Software and hardware configuration

The following tables try to sum up all available possibilities, and give some benchmark (average objects/second for writing or reading).

In these tables:
- 'SQLite3 (file full/off/exc)' indicates use of the internal SQLite3 engine, with or without Synchronous := smOff and/or DB.LockingMode := lmExclusive;
- 'SQLite3 (mem)' stands for the internal SQLite3 engine running in memory;
- 'SQLite3 (ext ...)' is about access to a SQLite3 engine as external database , either as file or memory;
- 'TObjectList' indicates a TSQLRestServerStaticInMemory instance, either static (with no SQL support) or virtual (i.e. SQL featured via SQLite3 virtual table mechanism) which may persist the data on disk as JSON or compressed binary;
- 'NexusDB' is the free embedded edition, available from official site;
- 'Jet' stands for a MSAccess database engine, accessed via OleDB.
- 'Oracle' shows the results of our direct OCI access layer (SynDBOracle.pas);
- 'Zeos *' indicates that the database was accessed directly via the ZDBC layer;
- 'FireDAC *' stands for FireDAC library;
- 'UniDAC *' stands for UniDAC library;
- 'BDE *' when using a BDE connection;
- 'ODBC *' for a direct access to ODBC.

This list of database providers is to be extended in the future. Any feedback is welcome!

Numbers are expressed in rows/second (or objects/second). This benchmark was compiled with Delphi XE4, since newer compilers tends to give better results, mainly thanks to function in-lining (which was not existing e.g. in Delphi 6-7).

Note that these tests are not about the relative speed of each database engine, but reflect the current status of the integration of several DB libraries within the mORMot database access.

Benchmark was run on a Core i7 notebook, running Windows 7, with a standard SSD, including anti-virus and background applications:
- Linked to a shared Oracle database over 100 Mb Ethernet;
- MS SQL Express 2008 R2 running locally in 64 bit mode;
- IBM DB2 Express-C edition 10.5 running locally in 64 bit mode;
- PostgreSQL 9.2.7 running locally in 64 bit mode;
- MySQL 5.6.16 running locally in 64 bit mode;
- Firebird embedded in revision 2.5.2;
- NexusDB 3.11 in Free Embedded Version.

So it was a development environment, very similar to low-cost production site, not dedicated to give best performance. During the process, CPU was noticeable used only for SQLite3 in-memory and TObjectList - most of the time, the bottleneck is not the CPU, but the storage or network. As a result, rates and timing may vary depending on network and server load, but you get results similar to what could be expected on customer side, with an average hardware configuration. When using high-head servers and storage, running on a tuned Linux configuration, you can expect even better numbers.

Tests were compiled with the Delphi XE4 32 bit mode target platform. Most of the tests do pass when compiled as a 64 bit executable, with the exception of some providers (like Jet), not available on this platform. Speed results are almost the same, only slightly slower; so we won't show them here.

You can compile the "15 - External DB performance" supplied sample code, and run the very same benchmark on your own configuration.
Feedback is welcome!

From our tests, the UniDAC version we were using had huge stability issues when used with DB2: the tests did not pass, and the DB2 server just hang processing the queries, whereas there was no problem with other libraries. It may have been fixed since, but you won't find any "UniDAC DB2" results in the benchmark below in the meanwhile.

Insertion speed

Here we insert 5,000 rows of data, with diverse scenarios:
- 'Direct' stands for a individual Client.Add() insertion;
- 'Batch' mode will be described 28;
- 'Trans' indicates that all insertion is nested within a transaction - which makes a great difference, e.g. with a SQlite3 database.

  Direct Batch Trans Batch Trans
SQLite3 (file full) 462 356 95377 130086
SQLite3 (file off) 844 821 100389 136675
SQLite3 (file off exc) 28847 35316 102599 144258
SQLite3 (mem) 89456 120513 104249 146933
TObjectList (static) 314465 543892 326370 542652
TObjectList (virtual) 325393 545672 298846 545018
SQLite3 (ext full) 424 11297 102049 164636
SQLite3 (ext off) 830 21406 109706 189250
SQLite3 (ext off exc) 41589 180759 108481 192071
SQLite3 (ext mem) 101440 234576 113530 190142
ODBC SQLite3 492 11746 35367 82425
ZEOS SQlite3 494 11851 56206 85705
FireDAC SQlite3 26369 50306 49755 155115
UniDAC SQlite3 477 8725 26552 38756
ODBC Firebird 1495 18056 13485 17731
ZEOS Firebird 9733 13429 26348 30616
FireDAC Firebird 24233 52021 24791 52111
UniDAC Firebird 5986 14809 6522 14948
Jet 4235 4424 4954 5094
NexusDB 5998 15494 7687 18619
Oracle 226 56112 1133 52367
ODBC Oracle 236 1664 1515 7709
FireDAC Oracle 118 48575 1519 12566
UniDAC Oracle 164 5701 1215 2884
BDE Oracle 489 927 839 1022
MSSQL local 5246 54360 12988 62453
ODBC MSSQL 4911 18652 11541 20976
FireDAC MSSQL 5016 7341 11686 51242
UniDAC MSSQL 4392 29768 8649 33464
ODBC DB2 4792 48387 14085 70104
FireDAC DB2 4452 48635 11014 52781
ZEOS PostgreSQL 4196 26663 9689 38735
ODBC PostgreSQL 4068 19515 5130 27843
FireDAC PostgreSQL 4181 37000 10111 36483
UniDAC PostgreSQL 2705 18563 4442 22317
ODBC MySQL 3160 38309 10856 47630
ZEOS MySQL 3426 34037 12217 40186
FireDAC MySQL 3078 43053 10955 45781
UniDAC MySQL 3119 27772 11246 33288

Due to its ACID implementation, SQLite3 process on file waits for the hard-disk to have finished flushing its data, therefore it is the reason why it is slower than other engines at individual row insertion (less than 10 objects per second with a mechanical hardrive instead of a SDD) outside the scope of a transaction.

So if you want to reach the best writing performance in your application with the default engine, you should better use transactions and regroup all writing into services or a BATCH process. Another possibility could be to execute DB.Synchronous := smOff and/or DB.LockingMode := lmExclusive at SQLite3 engine level before process: in case of power loss at wrong time it may corrupt the database file, but it will increase the rate by a factor of 50 (with hard drive), as stated by the "off" and "off exc" rows of the table. Note that by default, the FireDAC library set both options, so results above are to be compared with "SQLite3 off exc" rows.

For both our direct Oracle access SynDBOracle.pas unit and FireDAC library , BATCH process benefits of the array binding feature a lot (known as Array DML in FireDAC/AnyDAC).

For most engines, our ORM kernel is able to generate the appropriate SQL statement for speeding up bulk insertion. For instance:
- SQlite3, MySQL, PostgreSQL, MSSQL 2008, DB2, MySQL or NexusDB handle INSERT statements with multiple INSERT INTO .. VALUES (..),(..),(..)..;
- Oracle handles INSERT INTO .. INTO .. SELECT 1 FROM DUAL (weird syntax, isn't it?);
- Firebird implements EXECUTE BLOCK.

As a result, some engines show a nice speed boost when BatchAdd() is used. Even SQLite3 is faster when used as external engine, in respect to direct execution! This feature is at ORM/SQL level, so it benefits to any external database library. Of course, if a given library has a better implementation pattern (e.g. our direct Oracle or FireDAC with native array binding), it is used instead.

Reading speed

Now the same data is retrieved via the ORM layer:
- 'By one' states that one object is read per call (ORM generates a SELECT * FROM table WHERE ID=? for Client.Retrieve() method);
- 'All *' is when all 5000 objects are read in a single call (i.e. running SELECT * FROM table from a FillPrepare() method call), either forced to use the virtual table layer, or with direct static call.

Here are some reading speed values, in objects/second:

  By one All Virtual All Direct
SQLite3 (file full) 27284 558721 550842
SQLite3 (file off) 26896 549450 526149
SQLite3 (file off exc) 128077 557537 535905
SQLite3 (mem) 127106 557537 563316
TObjectList (static) 300012 912408 913742
TObjectList (virtual) 303287 402706 866551
SQLite3 (ext full) 135380 267436 553158
SQLite3 (ext off) 133696 262977 543065
SQLite3 (ext off exc) 134698 264186 558596
SQLite3 (ext mem) 137487 259713 557475
ODBC SQLite3 19461 136600 201280
ZEOS SQlite3 33541 200835 306955
FireDAC SQlite3 7683 83532 112470
UniDAC SQlite3 2522 74030 96420
ODBC Firebird 3446 69607 97585
ZEOS Firebird 20296 91974 107229
FireDAC Firebird 2376 46276 56269
UniDAC Firebird 2189 66886 88102
Jet 2640 166112 258277
NexusDB 1413 120845 208246
Oracle 1558 120977 159861
ODBC Oracle 1620 43441 45764
FireDAC Oracle 1231 42149 54795
UniDAC Oracle 688 27083 30093
BDE Oracle 860 3870 4036
MSSQL local 10135 210837 437905
ODBC MSSQL 12458 147544 256502
FireDAC MSSQL 3776 72123 94091
UniDAC MSSQL 2505 93231 135932
ODBC DB2 7649 84880 124486
FireDAC DB2 3155 71456 88264
ZEOS PostgreSQL 8833 158760 223583
ODBC PostgreSQL 10361 85680 120913
FireDAC PostgreSQL 2261 58252 79002
UniDAC PostgreSQL 864 86900 122856
ODBC MySQL 10143 65538 82447
ZEOS MySQL 2052 171803 245772
FireDAC MySQL 3636 75081 105028
UniDAC MySQL 4798 99940 146968

The SQLite3 layer gives amazing reading results, which makes it a perfect fit for most typical ORM use. When running with DB.LockingMode := lmExclusive defined (i.e. "off exc" rows), reading speed is very high, and benefits from exclusive access to the database file. External database access is only required when data is expected to be shared with other processes.

In the above table, it appears that all libraries based on DB.pas are slower than the others for reading speed. In fact, TDataSet sounds to be a real bottleneck, due to its internal data marshalling. Even FireDAC, which is known to be very optimized for speed, is limited by the TDataSet structure. Our direct classes, or even ZEOS/ZDBC performs better, since they are able to output JSON content with no additional marshalling.

For both writing and reading, TObjectList / TSQLRestServerStaticInMemory engine gives impressive results, but has the weakness of being in-memory, so it is not ACID by design, and the data has to fit in memory. Note that indexes are available for IDs and stored AS_UNIQUE properties.

As a consequence, search of non-unique values may be slow: the engine has to loop through all rows of data. But for unique values (defined as stored AS_UNIQUE), both insertion and search speed is awesome, due to its optimized O(1) hash algorithm - see the following benchmark, especially the "By name" row for "TObjectList" columns, which correspond to a search of an unique RawUTF8 property value via this hashing method.

SQLite3 (file full) SQLite3 (file off) SQLite3 (mem) TObjectList (static) TObjectList (virt.) SQLite3 (ext file full) SQLite3 (ext file off) SQLite3 (ext mem) Oracle Jet
By one 10461 10549 44737 103577 103553 43367 44099 45220 901 1074
By name 9694 9651 32350 70534 60153 22785 22240 23055 889 1071
All Virt. 167095 162956 168651 253292 118203 97083 90592 94688 56639 52764
All Direct 167123 144250 168577 254284 256383 170794 165601 168856 88342 75999

Above table results were run on a Core 2 duo laptop, so numbers are lower than with the previous tables.

During the tests, internal caching was disabled, so you may expect speed enhancements for real applications, when data is more read than written: for instance, when an object is retrieved from the cache, you achieve more than 1,00,000 read requests per second, whatever database is used.

Analysis and use case proposal

When declared as virtual table (via a VirtualTableRegister call), you have the full power of SQL (including JOINs) at hand, with incredibly fast CRUD operations: 100,000 requests per second for objects read and write, including serialization and Client-Server communication!

Some providers are first-class citizens to mORMot, like SQLite3, Oracle, MS SQL, PostgreSQL, MySQL or IBM DB2. You can connect to them without the bottleneck of the DB.pas unit, nor any restriction of your Delphi license (a Starter edition is enough).

First of all, SQLite3 is still to be considered, even for a production server. Thanks to mORMot's architecture and design, this "embedded" database could be used as main database engine for a client-server application with heavy concurrent access - if you have doubts about its scaling abilities, see this blog article. Here, "embedded" is not restricted to "mobile", but sounds like a self-contained, zero-configuration proven engine.

Most recognized closed source databases are available:
- Direct access to Oracle gives impressive results in BATCH mode (aka array binding). It may be an obligation if your end-customer stores already its data in such a server, for instance, and want to leverage the licensing cost of its own IT solution. Oracle Express edition is free, but somewhat heavy and limited in terms of data/hardware size (see its licensing terms);
- MS SQL Server, directly accessed via OleDB (or ODBC) gives pretty good timing. A MS SQL Server 2008 R2 Express instance is pretty well integrated with the Windows environment, for a very affordable price (i.e. for free) - the LocalDB (MSI installer) edition is enough to start with, but also with data/hardware size limitation, just like Oracle Express;
- IBM DB2 is another good candidate, and the Express-C ("C" standing for Community) offers a no-charge opportunity to run an industry standard engine, with no restriction on the data size, and somewhat high hardware limitations (16 GB of RAM and 2 CPU cores for the latest 10.5 release) or enterprise-level features;
- NexusDB may be considered, if you have existing Delphi code and data - but it is less known and recognized as the its commercial competitors.

Open Source databases are worth considering, especially in conjunction with an Open Source framework like mORMot:
- MySQL is the well-known engine used by a lot of web sites, mainly with LAMP (Linux MySQL Apache PHP) configurations. Windows is not the best platform to run it, but it could be a fairly good candidate, especially in its MariaDB fork, which sounds more attractive those days than the official main version, owned by Oracle;
- PostgreSQL is an Enterprise class database, with amazing features among its Open Source alternatives, and really competes with commercial solutions. Even under Windows, we think it is easy to install and administrate, and uses less resource than the other commercial engines.
- Firebird gave pretty consistent timing, when accessed via Zeos/ZDBC. We show here the embedded version, but the server edition is worth considering, even if it has a less

To access those databases, OleDB, ODBC or ZDBC providers may also be used, with direct access. But mORMot is very open-minded: you can use any DB.pas provider, e.g. FireDAC, UniDAC, DBExpress, NexusDB or even the BDE, but with the additional layer introduced by using a TDataSet instance, at reading.

Therefore, the typical use may be the following (int. meaning "internal", ext. for "external", mem for "in-memory"):

Database Use case
int. SQLite3 file Created by default.
General safe data handling, with amazing speed in "off exc" mode
int. SQLite3 mem Created with :memory: file name.
Fast data handling with no persistence (e.g. for testing or temporary storage)
TObjectList static Created with StaticDataCreate.
Best possible performance for small amount of data, without ACID nor SQL
TObjectList virtual Created with VirtualTableRegister.
Best possible performance for SQL over small amount of data (or even unlimited amount under Win64), if ACID is not required nor complex SQL
ext. SQLite3 file Created with VirtualTableExternalRegister
External back-end, e.g. for disk spanning
ext. SQLite3 mem Created with VirtualTableExternalRegister
Fast external back-end (e.g. for testing)
ext. Oracle / MS SQL / DB2 / PostgreSQL / MySQL / Firebird Created with VirtualTableExternalRegister
Fast, secure and industry standard back-ends; data can be shared outside mORMot
ext. NexusDB Created with VirtualTableExternalRegister
The free embedded version let the whole engine be included within your executable, and use any existing code, but SQlite3 sounds like a better option
ext. Jet Created with VirtualTableExternalRegister
Could be used as a data exchange format (e.g. with Office applications)
ext. Zeos Created with VirtualTableExternalRegister
Allow access to several external engines, with direct Zeos/ZDBC access which will by-pass the DB.pas unit and its TDataSet bottleneck - and we will also prefer an active Open Source project!
ext. FireDAC/UniDAC Created with VirtualTableExternalRegister
Allow access to several external engines, including the DB.pas unit and its TDataSet bottleneck

Whatever database back-end is used, don't forget that mORMot design will allow you to switch from one library to another, just by changing a TSQLDBConnectionProperties class type. And note that you can mix external engines, on purpose: you are not tied to one single engine, but the database access can be tuned for each ORM table, according to your project needs.

Feedback is welcome in our forum, as usual.