1 |
the OpenIsis database at a glance |
2 |
|
3 |
|
4 |
* characteristics |
5 |
|
6 |
- highly flexible data structure: |
7 |
potentially unlimited number of data fields ("columns") in record |
8 |
- highly efficient storage: |
9 |
unused data fields consume no space |
10 |
- natural modelling - ultra fast access: |
11 |
logically related data that would be artificially separated in |
12 |
a relational DB is stored in a single record |
13 |
- highly flexible index structure: |
14 |
index entries associated with a record are under full application control, |
15 |
can even be derived from associated text documents of any format. |
16 |
|
17 |
|
18 |
* standards |
19 |
|
20 |
- Z39.2 (IIF) based data structure |
21 |
record structure corresponds to ANSI Z39.2 |
22 |
(ISO 2709, "Information Interchange Format") |
23 |
- Z39.50 based retrieval |
24 |
the query language resembles Z39.50 "Type-1" queries as used by |
25 |
all major libraries (known from WAIS and most search engines). |
26 |
- all bibliographic data models |
27 |
like MAB and the various MARC standards can be represented in IIF |
28 |
and thus in OpenIsis |
29 |
- XML data |
30 |
is canonically mapped to IIF for highly efficient storage and retrieval |
31 |
|
32 |
|
33 |
* interfaces |
34 |
|
35 |
- variety of programming languages |
36 |
Java, Perl, Tcl, PHP, C-library for all compiler languages |
37 |
- web applications |
38 |
using Java, Perl or PHP as wanted |
39 |
- graphical user interface |
40 |
for Windows and Unix (incl. Mac OS X) based on Tk |
41 |
- usage in-process or as server |
42 |
- Z39.50 protocol |
43 |
server planned using indexdata's Yaz |
44 |
|
45 |
|
46 |
* performance |
47 |
|
48 |
benchmarks on an 800MHz Pentium III with standard IDE disk |
49 |
- 100.000 records/sec |
50 |
read in random order |
51 |
- 50.000 entries/sec |
52 |
index lookup |
53 |
- 15.000 records/sec |
54 |
write access with no index entries |
55 |
- 3.000 records/sec |
56 |
write access with ten index entries per record |
57 |
|
58 |
|
59 |
* limits |
60 |
|
61 |
There is a choice between traditional and 21st century database formats. |
62 |
- size of data field and record |
63 |
traditional: 32KB, 21st: no limit |
64 |
- number of records and database size |
65 |
traditional: 16 million records of 500 MB total size, 21st: no limit. |
66 |
logical database can span several physical databases (virtually no limit). |
67 |
- index size |
68 |
traditional: ~1 TB, 21st: 32 TB per physical index. |
69 |
logical index can span several physical indexes (virtually no limit). |