The Business Benefits of DB2 9

7/29/2019 The Business Benefits of DB2 9

1/20

The business benefits of DB2 9(formerly codenamed Viper)

Author: Philip HowardDate Published: July 2006

a white paper by


2/20


3/20

Te business benets o DB2 9

Bloor Research 2006 Page

Executive summary

e Viper release o 2 which is ocially version is the most im

portant release o this database or many years ndeed regards Viper as so

signicant that at one time it considered calling it 3 on the basis that thisrepresents the third generation o databases rom ollowing and 2

s may be imagined there are a large number o new eatures and capabilities as

well as incremental enhancements in this release is paper does not attempt to

discuss all o these and concentrates in particular on those that have signicant

business benet

at begs the question o what we mean by business benet ince a database is

inherently a part o a companys inrastructure identiying its business ben

ets can be complex because many o the advantages that a particular eature

may bring are indirect new eatures o a database make it easier to manage

say then the people responsible or administration need to spend less time doing

that which means that they can spend more time ullling other duties n other

words they become more productive o or example some other project may

get completed more quickly than would otherwise be the case However this act

in itsel may not be visible outside the department let alone the act that it

derived rom an easier to manage database

ther benets are o course more direct Perormance enhancements or exam

ple may delay a requirement to upgrade hardware which has a direct impact on

the bottom line n the other hand some eatures may open up new possibilities

or the business that were not previously easible you may or instance now be

able to develop applications that would not have been viable beore thereby opening up new business opportunities

nother point to make is that business benets may be competitive in the sense

that they enable capabilities that are not available rom other vendors; or they may

be comparative where is introducing enhancements that oer benets when

compared to previous versions o 2 but do not necessarily provide competitive

advantage relative to other database products

n the discussions that ollow we will identiy the types o benet that individual

eatures o 2 provide and whether or not they are competitive or merely

comparative n order to do this and bearing in mind that databases are intrinsi

cally technical products in the sections that ollow we will not merely be discuss

ing 2 in business terms but we will also explain how the various technologies

work and where they dier rom that o s competitors in order to under

stand how these benets are derived

e ormat o this paper is that each o the major new eatures o 2 has a sec

tion to itsel ollowed by a composite section that briefy discusses the remaining

enhancements that have been introduced with this release inally we summarise

our conclusions


4/20


Bloor Research 2006Page 2

XML

t is because o its XL support that was considering calling this release o

its database 3 is is because 2 is no longer just relational nstead it

oers a hybrid relational/XL environment eore we consider the implicationso this we need to explain exactly what has done and how it diers rom the

XL implementations o other database vendors

ere are various ways in which XL documents can be stored e simplest

method is to store a document as a single entity is is done by treating it as a

large object (L) However the problem with this approach is that you can only

access the XL in its entirety You cannot or example access a particular detail

within the document

n order to enable the ability to access detail within an XL document an alter

native approach is to parse the XL at is to split the document up into its

constituent parts and then store those parts as relational data within tables in a

conventional manner However this method has the drawback that this shred

ding approach is slow oreover i you want to inspect the whole document then

it has to be reconstituted which again is a slow process t is thus commonplace

or vendors to oer this method in conjunction with L support so that you

do not have to recombine tabular XL data i you want to view the original

document in its entirety

Perormance is not the only problem with shredding it is not generally a compli

ant approach or example it is sometimes possible that ater decomposing an

XL document storing it and then putting it back together you do not end up

with what you started with n particular it does not support compliance in so aras digital signatures are concerned

nother issue or relational environments when storing XL is that elds stor

ing XL data are not natively understood by the database is means amongst

other things that columns with XL in them are not recognised by the database

optimiser which means urther problems with perormance or this reason a

number o database vendors have introduced an XL datatype is has the ad

ditional advantage that once you have an appropriate datatype you can dene

indexes against that column

However some suppliers having introduced an XL datatype have gone on to

suggest that this thereore means that they have native support or XL storage

is is not the caseXL data is still either shredded or stored as a Lsup

port or a datatype means that the database can natively understand that type o

data but it does not mean that it is stored natively so perormance issues do not

go away

or all o these reasons some companies have introduced specialised XL databas

es evertheless while this overcomes perormance issues in storing and retrieving

XL data it creates new problems when you want to combine relational and XL

data n a queryonly environment these are not so much o a burden because


5/20



ederated query products such as s ebphere normation ntegrator can

access separate XL and QL databases within a single query without too much

degradation in perormance (though there will always be some slow down) How

ever in a transactional environment these problems are much more acute and are

not limited to perormance or example you may need to implement twophase

commit across these disparate databases in order to retain synchronicity development will be much more complex and management and administration will be

more time consuming or these reasons pure XL databases are only really suit

able or applications which are purely XLbased and do not require relational

data as well

is brie outline describes the position prior to the introduc

tion o 2 hat has done is to appreciate that the only

recourse is to have a database that can store both relational and

XL data natively which is calling pureXL each with

its own storage mechanisms (so that there is no perormance hit

when accessing XL data) as illustrated in igure

ote that provides an environment that has a single manage

ment and administrative ramework as well as access capabilities

that allow the comingling o relational and XL data at is

you can use QL to access relational data and XQuery to access

XL data but you can also mix these (and s version o QL has long since

been XLised anyway) so that you can access both QL and XL within a

single query

Development and app.performance re: XML data

with relational data serverwith DB2 Viperhybrid data server

Development ofsearch & retrieval businessprocesses

LOB: 8 hrsShred: 2 hrs

30 min.

Add field to schema 1 week 5 min.

Relative lines of I/O code(65% reduction)

100 35

Queries 2436 hrs 20 sec10 min

Query non-shreddedXML element

1 week day

n terms o direct benets there are two main advantages o this approach e rst

o these is that as discussed it eliminates the perormance hit that you get i XL

is not stored natively or example able shows the relative perormance o di

erent approaches to XL based on comparisons done at a beta customer o 2

Viper ote that perormance benets are not limited to the comparison between

hybrid storage on the one hand and the use o Ls or shredded approaches on

the other but also extends to database administration (adding a eld to a schema

which can be done on the fy thereby making the environment much more fex

ible) and to development is leads on to the second major benet oered thanks

Figure 1: Hybrid storage in DB2 9

Data Server Hybrid Data Server

Database

Physical

Storage

DatabaseLogical view ofstorage

Tables

Views

Physical StorageDatabaseles

Regular StorageData stored in a rowand column format

pureXML StorageData stored in a pre-parsedhierarchical format,not as asingletext object (CLOB)

Data ServerServices that manage, secure and

provide access to the database

Hybrid Data ServerDB2 supports both relational and pureXML

storage and provides all the necessary services

to support both data structures

Table 1: Relative perormance o diferent approaches to XML


6/20


Bloor Research 2006Page

to the hybrid storage method adopted by which is that it enables the devel

opment o applications that combine XL and relational data in ways that were

not previously realistic

course a lot o XL especially in an (service oriented architecture) envi

ronment is transient However much o it is not ne good example is anythingthat involves contracts ranging rom consumer contracts in the nancial services

sector to service level agreements within the sector e problem is that such

documents typically include both structured (relational) and unstructured (XL)

data and you not only want to be able to manipulate these separately (and extract

the structured data rom the contract in the rst place) but also together oreo

ver in some instances details change on a regular basis or example when it

comes to servicelevel agreements these will be reviewed and amended on a peri

odic basis imilarly you may have milestones built into a contract that you need

to monitor on an ongoing basis n such circumstances you only want to have to

change one part o the data (relational or XL) and you want those changes to

be refected across the contract thus you might change the price o a contract in

your relational data but you want that to be refected automatically in the relevant

XL document

o be more specic consider the number o standardised XL variants

inancial (insurance) XL (nancial inormation) PL

(nancial products) L (unds) XL (business reporting);

Lie ciencesGV (genomics) L (bioinormatics) L (chemi

cals);

PublishingportL ewsL X (book industry) XPL (publicrelations);

thersLandL (land development) L (textile/clothing sup

ply chain) atL (materials property) JX (global justice) ebXL

(electronic business)

nd so on and so on Huge amounts o inormation are captured using XL but

it is dicult and unwieldy to extract structured inormation rom those docu

ments and to use that as or combine it with relational data unless you use a

solution such as 2 which at present is one o a kind

nother example is the use o XL or business intelligence purposes n the

past you could not really use XL documents or conventional because it was

impractical in terms o the time taken to shred relevant documents unless the da

taset under consideration was very small ithout that need it makes it practical

to query large XL document stores in a relatively short period o time whereas

previously this was only easible or processes such as text mining

urther and in particular sees its hybrid support or XL as a cornerstone

in the move towards an environment s an example see igures 2 and 3


7/20



which show pre and postversions o the same application environment e

potential or using an XLbased approach should be obvious

inally it is worth remarking on s support or XQuery which is the standard

method or addressing XL data and in particular its introduction o Visual

XQuery uilder in this release as illustrated in igure 4 hile the ormer is not

out o the ordinary the Visual XQuery uilder should certainly be a boon or

developers t is also worth noting that has announced partnerships with a

number o vendors including Zend echnologies xegenix Justsystem and c

tiveGrid that have extended toolsets to help in the development o XLbased

applications

Figure 2: Original order ullment system

Figure 3: DB2 9 powering the new SOA basedsolution via XML

Figure 4: Visual XQuery Builder


8/20



Compression

ompression is the science o tting a quart into a pint pot r better yet a hal

pint pot n principle compression is very valuable i you have a 0 terabyte da

tabase and you can compress it by 50% then you can store it in 5 terabytes insteadthereby reducing both your storage costs and associated running costs However

compression is dicult or conventional relational databases is is because they

are rowbased and because you store data by row you have to compress it by row

(ote this is not the case or columnbased relational databases but as there are

none o these that support transactional processing and only a ew in the data

warehousing arena we will assume a rowbased approach or the purposes o this

discussion)

n a typical database row you might have one column with alphabetic data in it

another with alphanumeric inormation several with numeric data some with

foating point decimals one or two date elds and a variety o other exotica e

problem is as ar as compression is concerned that the best way to compress

an alphanumeric eld is dierent rom the best method o compressing foating

point decimals hat this means in practice is that i you simply compress rows o

data you have to select an algorithm that is based on the least common denomina

tor it compresses across the row bearing in mind that there are multiple datatypes

within a row so no one column is compressed optimally

Prior to this release had already introduced null and deault value compres

sion multidimensional clustering (which provides index compression through

the use o block indexes) and database backup compression though only the

rst and third o these could be described as generic with the multidimensional

clustering being specic to data warehousing t has also had value compressionwhich is eectively the lowest common denominator approach described above

or some time However in this release the company has added what it reers

to as row compression but which is actually tokenisation

which is to our minds not the same thing as compres

sion though it amounts to the same thing in the end in

that it means you need less storage space

Put briefy the aim o tokenisation is to separate data val

ues rom data use is has the eect o reducing data

requirements and improving perormance s a practi

cal example in a customer table you might have many

customers in ichigan (or in the case o igure 5 em

ployees in Plano exas) and each o them would have

ichigan stored as a part o their address o store this

several hundred or even thousands o times is wasteul

okenisation aims to minimise this redundancy

ere is more than one way o implementing tokenisa

tion but the method chosen by is that the database

sotware will look or repeated patterns within the data

and when these are ound it will create a relevant tokenFigure 5: Elements o tokenisation


9/20



(usually a numerical value) that represents ichigan (or Plano exas as illus

trated) each time that it appears within a table and then have what is eectively a

lookup table (held in the data dictionary) so that you can convert rom the token

to the data value ote that or numeric values tokens are not required

n terms o storage savings estimates that use o this technique will producesavings o anywhere between 35% and 0% depending on how much repeated

data is stored ertainly in customer and H applications or example there are

likely to be considerable savings urther we know o no other relational database

vendor (excepting the aorementioned columnbased products) that can oer this

level o compression hile we tend to treat benchmarks with caution igure

6 illustrates the dierent compression rates achieved by 2 when compared

to more conventional approaches (which would also typiy previous versions o

2)

ere is however a potential downside when you access tokenised data you have

to read rom the dictionary as well as the relational table so there is extra / in

volved in the process However this is oset by the act that data is still in token

ised orm when held in buer pools in memory is means that more data can

be held in these buers thus reducing the amount o / required n practice

estimates that the advantage gained rom holding more data in memory will

usually outweigh the act that you have to detokenise the data or example take

this customer quote rom the senior database administrator at utoZone with

the new compression technology in DB2 Viper, we realized an 80 Percent improvement

in space savings or our most critical tables in our Data Warehouse. We were even more

pleased with this technology when we ound that Vipers compression capability helped

us process queries to the database an average o 0 Percent aster than beore. Were

looking orward to seeing the same results with our Operational Data Store and OLP

systems n other words more oten than not you should get a perormance benet as well as a reduced storage requirement

o conclude this section even i we assume the lowest possible actor or reducing

storage requirements database size should be reduced by a third while at worst

perormance should stay approximately the same us there should be signicant

direct benets in terms o cost o ownership ne point to note however is that

compression is not available or XL (it is planned) Ls or indexes (except

when used with multidimensional clustering as discussed) so the benets out

lined here apply specically to relational data

Figure 6: Compression rates o DB2 9compared to other approaches


10/20



SAP optimisation

Given racles status (subsequent to its acquisition o Peopleot iebel et al)

as Ps main competitor it is perhaps not surprising that P and have

strengthened their partnership with respect to P and 2 is started withthe release in 2005 o 2 version 22 which was specically optimised or

P environments ith the Viper release o 2 this integration has gone ur

ther and the two companies (all development is perormed jointly) have an ongo

ing roadmap or continuing this partnership into the uture

e key to P optimisation is not so much that there are additional eatures or

P environments but that 2 understands the P environment within which

it is working or example one o the new eatures o 2 is or the sotware to

autodiscover its environment upon installation and to automatically set deault

values based on that conguration ormally speaking this would include the

hardware platorm operating system and other inrastructure details but in the

case o P 2 can now also recognise relevant details o the P conguration

in use and set these deaults accordingly as well

course you could argue that you will install 2 beore you install P so

that the onguration dvisor (which provides the capabilities just discussed)

wont know about the P deployment until later However there is also a silent

install capability which means that you can install 2 as a part o the P

installation processin eect implementing both together in which case the

onguration dvisor will indeed know about the P environment ote that

is the only database vendor to have this silent install capability

e other eatures that 2 oers in conjunction with P are also based aroundthe act that the ormer knows about the latter or example 2 is aware o

P workloads and the databases builtin tuning capabilities can use this when

it makes recommendations about building new indexes materialised query tables

and so on; and the same applies to troubleshooting whereby diagnostics also

understand the P environment eatures like multidimensional clustering can

also be used specically in conjunction with P

e benets deriving rom the partnership between P and are essentially

about easier management and administration and thereore reduced overhead in

these areas However this is one o those cases when there are indirect advantages

that accrue rom these benets oth P and 2 (and relational databases in

general) are complex environments Getting the best perormance out o them on

a consistent basis is not a trivial task the integration o the two products means

that the task o tuning or example is very much easier and as a result improved

perormance can also be expected t is also worth noting that although this is not

technology specic and P have aligned their maintenance and product

delivery plans n the case o the ormer this means a single port o call in the event

o a problem arising

inally it is worth bearing in mind that has never been the leading data

base vendor underpinning P solutions However there are signs that that could


11/20


Bloor Research 2006 Page 9

change ince 2 version 22 came out with P integration has won a

number o competitive P deals some o which have migrated away rom other

database products


12/20



Range partitioning

nlike any o the other eatures we have so ar discussed range partitioning in

itsel (as we shall see things are dierent when combined with multidimensional

clustering) does not oer any competitive advantage though it does oer comparative advantage with respect to earlier versions o 2 e should however

point out that while Viper applies to both the distributed and mainrame versions

o 2 in act range partitioning has been available in 2 or z/ or some

time

evertheless the introduction o range partitioning is late ome other vendors

have been able to oer it or a decade hat it does is allow you to partition data

by range such as a date range or geography or store us you could have separate

data partitions or January ebruary arch and so on; or or rance Germany

razil and so orth ome vendors allow you to have multiple ranges in the sense

that you can have subranges us you could have a partition by month sub

partitioned by geography say n theory at least there are products that allow you

to have an innite number o such subpartitions however does not need

to do this instead it uses range partitioning in conjunction with its existing hash

capabilities and the existing multidimensional clustering that we have already

mentioned e way that these work together is shown in igure 7

ow reers to this as distribution partitioning and

organisation but in practice these are all dierent methods

o partitioning use hash partitions to distribute the data

in a way that the data does not become unbalanced sub

partition using range techniques and then partition by

dimension ote that you can have any number o dimensions when using multidimensional clustering

ow in a great many instances a dimension could be a

range or example in igure 7 the data has been organ

ised by geography o the question is why opted or

multidimensional clusters rather than support or sub

ranges? oreover why did it introduce the ormer beore

the latter? e answer is that multidimensional clustering

is not just about how you store the data (close together so you get better perorm

ance) but is also intrinsically linked to the way that you want to do slice and dice

and similar things within a query environment e truth is that you could build

an ntier range partitioning model (where the database supports itmany do not)

but it would be complex and it would mirror the work that you have to do when

setting up relational cubes

ere is also the opposite question i multidimensional clustering is so good

why do you need range partitioning? ere are actually two answers to this ques

tion the rst is that you may want to segment data in a way that is not material

to your needs or slicing and dicing and the second is that you may want to use

range partitioning without multidimensional clustering relevant example that

applies to both these cases is when partitioning is used in conjunction with Label

Jan Feb Jan Feb Jan Feb

No rt h So ut h

E as t W es t

No rth So ut h

E as t W es t

No rt h So ut h

E as t W es t

No rt h So ut h

E as t W es t

No rth So ut h

E as t W es t

No rt h So ut h

E as t W es t

Node 1 Node 2 Node 3

TS1 TS2 TS1 TS2 TS1 TS2

T1 distributed across 3 database partitions

Distribute

Partition

Organise

DISTRIBUTE BY HASH

PARTITION BY RANGE

ORGANISE BY DIMENSIONS

Figure 7: DB2 data partitioning


13/20



based ccess ontrol (see next section) is allows you to partition by security

level so that you might have top secret inormation in one partition say and

condential inormation in another

n associated eature that 2 provides is roll in/roll out support or parti

tions pecically you can attach or detach partitions using the L Lstatement is ability can be particularly useul in a data warehouse environment

where you oten need to load or delete data to run decisionsupport queries

e bottom line is that in competitive terms range partitioning in conjunc

tion with the other eatures discussed may provide a perormance advantage

(when compared to products that do not support unlimited range partitioning)

and will oer an administrative advantage ne eature that we would like to

see introduce is support or replicated partitions which can enhance query

perormance in large parallel environments hile we are not aware o any o its

mainstream competitors having this acility some o the data warehouse appliance

vendors do


14/20



Security

ere are two major new security eatures in 2 the introduction o Label

based ccess ontrol (L) and support or trusted contexts

Labelbased ccess ontrol is an implementation o andatory ccess ontrol

where the latter is a security system based on the principle that an administrator

determines user access rights and that users may not assign less stringent access

rights to any data that they have control over is is in contrast to iscretionary

ccess ontrol () where at least in principle the owners o the data (users)

determine who may or may not have access to it hat has done is to imple

ment L as a complementary unction to the that has been the historic

basis or security within 2

e point about the complementary nature o access security with L is that

the historical approach taken by with has been to apply this at the

table level n other words a user could either look at data in a table or he or she

could not L on the other hand is implemented at the row and column level

either individually or in conjunction us is relatively coarsegrained and

L is much more granular though you can use L at the table level as well

i you wish to replace ote that you do not have to apply L to all tables

within a database

n terms o the way that L works it is not dissimilar rom conventional access

control data is assigned a label and so is each user (using a hierarchy group or

treebased approach as required) and the two are compared at run time How

ever it is not quite as simple as this imply comparing labels provides a very

monolithic security structure whereas you typically need something that is morefexible o L also includes the concept o security policies which are pre

dened rules delivered with 2 which you can apply whenever data is read

or written these rules are not enough it is also possible to assign exemptions

whereby particular individuals have special permission to access inormation that

they would not normally be allowed to see

e benets o L are twoold rst it provides more granular control over

who can see and do what econdly and perhaps more importantly the combi

nation o L and is much more fexible; in principle it allows you to

implement the security policies that best suit your organisation rather than being

orced down a particular security path by the constraints o the database is is

both a comparative and competitive advantage

e second aspect o security that is new in 2 is that o trusted contexts ese

are essentially a way o bridging between disparate systems and applications that

have dierent security models rusted contexts are dened on the server and reer

to the connections that exist between applications databases or whatever on

nections may qualiy as a trusted context based on one or more relevant attributes

(userid P address and so on) and a context may also coner membership in a ole

though it cannot (in this release) assign a label e big advantage o trusted con

texts is that it avoids authentication costs within (especially) 3tier systems


15/20



Tuning

ere are a number o dierent elements that relate to tuning that have been

extended in this release notably adaptive seltuning memory the design advisor

and automatic storage though it is arguable that the last o these is more o anadministrative unction than a tuning one

hat calls adaptive seltuning memory is the ability o the database to

detect the workload on the database (including P details i that solution is

running) and to tune the memory available based on the needs o that workload

redistributing memory between processes as required to optimise the workload

n eect automatic storage support does the same thing or storage except that

it is not dynamically based on workload but is instead based on dened policies

(rules) that you set up or dierent storage types t requires the use o the ata

anaged torage () model but what it allows you to do is to have aster

and slower disks (or other storage media) on the same system and then allows

you to allocate data that is say more than three months old to slower disks while

keeping more uptodate data on the astest disks e system can also allocate

and grow storage on demand which is a eature that specically supports P

(so policies might be based on inormation rom within the P rather than the

2 environment) n other words it is providing at least a part o the solution

or inormation liecycle management (L) though s oering here is not

yet complete

e esign dvisor itsel has not been especially enhanced in this release n other

words it can still recommend the creation o indexes materialised query tables and

multidimensional clustering dimensions and can then automatically create thesei required However the previously available acilities or recommending parti

tions has been extended to include the newly introduced range partitions and as

with the seltuning memory previously discussed it now has knowledge o the

workload on the system (including P)

ll o these eatures lead directly to administrative improvements and indirectly

to perormance benets in the case o seltuning memory and the esign dvisor;

and cost reductions in the case o automatic storage


16/20


Bloor Research 2006Page

Other features

s one would expect rom a major release o a major product there are a large

number o new and enhanced eatures and capabilities in 2 Viper apart rom

those that we have discussed ese range rom new online acilities such as dynamic buerpool operations; online index creation and maintenance and online

loading; to new automated eatures including backups and statistics collection

which will reduce administrative workload; and the removal o size limitations to

ablespaces

ere are also a number o new eatures to support developers not least o which

are the XL acilities already discussed n addition to these there is also a new

clipsebased developer workbench in place o the previous 2 evelopment

enter there is a new stored procedure debugger and there are a number o other

enhancements or developers n particular there is now support or online table

reorganisations Previously using the L L command (used or exam

ple to drop or change a column) meant that you had to drop the table completely

and recreate it which was not just time consuming and complex but you also had

to quiesce the database is eature will be a boon to developers though it is not

beore time (competitive products have been able to do this or some years)


17/20



Conclusion

ere are many new eatures in 2 (Viper) o which we have highlighted

some o the most important any o the benets deriving rom these new capa

bilities are conventional reduced administration and management overhead leadsto reduced cost and/or increased productivity; improved perormance leads to

better utilisation o existing hardware which in turn means that upgrades and re

placements can be put on hold with a direct impact on the bottom line; enhanced

security means that management can sleep easier in their beds at night and helps

to support compliance requirements; in the case o P understanding that en

vironment leads to perormance and administrative gains that lead to the benets

just outlined; and so on However there are two areas that we would particularly

like to highlight

e rst o these derives rom the row compression introduced in this release n

like perormance which has an indirect benet in terms o hardware requirement

improved compression has a direct impact on your hardware needs and thereore

related costs

e nal benet that we believe derives rom 2 is arguably the most impor

tant but it is also the most intangible it is the ability provided by the new hybrid

XL/relational storage (pureXL as calls it) to create applications com

bining these data types that were not previously realistic possibilities e do not

know how many o these applications there are nor what they might look like be

yond the discussions in this paper evertheless we believe that they are there and

that there are many o them us the greatest benet o 2 is likely not to

be perormance or reduced cost or any o those things but the new possibilities it

opens up to companies that want to take advantage o the new acilities on oer

o conclude this release o 2 introduces new capabilities that signicantly

exceed those o its competitors in a number o areas n particular its pureXL is

a major advantage and in any company where XL is an important consideration

(and this increasingly applies to all organisations) should be at the head o

the list when considering potential suppliers


18/20

Bloor Research Overview

Bloor Research has spent the last decade developing what is recognised as Europes

leading independent IT research organisation. With its core research activities

underpinning a range o services, rom research and consulting to events and pub-lishing, Bloor Research is committed to turning knowledge into client value across

all o its products and engagements. Our objectives are:

Save clients time by providing comparison and analysis that is clear and

succinct.

Update clients expertise, enabling them to have a clear understanding o IT

issues and acts and validate existing technology strategies.

Bring an independent perspective, minimising the inherent risks o product

selection and decision-making.

Communicate our visionary perspective o the uture o IT.

Founded in 1989, Bloor Research is one o the worlds leading IT research, analy-

sis and consultancy organisationsdistributing research and analysis to IT user

and vendor organisations throughout the world via online subscriptions, tailored

research services and consultancy projects.


19/20

Copyright & Disclaimer

is document is subject to copyright. No part o this publication may be repro-

duced by any method whatsoever without the prior consent o Bloor Research.

Due to the nature o this material, numerous hardware and sotware products

have been mentioned by name. In the majority, i not all, o the cases, these

product names are claimed as trademarks by the companies that manuacture the

products. It is not Bloor Researchs intent to claim these names or trademarks as

our own. All diagrams in this document have been provided by IBM.

Whilst every care has been taken in the preparation o this document to ensure

that the inormation is correct, the publishers cannot accept responsibility or any

errors or omissions.


20/20

ti i IT i t t

uite 4, Town Hall, 86 Watling Street EastOWCESTER, Northamptonshire, NN12 6BS, United Kingdom

el: +44 (0)870 345 9911 Fax: +44 (0)870 345 9922

The Business Benefits of DB2 9

Documents

Transcript of The Business Benefits of DB2 9