Posted by Mr. Bruce McKnight on 23 August 2012 at 12:54
I know it's just a draft but the grammar makes it exceptionally difficult to read.
Posted by Mr. Paolo Bruni on 26 September 2012 at 17:38
That's why we appreciate the hard work of our editors before publishing, I think it flows much better now.
Posted by Hari Shanmugadhasan on 5 January 2013 at 14:21
Page 213 makes this very peculiar statement, without qualification:
"DB2 inserts are expensive and take time."
Now, this is true if you are comparing to IMS Fast Path VSO DEDB, or the Transaction Processing Facility (TPF) which powers the Sabre and Amadeus airline reservation systems, but they are not the audience for this Redbook.
It would make more sense to use some of the DB2 10 performance data to indicate peak insert rates. Or the insert part of the SAP results mentioned on page 6.
Posted by Hari Shanmugadhasan on 5 January 2013 at 14:35
Page 213 strangely discusses MEMBER CLUSTER under "Append Processing" and oddly states that:
"This option allows for inserts to occur quickly because the inserts occur where space is available."
It would have been more useful to state that MEMBER CLUSTER's key use is to reduce P-lock contention between high volume concurrent inserts by multiple DB2 data sharing members to the same part of a single partition, especially with row locking (a SAP type scenario).
Posted by Hari Shanmugadhasan on 5 January 2013 at 14:43
Page 214 oddly states:
"The APPEND option on the table inserts or loads data into the table without clustering."
There is an important distinction between SQL Inserts and the regular LOAD or Crossloader utilities where having sorted input is important.
LOAD and the Crossloader, INCURSOR flavor of LOAD, are much better choices for conversions or migrations of data to DB2 for z/OS than DB2 SQL Inserts.
Posted by Hari Shanmugadhasan on 5 January 2013 at 14:50
Page 213 strangely brings up the expense of time of DB2 Inserts in the context of "Append Processing" instead of high volume concurrent Insert to the same part of the same partition.
It does not mention the potential relief provided by MRI (Multiple Row Insert) and Online LOAD (LOAD SHRLEVEL CHANGE) which reduce the CPU costs of inserts. Or how various PBR partitioning schemes can provide relief in addition to other benefits. Or the DB2 10 RANDOM index option.
Posted by Hari Shanmugadhasan on 5 January 2013 at 14:52
It also doesn't mention the potential use of the CURRENT MEMBER special register in DB2 data sharing environments to reduce potential contention between DB2 data sharing members.
Posted by Hari Shanmugadhasan on 6 January 2013 at 12:12
Peculiarly, page 213 gives the impression that the Redbook is stuck in pre-DB2 10 thinking, as it states:
"With MEMBER CLUSTER, each space map covers 199 data pages."
Actually, each space map covers 10 segments in a DB2 10 UTS with MEMBER CLUSTER.
Posted by Hari Shanmugadhasan on 6 January 2013 at 12:20
The page 213 discussion of Inserts neglects the opportunity to discuss the potential uses of DB2 10 Hash Access for some tables, especially when converting from applications that use similar mechanisms like IMS HDAM.
Posted by Hari Shanmugadhasan on 6 January 2013 at 12:25
With a title starting with "Streamline Business . . ." it is disappointing that the opportunity to use a native pureXML implementation appropriately doesn't get more focus.
DB2 10's ability to support volume XML activity along with XPath and XQuery access could be an important way of streamlining business and providing extra value to users.
Posted by Hari Shanmugadhasan on 7 January 2013 at 13:23
Unfortunately, the text does not show that DB2 10 for z/OS makes it easier to deliver secure, world-wide, high speed, highly reliable, highly mobile business functions to customers, supplier, business partners and employees, through effective implementation of technologies like XML, Unicode and Message Queuing.
It would have been better to provide a clear DB2 path to robust e-commerce mobile apps via industry standard data interchange (XML) to support B2C and B2B applications on smart phones.
Posted by Hari Shanmugadhasan on 7 January 2013 at 13:39
For example, show how DB2 10's pureXML implementation, including XPath and XQuery support would make it easier to develop robust end-to-end applications based on industry XML standards.
For example, going from an industry standard XML Purchase Order through all the stages to industry standard XML Waybill, Lading, Shipping, Invoicing and Customs documents for each truck/train/ship/plane shipment.
"Streamline Business . . ." should be illustrated with DB2 10.