Best Practices in One Place for DB2ers!

November 7, 2012 at 10:13 am  -  6 Comments

 

 

 

 

 

Any executive and seasoned manager knows recreating the wheel makes no sense, especially when great best practices are available to learn and leverage in your own business environment.  Best practices make your job easier and can help you achieve your business goals more efficiently. Great leaders listen to techniques that work with other customers and seek out valuable knowledge directly from the experts.  That’s why IBM has launched a new IBM Information Management Best Practices portal.

The portal is the new home to IBM Information Management best practices, and extends the existing sites from IBM developerWorks.

http://www.ibm.com/developerworks/data/bestpractices/db2luw/

The portal connects you to best practice communities for a variety of IBM Information Management products, including:

DB2 for Linux, UNIX, and Windows (including InfoSphere Warehouse, IBM Smart Analytics System, DB2 pureScale)
DB2 for z/OS
InfoSphere Master Data Management
InfoSphere Information Server
IMS
with more products joining us in upcoming months.

The best practice papers, technical presentations, and videos published in each community cover a wide range of topics to help you plan, design, implement, tune, and monitor your solutions so they perform at their best in your environments, potentially saving you time and resources.  The best practices are written and tested by IBM technical experts who work in the development and quality assurance teams that created the product features, and who work with customers like you to determine the best ways to efficiently use Information Management products in real world environments.

The best practices portal is also a great vehicle for you to share your experiences and opinions about our best practices, to help us improve them and publish new ones that answer your needs. You can join any or all of the best practice communities to be able to rank best practices and leave comments on any of them. You can also easily share best practices with your friends and colleagues on a variety of social media.

Come visit us and let us know what you like and what you need!

The DB2 Guys

I Love Seeing the Looks on their Faces…

October 3, 2012 at 11:35 am  -  219 Comments

 

 

 

Steve Astorino, Program Director, DB2 LUW – User Technology

Think about this. You have your entire IT team running around trying to figure out how to speed up your DBMS. You are working crazy hours to tune this parameter and that parameter so that you can get some minuscule performance improvement, which in the grand scheme of things, seems to make almost no difference to your customers’ experience of your product or services. You feel good about having spent so many weeks, months and so many resources to get this little improvement…I have seen this happen over and over again…This is when I say STOP!! You need to step back and look at the bigger picture. Just because you have been using the same DBMS for the last ten years, it doesn’t mean it’s the right one for you.

I know users who have seen multiple times performance gains on their system by moving to DB2. Don’t lower your expectations to meet the performance of your DBMS. Instead, raise your DBMS’ performance to meet your expectations and your customer’s expectations. DB2 does exactly that!! When showing off DB2 to customers, I love seeing the looks on their faces when they observe these results. It’s pure shock. I know what they are asking themselves…Why didn’t we switch to DB2 before? And by the way the cost of moving to DB2 is very small compared to the ongoing cost and lack of performance you currently experience.

DB2 10.1 has even better features which drive your performance through the roof. There are significant performance improvements:

Enhanced query performance for a range of common SQL statements
Intra-partition parallelism improvements better utilize multi-core processors
Improved performance of star schema based queries
Improved query performance through more efficient data prefetching
Improved query performance on tables with composite indexes
Statistical views improve the accuracy of cost estimation for queries with predicates containing expressions
Runstats now supports index sampling
SQL compiler registry variables can now be set for a specific SQL statement or application
Enhanced memory sharing on large POWER7 systems running AIX

I was working with a customer recently and here is what they shared: after we switched to DB2 we got 44% performance improvement for batch/reporting processes, 71% improvement for backup jobs and 25% lower query response time….

A different customer shares: one of the standard SAP programs go from 30+ hours down to 2 hours, a 93% increase in performance.

Forget the minuscule 1% or 2% improvement!! And also think about the cost savings that would come along with these performance gains…

See some of the related published material.

http://www.ibm.com/developerworks/data/library/techarticle/dm-0909db2whichedition/

http://www.ibm.com/developerworks/data/library/techarticle/dm-0909db2halicensing/

http://www.ibm.com/developerworks/data/library/techarticle/dm-0909db2compare/

Steven Astorino, BSc – Computer Science is a Program Director of DB2 Development overseeing Information Development, User Experience and DB2 Development. He has many years of experience in Databases including DB2 as well real time Database Replication. He began his career as a developer and has held a vast range of roles from software development and quality assurance to information development and user experience. Early in his career, Steven has spent several years working with network testing technologies for the Telecom Industry and played a key role in providing VoIP testing solutions. High quality, efficiency and customer focus are amongst his highest goals and directives to ensure outstanding customer satisfaction and experience.You can reach him at: astorino@ca.ibm.com.

DB2 Advantages Over Oracle in Compression

September 25, 2012 at 9:14 am  -  67 Comments

 

 

 

 

 

Chris Eaton, Technical Specialist, IBM

I have previously posted about compression and the new adaptive compression capabilities in DB2 10. In this posting I want to clearly articulate why DB2 is the leader in compression capabilities and how this translates into big savings for clients along with big performance gains for database workloads.

What is compression and what’s available in DB2?

No matter what form of compression we look at, they all work on the same basic premise which is to find repeating patterns in larger amounts of data and replace those repeating patterns with smaller symbols. That’s compression in a nutshell. The algorithm that delivers the best compression is the one that can find the most repeating patterns. However, for databases you have to weigh the cost of finding and replacing patterns against the benefits of smaller objects. With DB2 we leverage a number of different techniques to find the most patterns to compress out while at the same time actually improving performance.

How do we use more CPU to compress/decompress yet get better performance?

DB2 has the most compression algorithms to offer of any DBMS vendor with 2 algorithms for table compression, 3 for index compression, 2 for temporary table compression and 1 for log archive and backup compression. The beauty of the DB2 10 solution is that you as a user don’t pick the algorithms; DB2 looks at the data, decides on the best way to compress the data and presto, smaller tables, indexes, temporary objects and log files. With compression on tables in the range of 7x the objects are significantly smaller than they are uncompressed so every I/O brings in seven times more rows into memory and every GB of memory has seven times more rows packed in there (because DB2 keeps rows compressed in memory and on disk). By doing up to 7 times fewer I/Os we actually are seeing large performance boosts at most clients and yes we use more CPU to do that but most clients don’t run their Linux, UNIX or Windows servers at 100% CPU utilization all the time. Quite on the contrary, most are sitting at around 40% utilization and so there is lots of CPU to accommodate the 5-7% higher compression/decompression requirements where as I/O bandwidth is at a premium. So compression is a win/win.

What sets DB2 apart?

As I mentioned we have a lot of algorithms to leverage to find the most patterns. In fact, compared to Oracle 11gR2 we have a huge advantage. The reason is that Oracle (and Microsoft as well as Sybase) all use what is called a page level dictionary looking for repeating patterns only on a 4k, 8k, 16k or 32k data page. Where as with DB2 we use a page level approach but also a global table level scan for repeating patterns. This 2 level compression approach in DB2 means we find more patterns and therefore get better compression. Let me end off with an analogy. How many people in the department you work in have a birth date of June 15?  I would bet that the vast majority of readers of this blog do not share my birth date (of June 15).   But how many people on the planet share the same birthday as me?  Millions I’m sure.  Why?  Because if I only look at a small amount of data (1 page worth) I’m not likely to find the same repeating patterns that I would find if I looked at a much larger set of data (a table worth or a planet worth).  Since DB2 does both we deliver superior compression and improved performance.

Tune in to this webcast this week for more on compression.  http://www.idug-db2.com/?commid=53961

Look for more of my blogs at http://it.toolbox.com/blogs/db2luw

 

Chris Eaton is a worldwide technical specialist for IBM’s Information Management products focused on Database Technology, Information Governance and Workload Optimization.Chris has been working with DB2 on the Linux, UNIX, and Windows platform for over 19 years. He is the author of several books in the data management space and is an international award winning speaker.

Welcome to The DB2 Guys!

September 24, 2012 at 9:29 pm  -  2 Comments

Welcome to The DB2 Guys, your source for expert opinion on everything DB2! These veteran ‘DB2ers’ share their unique but, yes, independent ideas on DB2 and the industry in general. While the guys are all IBMers, this blog is intended for customers, buyers and all practitioners who need to learn, understand, and compare database solutions. In a Big Data world we need to fully maximize the benefits of these solutions to more aggressively impact the bottom line.  The DB2 Guys bring you knowledge paired with objectivity. Join in the conversation by sending in your comments.