[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference ulysse::rdb_vms_competition

Title:DEC Rdb against the World
Moderator:HERON::GODFRIND
Created:Fri Jun 12 1987
Last Modified:Thu Feb 23 1995
Last Successful Update:Fri Jun 06 1997
Number of topics:1348
Total number of notes:5438

204.0. "Competing Against Oracle" by BROKE::DREYFUS () Tue Sep 27 1988 00:43

The following three articles are part of a series on Oracle.
Please forward any comments on their accuracy or usefulness.
The articles have also been submitted to Competitive Update.

Regards,

David Dreyfus
T.RTitleUserPersonal
Name
DateLines
204.1Review of Oracle's V6.0 AnnouncementBROKE::DREYFUSTue Sep 27 1988 00:44432
 







                                                    David Dreyfus
                                           Database Systems Group
                                                     DTN 381-2893
                                                   DEBIT::DREYFUS
          26-September-1988

               Review of Oracle's V6.0 with TPS Announcement

          The enclosed Internal-Use-Only information is based
          upon Oracle publications and press reactions to the
          Oracle 18-JUL-1988 announcement. As this note contains
          both fact and impressions, some of the statements I
          make are debatable.



          1  What Oracle Announced

          On July 18, 1988, Oracle announced version 6.0 of
          the Oracle relational DBMS and TPS (a transaction
          processing subsystem). According to Oracle, the new
          and upgraded products provide fault-tolerance, online
          backup and recovery, online reconfigurability, and
          high performance. Some of the lower level functions
          that have been introduced are fast commit, row level
          locking, a read-consistency model, new DBA tools, a
          new report writer, and a new query language (PL/SQL).

          For the most part, these new capabilities have been in
          the VAX Rdb product set for years. Oracle requires the
          customer to purchase an additional product, TPS, to
          get these capabilities.











                                                                1

 







          2  The market reaction

          The most exciting part of the announcement was
          Oracle's benchmarking technique which allowed them to
          show 49 transactions per second (TPS) on a 6240 with
          a 0.1 million record database. Oracle also showed 265
          TPS on an IBM MVS 3090 Model 600E with a 0.1 million
          record database. These tests used a heavily modified
          version of the TP1 benchmark. The results bear no
          relation to any number Digital, or any other vendor,
          has produced.

          The press and consultants have had a field-day
          discrediting Oracle's benchmark results. Oracle has
          had a very difficult time finding auditors who would
          verify the results (Computerworld, July 18, 1988).
          In fact, they searched for multiple auditors in an
          attempt to get their tests verified.

          Lawrence Ellison, President of Oracle, claims that
          their TP1 is a "Debit/Credit benchmark, absolutely the
          same as Tandem's." (Computerworld, August 1, 1988).
          However, unlike Tandem, the article says, Oracle did
          no front-end communication, no mirrored journaling, no
          simulated terminal networking (transactions initiated
          from main memory), etc. According to consultant Omri
          Serlin, "Oracle has gone to such an extreme in what
          they have done. It had no on-line flavor whatsoever."
          The author, referring to the material, says, "Yes,
          ladies and gentlemen, it was a thick document, but we
          should read it as well as weigh it."

          Oracle claims that Codd and Date have audited their
          benchmark. No report is available from Codd and Date
          who generally refuse to comment when asked about the
          audit (Computerworld, July 18, 1988). Many articles
          that provide analysis of the Oracle announcement
          generally dwell on the poor benchmarking techniques
          and ethics of Oracle.





          2

 







          3  Examination of Oracle's announcement



          3.1  Fault-tolerance

          Oracle now claims that their DBMS is fault-tolerant.
          From a reading of their promotional literature
          and other documents, this means that they now have
          automatic recovery if a disk or CPU crashes; they can
          continue to run in a VAXcluster environment if a CPU
          is off-line; and they can continue to run even if part
          of the database is off-line.

          In all of these areas, Oracle is playing catch-up to
          VAX Rdb/VMS. Rdb has had VAXcluster automatic recovery
          capabilities since V2.0, released November, 1985. Keep
          in mind that Oracle is releasing the first version of
          this functionality.


          3.2  Online backup and recovery

          Oracle now has online backup, the ability to backup
          a database while users are still running database
          applications. However, Oracle uses the native backup
          command provided by the host operating system. On a
          VAX, this is VMS backup. VAX Rdb uses its own backup
          format that is based upon the structure of an Rdb
          database for greater speed.

          In addition to online backup, VAX Rdb also does
          incremental backup. Incremental backup increases
          backup speed and ease of maintenance by reducing the
          amount of data that needs to be backed-up.

          We have taken advantage of our position as both the
          hardware and SOFTWARE vendor to provide the fastest
          and most reliable backup capabilities of any database
          system on the VAX. We can backup multiple areas
          (similar to Oracle's tablespace) at the same time,
          concurrently streaming three tape drives. Rdb still
          wins the maintenance battle.


                                                                3

 







          3.3  High performance

          Oracle would have you believe that they are the
          only high performance database system. Their recent
          testing, however, doesn't prove it. The TP1 test, as
          they ran it, is meaningless and should not be confused
          with the Debit/Credit test that Digital ran with Rdb.

          It is very difficult to rely upon any of Oracle's
          performance numbers and the conclusions that they draw
          from them (such as linear scaling on SMP machines).
          They generated all the transaction from memory (no
          users), there was no communication costs, and no
          application costs. No auditor with experience with
          relational databases would verify their tests as
          conforming to the Debit/Credit standard.

          There are many tricks that a vendor can play to
          generate high TPS rates. It surprises me that
          Oracle's are as low as they are. In an environment
          where IO, locking, and communication overheads can
          be eliminated, it is no great surprise that high
          performance and linear scaling can be observed. Give
          no credence to the Oracle results.

          Rdb V3.0 customers, however, have been very pleased
          with the performance of their new version. Check
          announcement materials for more details on all the
          benefits of V3.0. Also use the RDB_VMS_COMPETITION
          notes file and recent editions of Competitive Update
          for detail. (Competitive Update: July 12, 1988; August
          22, 1988. Sales Update: July 12, 1988).












          4

 







          3.4  Fast Commit

          Fast commit allows transactions to commit without
          writing the data pages to the database. The changed
          data pages are stored in memory in the expectation
          that future transactions will access the same page.
          This is a method for reducing IO per transaction.

          If system failure occurs, this method uses the log
          file to redo committed transactions. Often, this will
          cause much longer recovery times than Rdb.

          In large, OLTP, databases this feature probably has
          no usefulness unless the user of this feature just
          wants to show good benchmark results. Large OLTP
          applications will touch many different pages causing
          any cached pages to be written to disk. In this
          scenario fast commit will have no significant impact.

          Fast commit does help if you run benchmarks for short
          duration so that no IO is ever done (all the dirty
          pages stay in memory until after the benchmark is
          run). Fast commit can help on small databases where
          the database can reside in real memory. And, fast
          commit can help if access to the database is not
          random. The extent that fast commit helped Oracle
          in their benchmark is probably a measure of their
          divergence from the Debit/Credit benchmark standard.


          3.5  Row level locking

          With TPS, Oracle has introduced row level locking for
          the first time. In previous releases Oracle claimed
          that they had row level locking. In reality, they
          locked the entire table for update. There may be other
          features or benefits that Oracle claims that they have
          that they haven't yet implemented.

          Rdb has had row level locking since version 1.0. Rdb
          has been the leader in providing high performance
          techniques for OLTP.



                                                                5

 







          3.6  Read-consistency model

          Oracle likes to claim that their read-consistency
          model is unique. In a sense it is. It is the only one
          that forces all reads through a single before image
          log.

          Rdb has had read-consistency through SNAPSHOT files
          since version 1.0 in 1984. Rdb, however, provides a
          great deal more flexibility in its implementation.
          Perhaps the most important difference is that you can
          have separate snapshot files for each database area
          and can place these snapshots on different disks to
          eliminate the file contention that would result with
          the Oracle implementation.


          3.7  DBA tools

          Oracle is finally trying to make the running of an
          Oracle installation easier. They have created a single
          interface that allows the startup and shutdown of
          the six or so processes that must run in order to
          support even a single user (at least four of these are
          detached processes that support multiple users).

          Their DBA tool also allows some small amount of
          database monitoring capability: reads, writes,
          commits, locks, IO, etc. The granularity of these
          items is not clear. Some of the information is only
          available on a per-process or per database basis.

          Rdb has recognized the importance of simple database
          administration for a long time. That is why Rdb
          has the Rdb Monitoring Utility (RMU). RMU provides
          sophisticated statistics and graphs to help monitor
          and then tune the database for maximum performance.








          6

 







          3.8  PL/SQL

          PL/SQL is an Ada-like language to which users must
          convert in order to get high performance from
          Oracle. Oracle presentations go into great length
          on the conversion of SQL to PL/SQL. According to
          the president of Oracle, PL/SQL gives the user a 15%
          performance increase over regular Oracle.

          By using PL/SQL the application is no longer based
          upon the SQL standard and is no longer portable across
          DBMSs. The customer is tied to Oracle.

          Compare this to Rdb where high performance is built
          into the product and not through language enhancements
          (PL/SQL) and add-ons (TPS). In addition, high
          performance Rdb applications can be written in SQL
          and retain portability across database platforms.

          Although Oracle may try to sell this product as a
          TP monitor, it is not. In addition, it is not clear
          how well (if at all) this product will work with
          existing TP monitors offered by Digital. See the
          article "Selling OLTP Against Oracle" in XXX for more
          information.


          4  Summary

          On July 18, 1988, Oracle announced a number of new
          features that improve Oracle's DBMS performance.
          Most of these features require the purchase of a new
          product, TPS, that costs an additional $75,000 on a
          VAX 6240.

          The benchmarks that Oracle produced with this new
          software demonstrate very high TP1 performance but
          are not comparable to Digital's Debit/Credit benchmark
          results. Any comparisons between the two tests are
          meaningless.





                                                                7

 







          The new features that Oracle introduced have generally
          either been in Rdb for years or have been added to
          Rdb in version 3.0. In addition, Rdb generally has
          a more complete and flexible implementation that
          should provide users with higher performance than that
          available with Oracle systems.




































          8
204.2Selling OLTP against OracleBROKE::DREYFUSTue Sep 27 1988 00:45320
 







                                                    David Dreyfus
                                           Database Systems Group
                                                     DTN 381-2893
                                                   DEBIT::DREYFUS
          26-September-1988

                        Selling OLTP against ORACLE

          On July 18, 1988, Oracle announced a number of new
          features that improve Oracle's DBMS performance.
          Oracle now sells the idea that they can provide OLTP
          (On-Line Transaction Processing) solutions. The claim
          is supported by the announcement of TPS (Transaction
          Processing Subsystem), a $75,000 add-on for a VAX
          6240. In fact, TPS brings little to the customer that
          is not already available with VAX Rdb/VMS V3.0.
          The benchmarks that Oracle produced with this new
          software demonstrate very high TP1 performance but
          are not comparable to Digital's Debit/Credit benchmark
          results. Any comparisons between the two tests are
          meaningless. In fact, most of the OLTP characteristics
          of the Debit/Credit benchmark were ignored in Oracle's
          tests.

          Oracle can only satisfy a small portion of customers'
          OLTP needs. Digital has the architecture, products,
          and experience to support OLTP applications. Where
          Oracle is introducing first-time capabilities, Digital
          has mature products and many satisfied customers.













                                                                1

 







          1  Selling OLTP

          1.1  Capabilities

          Transaction processing (TP) is much more than driving
          high TPS (transactions per second) through a database
          system in a controlled benchmark environment. TP is
          driving high TPS through a complete computing system
          in a dynamic environment.

          The TP system much be able to handle many users using
          many different types of applications while still
          demanding high performance at minimal cost. In order
          to satisfy this environment, the database system must
          be designed to handle complex database designs and
          a wide variety of transaction types. This is the
          exact opposite of the environment described by the
          TP1 benchmark used by Oracle. It is also the type of
          environment that VAX Rdb is designed to handle.

          Other critical capabilities of a TP system are high
          levels of security for applications, security for
          users, dynamic tuning capabilities, and the ability
          to queue work for later processing. It is only
          through the use of a TP monitor that many of these
          capabilities are realized. These too are capabilities
          that Oracle lacks.

          As Digital improves its TP capabilities, important
          functions will be moved into the operating system.
          For the customer, it is important that their TP system
          continue to work as the underlying system services are
          enhanced. It is important that their database system
          continue to work with and take advantage of changes
          to the operating system. The level of integration
          required for TP systems is much greater than that
          required for ad-hoc reporting and 4GL applications.
          As the developer of the operating system, database
          system, TP monitor, forms, etc Digital is uniquely
          qualified to provide the required capabilities and
          integration for TP applications.



          2

 







          1.2  Experience

          Digital has been selling OLTP products and solutions
          for years (one consultant's estimates put Digital
          OLTP revenue at 1 billion dollars). Oracle is just
          entering the market. As a result, Oracle has a number
          of problems delivering OLTP solutions. For starters,
          TPS, which is required for all the features described
          above, is not yet available. After all the delays
          in getting V6.0 out, who knows how long it will take
          before TPS is delivered. In addition, there is some
          doubt about the reliability of Oracle products.

          In contrast, Rdb, Transaction Processing (TP) monitors
          (ACMS and DECintact), forms products (FMS and TDMS),
          and the other components for OLTP from Digital have
          been released for years. So, a big difference between
          Oracle and Digital in OLTP is experience.


          1.3  Architecture

          OLTP requires an architecture and products in order
          to deliver solid solutions that will protect the
          customers investment for years to come. Digital has
          an architecture (DECtp) and the products. Oracle has
          neither.

          All that Oracle can offer is a DBMS. It is not clear
          what forms package Oracle can offer. Their SQL*FORMS
          is not suitable for OLTP for lack of performance.
          It is not clear that PL/SQL introduces a new forms
          package or will work with any existing forms package.
          Oracle doesn't have a TP monitor. Thus, while they may
          provide efficient database services (no guarantee of
          this), they may not provide efficient application or
          form execution services.

          Digital can support hundreds of users on a MicroVAX
          with a product like ACMS. It is not clear that Oracle
          can. In addition, with a product like ACMS, forms
          processing can be distributed to multiple inexpensive
          CPUs for a distributed OLTP solution. TP monitors


                                                                3

 







          that integrate with Rdb allow a small amount of
          VAX resources to provide high performance and fast
          response times for hundreds of users.


          1.4  Development

          Oracle will probably require customers to use PL/SQL
          for any OLTP application. Otherwise, the communication
          costs will be excessive between applications and the
          database. This solution uses a 3GL (Ada-like PL/SQL)
          and is proprietary. An Rdb solution can use either
          a 3GL approach with DECintact or a 4GL approach
          with ACMS. In either case, database access is still
          provided by industry standard SQL.

          All the pieces of an OLTP solution should be tracked
          through a data dictionary. With CDD, Digital offers
          a distributed active data dictionary that tracks
          applications, databases and forms. CDD works with
          database objects and non-database (RMS) objects
          and acts as the integration point for many Digital
          products. Oracle is not nearly as advanced in this
          area, providing little more than the system catalogs
          every database management system provides.

          With Oracle, form, application, and database
          definitions cannot be easily shared between
          applications or databases. With VAX Rdb, definitions
          for forms, applications, and databases are centralized
          and shared.

          Another important part of OLTP is the management of
          the database itself. Oracle has tried to make this
          easier with their recent announcement. This includes
          on-line backup and a DBA tool.








          4

 







          1.5  Management

          OLTP databases are often complex and dynamic. They are
          complex because they support multiple types of users
          with a wide variety of transaction and record types.
          It is unfortunate that neither TP1 or Debit/Credit
          benchmarks measure this because it is in exactly this
          type of environment where Rdb excels.

          Unlike Oracle, Rdb provides a sophisticated monitoring
          and tuning tool with RMU. With RMU the DBA can monitor
          disk usage by table and partition data across the
          available disks to provide low IO rates and optimum
          performance. Oracle doesn't have the capability to
          control the placement of records from a single table
          on multiple disks. The key here is control. Oracle
          does have some capability for a table to use multiple
          disks. The question is, can the DBA determine what
          ranges of values go on each disk.

          Another important area of database management is the
          ability to control page sizes on disk and re-use pages
          as the database changes. Oracle uses fixed page sizes
          which may not necessarily be the correct size for
          maximum database performance. Although it may have
          changed in this new version, Oracle also doesn't allow
          space deleted from one table to be automatically used
          by other tables. Rdb, of course, allows both flexible
          page sizes and automatically reuses space.

          One of the biggest differences between Oracle and
          Rdb are in the area of space management. Oracle uses
          the IBM, DBA intensive, approaches where space must
          be preallocated for databases and tables. If table
          space fills up with Oracle, data must be unloaded and
          reloaded after new space is allocated. Rdb, of course,
          uses the Digital style of file management where space
          is allocated as needed and the database is extended
          dynamically.





                                                                5

 







          Rdb recognizes that OLTP is dynamic. You can't know
          how much space to allocate beforehand. You also don't
          want to allocate space that isn't going to be used
          right away. Finally, you don't want to hire people who
          spend all their time worrying about managing database
          disk space allocations.

          Oracle may be very easy to use from an end-user point
          of view. Most products are. The difficulty is managing
          the database and application development. It is in
          these more difficult areas that Digital excels.































          6
204.3Combating Oracle's Marketing StrengthsBROKE::DREYFUSTue Sep 27 1988 00:47486
 







                                                    David Dreyfus
                                           Database Systems Group
                                                     DTN 381-2893
                                                   DEBIT::DREYFUS
          26-September-1988

                   Combating Oracle's Marketing Strengths

          Like the Sirens of Greek mythology, Oracle has a few
          high level messages that can be difficult to overcome
          when selling against them. The difficulty increases
          considerably when Oracle has had a chance to set
          expectations in the account.

          The messages are quite seductive until you really
          understand what it is that Oracle is saying and, more
          importantly, not saying. We must show the customer
          that Digital understands what is seductive about
          Oracle's promises and can deliver the promise better
          than Oracle.

          This involves showing the customer that Oracle is not
          actually living up to the expectations set by their
          messages and, in fact, that a solution based upon
          Oracle may be counter-productive.


          1  Portability

          The "Oracle is portable, runs on many machines,
          and provides hardware independence" argument goes
          to the heart of the software/hardware dependency
          issue. Oracle uses prospects' belief that they a)
          want to be independent of their hardware vendor and
          that b) Oracle can provide this independence. By
          working across most computing platforms, Oracle tells
          prospects that they can move their applications to the
          newer, faster hardware platforms.






                                                                1

 







          Rdb, by contrast, only works on VAX/VMS and therefore
          locks the prospect into VAXes - or so goes Oracle's
          argument. These are difficult arguments to overcome.
          Sometimes they take on a religious nature. There are,
          however, a number of ways to respond.


          1.1  Application layer portability

          The real cost to users is application development
          time. Users want to be able to move their applications
          to new hardware without rewriting it. Thus, customers
          want to write software that is hardware independent in
          order to effect this portability requirement.

          This does not, however, require that the same system
          software run on each hardware platform. What is
          required is that system software provide an industry
          standard interface and that the tools used to produce
          the software do the same. With VAX SQL and the many
          industry standard languages it integrates with,
          Digital provides application portability. Most, if
          not all, the software vendors that provide "portable"
          software build it on VAXes with portable tools and
          languages offered by Digital.

          Oracle fails to explain (unsurprisingly) what is given
          up by using a major piece of system software (the
          DBMS) not provided by the hardware/operating system
          vendor.

          Rdb provides application portability and takes
          advantage of the richness of the underlying platform.
          In order to produce a piece of system software that
          runs on all platforms, Oracle fails to use a VAX
          to its greatest advantage. Oracle must build to the
          lowest common denominator (a PC DOS system?).

          By selecting Rdb as the database engine, the customer
          has both the best database for VAXes and application
          portability - a win-win situation.




          2

 







          1.2  Software vs hardware dependency

          The real threat to application portability and
          development is software dependency. Customers are
          dependent on the applicability of the tools they
          select in order to complete their development goals.
          They are dependent upon the continued maintenance,
          support, and improvements of database and application
          runtime systems for application execution.

          This dependency is OK as long as the tools and runtime
          systems meet customers' needs at an appropriate price.
          However, if the tools and runtime systems fail to meet
          expectations, customers will want to migrate their
          applications to a different software platform. It is
          during this migration that software dependency becomes
          a major issue.

          All 4GLs are proprietary and therefore cause a
          dependency. The difference is that Digital offers
          a standard environment for application development
          that doesn't require the use of a proprietary 4GL.
          Applications can be written in a standard language
          with SQL with calls to a forms package or built with
          the VAX COBOL Generator. Oracle applications that
          use a form interface require the use of SQL*FORMS,
          Oracle's proprietary 4GL.

          VAX Rdb supports a whole range of 4GL tools. This
          eases migration for applications that require new
          tools because both the new and old tools can continue
          to use the same database. If Oracle's tools are no
          longer effective, the customer must select a new tool
          from a very short list of third-party products or the
          customer must migrate to a new database system.

          In order to provide an environment that further
          minimizes the cost of software dependency, Digital
          supports broad, standardized interfaces to many
          important system services. A partial list includes
          SQL, RDA (remote database access), X-windows, PHIGS,



                                                                3

 







          and OSI. Oracle does not support nearly as many
          standards as Digital.

          Oracle customers are trapped by Oracle's proprietary
          software environment. Their SQL is not ANSI or DB2
          compliant. The addition of so many extensions to
          Oracle's SQL makes their SQL non-standard. In order
          to improve performance, Oracle's customers must use
          PL/SQL (which is both non-standard and incompatible
          with Oracle's own SQL). The tools used to build
          software (4GLs, SQL*FORMS, PL/SQL, dictionaries,
          precompilers, etc) are not industry standard.

          If we can get the customer to see the software
          dependency issue, we can offer the VAX Rdb platform as
          a solution. The lock-in is much less, the dependency
          is much less, and there is a far greater variety of
          tools and software with which to integrate and build.


          1.3  Broad based integration

          Oracle has a very narrow product set in comparison to
          that which is available from Digital. Digital has
          a wide variety of office automation, commercial,
          scientific, CASE, realtime, and OLTP software that
          integrates with Rdb. The level of integration and the
          variety of tools and applications will only increase
          in the future. A decision to go with Oracle is a
          decision to go with a narrowly defined product set
          and to forgo the richness of the VAX environment.













          4

 







          2  A PC product

          Another convincing Oracle argument is the availability
          of a PC based DBMS. As PCs become the desktop device
          of choice for many users, prospects want database
          software to run on the PC. Rdb, as you know, only runs
          on VAX/VMS.

          The first step in combating this argument is to
          assess what is really needed on the PC. Does the user
          really need the data to be stored on the PC or will
          applications and application development suffice? We
          need to convince the user that VAX stored data is OK
          as long as PCs can be used for application development
          and execution.

          The argument for not storing data on the PC is
          that the data can't be shared, data management is
          complicated, backup is difficult, and integrity is
          diminished. By storing data on the VAX, backups are
          easy, data can be shared, and high performance can be
          achieved.

          Using standard languages and DECnet-DOS, users can
          build applications on the PC for PC use or VAX
          use that connect to a VAX and Rdb. The "Colonial
          Management Testimonial" (available at a Lit. Center
          near you) talks about one company's experience doing
          just this.

          With SQL Services, program announced July 19, 1988,
          application builders will have a direct SQL connection
          from their favorite workstation (PC, Unix, DECwindow,
          etc) to Rdb. SQL services will provide a client-server
          environment to PC database users. A VAX Rdb server on
          VMS will handle the needs of multiple PCs quickly and
          efficiently. Applications on the PC will be written
          to a standard API (application programming interface)
          so that the same application can be executed on any
          supported workstation without change.




                                                                5

 







          In addition, there are existing third-party products
          providing PC to Rdb connectivity. Currently, Multiplex
          (Network Innovations) allows users to access VAX
          Rdb and move data into multiple PC data formats.
          Powerhouse (Cognos) and Today (BBJ Computers) allow
          users to build applications on their PCs that access
          VAX Rdb.


          3  Integrated SQL toolset

          Oracle likes to show how all their tools talk SQL to
          the Oracle DBMS. This gives prospects the impression
          that Oracle is industry standard. This argument should
          fail with sophisticated prospects; but, they don't
          make all the decisions.

          The argument can be attacked in a variety of ways.


          3.1  It's not SQL

          SQL is a data definition and manipulation language.
          As such, it is not capable of providing a complete
          interface between applications, tools, etc and
          the DBMS. Any extensions made to SQL to allow this
          connection are, by definition, not standard. That they
          may be called SQL is marketing hype. The user cannot
          take Oracle tools that talk SQL and connect them to
          another SQL database (say Rdb) and expect a working
          product.













          6

 







          3.2  Third-party integrated tools

          One of the great advantages of Rdb is that it is
          accessible by a large variety of third-party tools.
          These tools are often better integrated than Oracle's
          and work very well with Rdb. Many of them even speak
          SQL and many of them work with RMS.

          VAX Rdb's ability to support a large number of tools
          increases Rdb's value to the customer. While an
          Oracle customer is limited by Oracle's development
          tools, an Rdb customer has a very wide range of tools.
          Sometimes it takes more than one type of tool to solve
          a problem.


          3.3  Separate the DBMS and tools

          The key to selling against Oracle may be to divide
          and conquer. We need to get the user to consider the
          DBMS separately from the 4GLs and tools. Oracle will
          get their customers to buy their database system on
          the strength of their tools. Oracle demonstrates their
          tools - it is what users will see first.

          We need to show that Rdb supports an excellent
          tool set with products from Digital and third-party
          companies. We also need to impress upon the customer
          the importance of their DBMS selection. Position
          Oracle as a single-tool, 4GL company with a database.

          If both system administration and end-user groups can
          be involved in the product evaluation, the fact that
          there are really two different purchases may become
          clear. We can then sell the end-users on the idea
          that Digital can find them a wide variety of excellent
          tools. We can sell the system administration people
          Rdb's ease of maintenance, tunability, and performance
          features. We can satisfy both groups by selling a more
          complete package than Oracle.





                                                                7

 







          4  Performance

          Oracle has jumped on the OLTP bandwagon with their
          version 6.0 and TPS (transaction processing subsystem)
          announcement. As part of this announcement, Oracle
          released the results of their TP1 benchmark. Oracle
          is trying to use this benchmark to confuse the market
          about the differences between TP1 and Debit/Credit.
          They are also trying to confuse the market about their
          relative performance with VAX Rdb.

          The features added to version 6.0 and TPS are also in
          VAX Rdb. However, where Oracle has placed much of its
          functionality in add-ons (TPS) and languages (PL/SQL),
          which requires applications to be rewritten to take
          advantage of the enhancements, VAX Rdb has performance
          built into the product so that applications can
          continue to work without change.

          Unlike Digital's Debit/Credit benchmark, Oracle did
          no front-end communication, no mirrored journaling, no
          simulated terminal networking (transactions initiated
          from main memory), etc. According to consultant Omri
          Serlin, "Oracle has gone to such an extreme in what
          they have done. It had no on-line flavor whatsoever."
          The author, referring to the material, says, "Yes,
          ladies and gentlemen, it was a thick document, but we
          should read it as well as weigh it."

          For a more detailed analysis of the Oracle
          announcement, read the report titled "Review of
          Oracle's V6.0 with TPS Announcement" in XXX.












          8

 







          5  The 4GL demo

          Oracle uses their 4GLs and tools to sell the database
          system. They talk SQL and portability to comfort the
          user into an Oracle purchase. They use the 4GL to show
          ease of use and productivity.

          Customers who purchase Oracle because of the 4GL walk
          into Oracle's trap. In an effort to reach some level
          of independence, they sacrifice all independence by
          getting locked into a proprietary Oracle software
          environment.

          The first way of addressing this issue is to outline
          this progression to the customer. Help them "see
          the light" and separate the DBMS and 4GL purchase.
          Through a combination of standardized hardware and
          software platforms as well as partnerships with third-
          pary companies, Digital can maximize application
          portability and minimize the customer's involvement
          with proprietary dependencies.






















                                                                9