Hi,
I cannot be convinced it is a relational database. Simple joins in select
statements take ages to complete, exactly the thing rdms-es are build to do.
TSM uses a b-tree to store its data. There is a world of possibilities other
that rdbms or flat files. Tivoli has never stated exactly what db it is, but
Kelly will be happy to tell you that it is best desribed as a relation like
database. My guess is that an 'object database' is a better desription. Google
will be your friend on finding out what an object database is (an no, oracle
does not count as such, while it still has the word object added to what they
do).
I will not fight your argument that there are multiple tables in the db, and
that there are relations between those tables, but that does not make the
database a relational database. I can have a (hardcopy) cardfile with relations
to another cardfile, does that make it a rdms? Not really, don't you agree?
What counts is what the db's internal structures are designed to do, in TSM'
s case that is storing and retrieving objects refered to by some key, not
calculation relations between tables.
I'm very well aware that the select command has later been added to tsm, or
at least later exposed to us. This was done to fill the need with some admins
to do more complex queries that the standaard ones available. Unless someone
from tivoli speaks up about this matter, I'll stick to what Kelly has told
me and the behaviour I see in TSM.
On zaterdag, juli 20, 2002, at 01:15 , Zlatko Krastev wrote:
I hope you can be convinced TSM DB is pure relational and not flat files
with few SQL queries on top of them:
- Internal addressing of all data archived/backed up is through OBJECT_ID.
All links between the object (B/A client file, TDP backup/logs, Content
Manager documents, etc.), the object's name and physical placement on
volumes is through relations.
- SELECT queries are not some kind of add-on but integral part of TSM
server with very wide subset of DB2 features (field types, SQL functions,
etc.). If in doubt about any TSM SQL function look at its description in
DB2 manuals.
Have a look at nice post made by Roger Deschner on thread "tsm internal
design?" 6 weeks ago. His definition for WDSF/ADSM/TSM is very nice and
hits the bullseye. Thank you, Roger.
Lets say TSM DB is not DB2 but a cousin of it. Actually they are derived
from the same code. Also TSM DB physical layout is using same LVM code AIX
is using for filesystems and raw devices (but this is another story).
If you are interested in the history of the product (and try to dig down
IBM Announcement Letters archives) you can find TSM originate from
mainframe world:
1. '80s - IBM Data Facility Storage Management Subsystem (DFSMS for
System/370; still current and developed as DFSMS for S390/zSeries; very
good product but mainframe guys can comment it better than me)
2. '91 - IBM Workstation Data Save Facility/VM (VM server to backup PCs
running DOS)
3. '93 - IBM Data Facility Distributed Storage Manager (v1.0, from there
come all dsm* names)
4. '93 - in version 1.1 renamed to IBM ADSTAR Distributed Storage Manager
(ADSM)
5. 1994-2002 - You probably know the rest.
Same story for the DB2 - IBM had relational database (with SQL queries)
for System/370; later ported to RS/6000, S/390, AS/400 (integrated in the
OS/400); got new name in '90s as DB2; later become DB2 UDB. Same
research&concepts evolved from the old mainframes.
Zlatko Krastev
IT Consultant
---
Met vriendelijke groeten,
Met vriendelijke groeten,
Remco Post
SARA - Stichting Academisch Rekencentrum Amsterdam http://www.sara.nl
High Performance Computing Tel. +31 20 592 8008 Fax. +31 20 668 3167
"I really didn't foresee the Internet. But then, neither did the computer
industry. Not that that tells us very much of course - the computer industry
didn't even foresee that the century was going to end." -- Douglas Adams
|
