| |
This is a legally copyrighted work,
so none of it can,
legally or honorably,
be copied, used, or claimed by anyone other than the owner.
AxleBase Copyright 2003 - 2022 John E. Ragan.
Errors :
Simplicity and small size of this report hides the huge amount of thought in it. Notification of any error in it will sincerely be appreciated.
__________________________________________________
Limits
AxleBase is designed and engineered to the following limits.
| _______________ | _____________ | ___________ | ___ |
| object | elements | limit | note |
| _______________ | _____________ | ___________ | ___ |
| | | | |
| column width | bytes | 2 billion | |
| | | | |
| row | bytes | 2 billion | |
| | columns | 2 billion | |
| | | | |
| table | rows | 20 quintillion | [1] |
| | bytes | 20 exabytes | [1] |
| | computers | 10 billion | [2] |
| | DASD | 10 billion | [2] |
| | | | |
| table join | table sizes | no limit | [13] |
| | | | |
| database | tables | 20 million | [4] |
| | virtual tables | 20 million | |
| | files | no limit | |
| | bytes | no limit | [5] |
| | | | |
| virtual table | per database | 20 million | |
| | concatenations | no limit | |
| | | | |
| domain | databases | no limit | |
| | | | |
| installation | domains | no limit | |
| | databases | no limit | |
| | instances | no limit | |
| | | | |
| name length | characters | 80 | |
| | | | |
| index | columns | no limit | |
| | bytes | no limit | [7] |
| | rows | no limit | [7] |
| | | | |
| SQL statement | columns | no limit | |
| | characters | no limit | |
| | | | |
| query return | rows | no limit | [6] |
| | bytes | no limit | [6] |
| | | | |
| query sort | bytes | 2 billion | [12] |
| | rows | no limit | |
| | sort columns | no limit | |
| | | | |
| concurrency | clients | no limit | [8] |
| | | | |
| operating system | database | any | [10] |
| | controller | Ms. Windows | [9] |
| | super-system | Ms. Windows | |
| | | | |
| network | type | any | |
| _______________ | _____________ | ___________ | ___ |
Quantities are rounded to speed comprehension, and are rounded down to avoid deceit.
The 'no limit' may indicate one or more of the following :
A limit has not been conceived.
An engineering specification is not yet established.
Variables involved preclude precise specification.
The value may surpass practical requirements of the current state of Mankind.
The value may surpass current practical limits of computer technology.
Footnotes follow:
| ____ | _____________________________________________ |
| key | comment |
| | |
| [1] |
20 exabytes, or 20 quintillion, or 2 x (10^19), or
20,000,000,000,000,000,000. |
| [1] |
Bytes and rows are limited by the same mechanism. The
one that reaches the limit first will halt. |
| [1] |
Theoretically, the AxleBase architecture will allow
a large data increase. See
Theoretical Data limits
on the description document. |
| [2] |
The number of computers and/or storage devices that can
be used to store each table. Total cannot exceed that value.
Hardware, operating systems, and networks that house data objects are transparently managed by AxleBase. |
| [4] |
The number of tables in each database is limited to
the number of table configuration records that can be
stored in the table table, which is limited to two
gigabytes. Those records are variable length
of maybe a few hundred bytes. |
| [5] |
(Bytes per table) x (tables per database) = 4 x (10^25)
bytes. |
| [6] |
Largest tested dataset was an 8 gigabyte return that
held around a 100 million rows.
AxleBase uses RAM to build queries like conventional
systems, but a configuration toggle can tell him to use the
computer's disks. |
| [7] |
Index size can impact hardware-dependent performance. |
| [8] |
Although he might do it, high concurrency is not an AxleBase
objective. He is primarily a workhorse for carrying massive loads up to any conceivable size. See the design
objectives and concurrency tests. |
| [9] |
Tested on Windows 2000. |
| [10] |
AxleBase data can be distributed across any computers
that can express and share the Ms. Windows file system,
including mainframes, Linux, Unix, etc. |
| [12] |
Sort limit is for tuples, rows, elements, or bytes; whichever comes first.
Small sorts are quickly done
in RAM. Large sorts are done on disk, so twenty million
rows of two gigabytes can be much slower. |
| [13] |
Table joins are limited by the hardware and even that
limit is managed by AxleBase. Billion-row tables are
easily joined. A virtual table of trillions of rows
was joined to billion-row tables. |
| ____ | _____________________________________________ |
| | |
__________________________________________________
|
|
