQL" databases such - although most databases very different from each other, as they do with SQL different.
Given the number and variety of data rapid growth - due to social media content richer, more social media users as well as automatic data acquisition Things shock surge more than the other companies in view of the increasing desire to get the data analysis, we We believe that the use cases for NoSQL will soon be far more than the number for SQL usage scenarios.
The rise of these new use cases will drive large enterprises greater use of open-source NoSQL solutions, the result of an SQL database business became a victim.
NoSQL databases and read a mode (Schema-on-Read) method of Moore's Law
SQL database and write mode (Schema-on-Write) did not follow Moore's Law. Generally speaking, a processor, memory and storage resources required NoSQL database is much denser than the SQL database , even taking into account the fact that: they relax the SQL many constraints in terms of data organization. While NoSQL database schema as early as 2005 has existed - at least in theory is the case, but it was simply not enough processing power, memory and storage space, so they are in areas outside academia put to practical use.
In the relative scarcity of the processor / memory / bandwidth resources in the 1980s and 1990s, to comply with more stringent standard SQL is actually necessary, so in order to ensure that enterprises migrating applications from mainframes and minicomputers to a more distributed, that rely on network performance and reliability of client / server architecture needs, especially those more important mission-critical applications.
SQL is acceptable to obtain the performance and reliability comes at a price , which was mainly due to:
- Lack of flexibility on the whole data structure called pattern;
- Definition of onerous requirements brought about by the structure prior to deployment databases and related applications;
- Over time, it is difficult to change the structure, in order to obtain different types of data in order to better reflect changes in data structures and organizational aspects; and
- When the need to write mode method, that is, when the input data so that data for patterns, rather than read mode, only the data down to a big "container", after which it will be organized into a pattern.
However, over time, Moore's Law to promote the processing power, memory capacity and speed, storage capacity and network speed and throughput have been improved, which makes the user does not need more and more rigid and write SQL standard mode method , that momentum will only continue. Therefore, with each cycle of Moore's Law, SQL and save resources when writing mode method increasingly losing its competitive advantage, NoSQL and its resources are relatively inefficient, but extremely flexible read mode to get rid of the original method is becoming increasingly hampered it is used to obtain the constraint.
Memory Technology (In-Memory) brings good prospects, eliminating the shortcomings of traditional hard drives
When SQL became standardized, traditional hard drive (HDD) storage media is the only cost-effective real-time access. Thus, many of the basic code is written to the SQL database software is intended to bear HDD disadvantages, such as the read data and the data transfer request to the memory speed is slow, relatively high failure rate - at least, the main solid component and system (such as CPU, memory, and network throughput) is compared. Now that the solid-state drive (SSD) HDD design is fast becoming a cost-effective alternative to traditional SQL database software, of course, most of the code is now unnecessary - and had to make such a compromise is in order adapt to a much slower HDD.
In contrast, many NoSQL database development is to maximize the use of SSD storage media, these databases may be updated to take advantage of more innovative non-volatile memory technology , such as Intel / Micron joint development of 3D xPoint, this memory is the market. We believe that, in view of the continued compliance with the SQL standard, to maintain backward compatibility with their strong demand, SQL database vendors can not, like many open source projects as quickly optimize their code for SSD , which makes them further at a competitive disadvantage - this is g Lleyton Christensen (Clayton Christensen) a typical example called "innovator's dilemma".
Market "SQLMethadone" more and more solutions that allow enterprises to get rid of expensive commercial database
We see more and more software tools and services designed to help businesses migrate away from a commercial SQL database. Over time, even the most important mission-critical applications running on Oracle or IBM database may increasingly be surrounded, "isolated", was demolished, this scene in the 1990s as the transition to client / server architecture encounter many of the traditional mainframe applications during that.
Although from Oracle, IBM and Microsoft SQL database giant will survive for many years in some companies - once again exactly like mainframes, but they will become increasingly marginalized, and their costs will be reduced as much as possible. We see many of these tools have matured in the Hadoop ecosystem inside, the ecosystem has several ways to integrate with SQL databases, and / or SQL interface and query features on NoSQL database. In our view, this scene with a variety of methods mainframe applications with PC and client / server applications to integrate the emergence of the early 1990s to repeat itself.