summaryrefslogtreecommitdiff
path: root/src/site/apt/index.apt
blob: 197d3d2b2922f79ecaef0092c6a6409fdd0ca4ed (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
 -----
 Home
 -----

Carbonado

    Carbonado is an extensible, high performance persistence abstraction layer
for Java applications, providing a relational view to the underlying
persistence technology. Persistence can be provided by a JDBC accessible SQL
relational database, or it can be a BDB. It can also be fully replicated
between the two.

    Even if the backing database is not SQL based, Carbonado still supports
many of the core features found in any kind of relational database. It supports
queries, joins, indexes, and it performs query optimization. When used in this
way, Carbonado is not merely a layer to a relational database, it <is> the
relational database. SQL is not a requirement for implementing relational
databases.

    Defining new types in Carbonado involves creating an interface or abstract
class which follows Java bean conventions. Additional information is specified
by inserting special annotations. At the very least, an annotation is required
to specify the primary key. Annotations are feature first available in Java 5,
and as a result, Carbonado depends on Java 5.

    On the surface, it may appear that Carbonado types are defined like POJOs.
The difference is that in Carbonado, types are object representations of
relations. It is not an object database nor an object-relational bridge. In
addition, data type defintions are simply interfaces, and there are no external
configuration files. All the code to implement types is auto-generated, yet
there are no extra build time steps.

    Carbonado is able to achieve high performance by imposing very low overhead
when accessing the actual storage. Low overhead is achieved in part by auto
generating performance critical code, via the open source Cojen library.

* Terminology

    Loose mapping from Carbonado terminology to SQL terminology:

*---------------------+-------------------------+
| <<Carbonado>>       | <<SQL>>                 |
*---------------------+-------------------------+
| Repository          | database                |
*---------------------+-------------------------+
| Storage             | table                   |
*---------------------+-------------------------+
| Storable definition | table definition        |
*---------------------+-------------------------+
| Storable instance   | table row               |
*---------------------+-------------------------+
| property            | column                  |
*---------------------+-------------------------+
| Query               | select/delete statement |
*---------------------+-------------------------+
| Cursor              | result set              |
*---------------------+-------------------------+

* Limitations

    Carbonado queries are not as expressive as SQL selects. Unlike SQL,
Carbonado queries do not support things like sub-selects or data
processing. See Carbonado
{{{http://sourceforge.net/tracker/?group_id=171277&atid=857357}feature
requests}}.

    Carbonado supports the minimal querying capability that makes automatic
index selection possible. Other features available in SQL can be emulated in
code. If the database is local, then this offers no loss of performance.

    Applications that wish to use Carbonado only as a convenient layer over SQL
will not be able to use full SQL features. Carbonado is by no means a
replacement for JDBC. These kinds of applications may choose a blend of
Carbonado and JDBC. To facilitate this, access to the JDBC connection in use by
the current transaction is supported.

    The Carbonado repositories that are backed by BDB use a rule-based query
optimizer to come up with a query plan. Cost-based optimizers are generally
much more effective, since they estimate I/O costs. Carbonado has a rule-based
optimizer mainly because it is easier to write.

*  Persistence Technology Requirements

    Carbonado is capable of supporting many different kinds of persistence
technologies. A minimum set of features is required, however, in order to
provide enough Carbonado features to justify the effort:

    * Arbitrary keys and values

    * Get value by key

    * Put value by key (does not need to distinguish insert vs update)

    * Delete value by key

    * Ordered key iteration

    * Iteration start specified via full or partial key

    []

    Ideally, the persistence technology should support transactions. If it does
not, then its transactions must be implemented by batching updates in
memory. The updates are not persisted until the transaction is committed. If
atomic batch updates are supported, then the repository can report supporting
an isolation level of "read committed". Otherwise, it can only support the
lowest level of "read uncommitted".

    Additional features which are nice to have, but not strictly required:

    * Reverse iteration

    * ACID transactions

    * Storable type segregation (eliminates need to define key prefixes)

    * Truncation by storable type, if segregated 

    []