by Archive-name: databases/sybase-faq/part13
URL: http://reality.sgi.com/pablo/Sybase_FAQ
Q12.1: SYBASE IQ
_________________________________________________________________
Sybase IQ isn't meant as just an indexing scheme, per se. It is meant
as a means of providing a low cost data warehousing solution for
unplanned queries.
By the way, Sybase IQ does not use bitmapped indexes, it uses bitwise
indexes, which are quite different.
In data warehousing MIS generally does not know what the queries are.
That also means that the end users often don't know what the queries
are. Not knowing what the queries are turning end users loose on a
500GB operational database to perform huge queries could prove to be
unacceptable (it may bring the system down a crawl). So, many
customers are resorting to separating their operational databases
(OLTP) and data warehousing databases. By providing this separation
the operational database can continue about its business and the data
warehouse users can issue blind queries without affecting the
operational systems. Realize that operational systems may handle
anywhere from hundreds to a few thousand users and, more likely than
not, require data that is highly accurate. However, data warehouse
users often don't require up to the second information and can often
wait several hours, 24 hours or even days for the most current
snapshot and generally don't require updates to be made to the data.
So, Sybase IQ can be updated a few times a day, once a day or a few
times a week. Realize that Sybase IQ is _strictly_ a data warehousing
solution. It is not meant for OLTP systems.
Sybase IQ can also sit on top of Sybase SQL Server:
[end user]
|
|
[Sybase IQ]
[Sybase SQL Server]
What happens in this environment is that a data warehouse user can
connect to Sybase IQ. Sybase IQ will then take care of processing the
query or forwarding the query to SQL Server if it determines that the
access paths in SQL Server are faster. An example where SQL Server
will be faster than Sybase IQ in queries is when SQL Server can
perform query coverate with the indexes built in SQL Server.
The obvious question is: why not index every column in SQL Server?
Because it would be prohibitive to update any of the data. Hence,
Sybase IQ, where all the columns are making use of the bitwise index
scheme. By the way, you can choose which columns will be part of an IQ
implementation. So, you may choose to have only 30% of your columns as
part of your Sybase IQ implementation. Again, I can't stress enough
that Sybase IQ is _strictly_ for data warehousing solutions, not OLTP
solutions.
_________________________________________________________________
Q12.2: NET BOOK REVIEW
_________________________________________________________________
* Sybase
* Sybase Architecture and Administration
* Developing Sybase Applications
* Sybase Developer's Guide
* Sybase DBA Survival Guide
* Guide to SQL Server
* Client/Server Development with Sybase
* Physical Database Design for Sybase SQL Server
* Sybase Performance Tuning
Sybase - McGoveran and Date
ISBN: 0-201-55710-X Published by Addison-Wesley. 450 pages.
I think that once, not too long ago, this used to be the only book on
Sybase available. Now it seems to be totally out of print! It covered
versions of Sybase SQL server up to 4.8. It covered a number of
aspects of Sybase, including APT.
Sybase Architecture and Administration - Kirkwood
ISBN: 0-13-100330-5 Published by Ellis Horwood. 404 pages.
This is a good book covering Sybase systems up to and including System
10. It deals to a good depth the architecture and how most of the
functions such as the optimiser work. It explains in a readable style
how devices work, and how indexes are stored and manipulated.
Developing Sybase Applications - Worden
ISBN: 0-672-30700-6 Published by SAMS. ??? pages. (Inc CD.)
This books seems very similar to number 4 to me and so I have not
bought it. I have browsed through several times in the book shop, and
decided that his other book covers a good deal of this. There are
chapters on Visual Basic and Powerbuilder.
Sybase Developer's Guide - Worden
ISBN: 0-672-30467-8 Published by SAMS. 698 pages. (Inc disk.)
This is a big book that does not, in my opinion, cover very much. In
fact the disk that is included contains DBATools, and that seems to
sum up the first 50% of the book. There is a fair amount of coverage
of the general architecture and how to install Sybase. Transact SQL,
cursors and stored procedures get a fair covering, as does using C/C++
with DB-Library. (I can find no mention of CT-Library.) Unfortunately
quite a lot of the book covers general issues which are not covered in
sufficient depth to be useful, and just seem to be there to give the
book bulk. Maybe as a developer's guide, his other book would be a
better buy. This would probably be most useful to a small company
implementing a Sybase database.
Sybase DBA Survival Guide - Jeff Garbus, David Solomon, Brian Tretter
ISBN: 0-672-30651-4 Published by SAMS. 506 pages. (Inc disk.)
This book is good, and is a great help in a crisis. It includes lots
of useful ideas and strategies for most (if not all) of the DBA tasks.
It covers Sybase SQL Server on all platforms. It does not specifically
cover any of the Microsoft versions, and certainly not version 6. It
does cover System 10. It is very good at explaining the output from
things like the DBCC commands. There is also a good section on what to
look for in the errorlog. If you are a DBA and want to buy just one
book, I would recommend this one since it covers just about everything
you will need to know. This book is filled with little hints, tips and
warnings which are _very_ useful. They have certainly saved my bacon
on a number of occasions, and have even made me look a real star more
than once.
Guide to SQL Server - Aloke Nath
ISBN: 0-201-62631-4 Published by Addison-Wesley. 567 pages.
This book is solely about MS SQL Server, covering 4.2 for OS/2 and SQL
Server NT. It is not bad, but does seem to regurgitate a lot from the
Sybase [sic] manuals. Its coverage is fairly broad dealing with
Transact SQL on the one hand through to client configuration on the
other. It does cover the aspects of MS Sqlserver that are different
from Sybase, (dbcc perfmon for instance) but it does not flag any as
such. Probably a good buy if you only have MS Sqlserver and never
intend looking at Sybase.
Client/Server Development with Sybase - Alex Berson and George Anderson,
ISBN: 0-07-005203-4 Published by McGraw-Hill. 743 pages.
I have used this book as a reference when system manuals where not
available. It is much more useful on how thing work and what approach
to use rather than syntax.
The breadth of topics pleased me - all the right jargon is mentioned.
The introduction mentions CORBA and DCE. Sybase RPC is compared to
UNIX RPCs. Middle ware products are discussed. Talks with our sales
rep. about the OMNI and NetGateway product where greatly assisted by
using the diagrams in the Open Server and Gateways chapter.
Like any text, it is dated (as it is printed). The Netgateway diagram
does not show a TCP/IP interface to MVS. However, the information
provided is not really diminished. This goes back to the fact that
this is a How Things Work and How to Use Things book, not a
compilation of details on a single version.
Physical Database Design for Sybase SQL - Rob Gillette, Dean Meunch, Jean
Tabaka
ISBN: 0-13-161523-8 Published by Prentice-Hall. 225 pages.
Supposedly the first in a series from Sybase Professional Services,
espousing the Sybase Development Framework or SDF (tm). I've seen no
more books, and have never heard any more about SDF. This book is a
reasonable attempt to guide developers through the process of turning
a logical database design into a physical Sybase implementation.
Topics include:
* Defining Tables and Columns
* Defining Keys
* Identifying Critical Transactions
* Adding Redundant Columns
* Adding Derived Columns
* Collapsing Tables
* Splitting Tables
* Handling Supertypes and Subtypes
* Duplicating Parts of Tables
* Adding Tables for derived Data
* Handling Vector Data
* Generating Sequence Numbers
* Specifying Indexes
* Maintaining Row Uniqueness
* Handling Domain Restrictions
* Handling Referential Integrity
* Maintaining Derived and Redundant data
* Handling Complex Integrity Constraints
* Controlling Access to Data
* Managing Object Sizes
* Recommending Object Placement
* Required Inputs to Physical DB Design
* Naming Guidelines
Covers System 10. Lots of good practical hints and guidelines on
database design. In the absence of any competition - a definite
recommendation for newcomers to Sybase database design.
Sybase Performance Tuning - Shaibal Roy & Marc B. Sugiyama
ISBN 0-13-442997-4 Published by Prentice Hall (www.prenhall.com). 622
pages.
Covers the topics:
* Tuning for performance
* Hardware and system software
* Sybase productand feature overview
* SQL Server - form and structure
* SQL Server - methods and features
* Physical database design
* Application development
* Monitoring SQL Server
* Instrumenting SQL Code
* Transaction processing performance
* Query processing performance
* Batch processing performance
* Advanced topics - I/O subsystems, named caches and buffer pools
and other enhancements
* Also a load of extra configuration details.
Just a quick note to let you know of a very good book on Performance
Tuning that isn't mentioned in the Sybase FAQ. I bought it a little
while ago and has quickly become invaluable. It's by two pretty gifted
Sybase Engineers in the SQL Server Performance Team and covers loads
of things up to and including System 11. It deserves to become as big
as the bible :)
This I believe is the Holy Grail of Sybase books that a lot of people
have been looking for - an exaggerated claim perhaps - but a damn fine
book.
_________________________________________________________________
--
Pablo Sanchez | Ph # (415) 933.3812 Fax # (415) 933.2821
pa...@sgi.com | Pg # (800) 930.5635 -or- pabl...@corp.sgi.com
===========================================================================
I am accountable for my actions. http://reality.sgi.com/pablo [ /Sybase_FAQ ]