Conor O'Mahony's Database Diary

NYSE Euronext uses Netezza to Manage their “Big Data”

Conor O'Mahony — Fri, 24 Feb 2012 16:02:26 +0000

NYSE Euronext operates multiple securities exchanges, including the New York Stock Exchange and Euronext. As you might imagine, securities exchanges present significant data management challenges. But NYSE Euronext didn’t just want to have a transactional system, they wanted to do much more with their data, further increasing the challenges. At the 2011 IBM Information On Demand (IOD) conference, NYSE Euronext described their challenges and the solution they chose. In particular, they highlight Netezza’s tremendous performance and how fast it is to get up-and-running with Netezza.

Not only is it easy to get up-and-running with Netezza, but it is easy to manage your environment on an ongoing basis. You can hear for yourself in this short video segment…

Coca Cola Bottling Move from Oracle Database to IBM DB2

Conor O'Mahony — Thu, 23 Feb 2012 15:07:38 +0000

At the 2011 IBM Information On Demand (IOD) Conference, Coca Cola Bottling spoke about their experiences when moving from Oracle Database to IBM DB2. I have included some very brief video segments shot at the conference below. It is really interesting to heard about the experiences and impact of switching from Oracle to IBM from the people involved.

In the following short video segment, hear how Coca Cola Bottling have changed their fix pack philosophy as a result of moving. With Oracle Database, they would avoid fix packs unless they “had to”. But with DB2, applying fix packs is much easier and faster, providing faster access to new functionality, performance improvements, and bug fixes. Also, hear about how Coca Cola Bottling have had significant data storage savings thanks to moving to DB2. Who wouldn’t want to reclaim some of that IT budget allocated for storage purchases

And finally, hear about their experiences with performance boosts and the autonomic computing capabilities in DB2.

Get a Free Copy of the Forrester Wave™ for Enterprise Hadoop Solutions

Conor O'Mahony — Thu, 02 Feb 2012 18:41:04 +0000

Today, Forrester published its Wave analysis for enterprise Hadoop solutions. It has detailed coverage of the Hadoop solutions from vendors like IBM, MapR, Cloudera, Hortonworks, and others. If you are considering an enterprise Hadoop solution, such as IBM InfoSphere BigInsights, it will make for very interesting reading. You can download a free copy of the report from The Forrester Wave™: Enterprise Hadoop Solutions, Q1 2012.

Oracle Reduce their Exadata Projections

Conor O'Mahony — Mon, 30 Jan 2012 17:17:13 +0000

In June of last year, during Oracle’s FYQ4 2011 earnings call, Larry Ellison claimed that Oracle expect more than 2,000 Exadata systems to be installed in fiscal year 2012. His exact quote follows. You can read the full transcript on SeekingAlpha at Oracle’s CEO Discusses Q4 2011 Results.

Today, more than 1,000 Exadatas are installed, and we plan on tripling that number this year.

He noted that more than 1,000 systems had been installed at that time. Tripling this number yields more than 3,000. This implies that there would be more than two thousand new systems installed in FY2012.

Last month, during Oracle’s FYQ2 2012 earnings call, Larry Ellison said:

This past Q2, Oracle sold over 200 Exadata and Exalogic engineered systems. In Q3, we plan to sell over 300 Exadata and Exalogic engineered systems. In Q4, we plan to sell over 400 Exadata and Exalogic engineered systems.

Again, the full transcript is available on SeekingAlpha at Oracle’s CEO Discusses Q2 2012 Results. There is no reference to Q1 sales, but Oracle projects that Q2 + Q3 + Q4 sales of both Exadata and Exalogic will be more than 900.

A couple of things stand out here. The first is that these latest projections from Oracle are for both Exadata and Exalogic systems combined, whereas the original projection was for Exadata systems only. The second is that these latest projections from Oracle are significantly down (more than 2,000 has been revised down to whatever business they did in Q1 + more than 900 in Q2, Q3, and Q4 combined). And this significant downward revision in projections has happened in the space of just 6 months.

If you read the Q&A segment from the Q2 earnings call, it is quite interesting. An analyst asks Oracle about the downward revision in projections. There are some semi-coherent responses from Ellison and Hurd, before Hurd claims that instead of 3x growth in engineered systems, they are on track for 2.5x growth. Hmmm, unless they had a monster Q1, that doesn’t quite add up either

Win a Trip to the IDUG Conference of your Choice

Conor O'Mahony — Wed, 25 Jan 2012 18:01:38 +0000

The International DB2 User Group (IDUG) is a user-run organization. If you want independent information about DB2, IDUG is the place to go. This year, IDUG are have conferences in the US (Denver), Germany (Berlin), and Australia (Sydney). The good news is that the DB2night Show is holding a contest, and the prize is an all expenses-paid trip to the IDUG conference of your choice. The contest aims to identify new users who can speak about their experiences with DB2. It’s a talent contest of sorts, where the talent is sharing your experiences. If you have ever considered speaking at a conference, this contest is the ideal way to see how you might do in a fun setting.

Anatomy of an Oracle Marketing Claim

Conor O'Mahony — Wed, 18 Jan 2012 15:01:50 +0000

Yesterday, Oracle announced a new TPC-C benchmark result. They claim:

In this benchmark, the Sun Fire X4800 M2 server equipped with eight Intel® Xeon® E7-8870 processors and 4TB of Samsung’s Green DDR3 memory, is nearly 3x faster than the best published eight-processor result posted by an IBM p570 server equipped with eight Power 6 processors and running DB2. Moreover, Oracle Database 11g running on the Sun Fire X4800 M2 server is nearly 60 percent faster than the best DB2 result running on IBM’s x86 server.

Let’s have a closer look at this claim, starting with the first part: “nearly 3x faster than the best published eight-processor result posted by an IBM p570 server“. Interestingly, Oracle do not lead by comparing their new leading x86 result with IBM’s leading x86 result. Instead they choose to compare their new result to an IBM result from 2007, exploiting the fact that even though this IBM result was on a different platform, it uses the same number of processors. Of course, we all know that the advances in hardware, storage, networking, and software technology over half a decade are simply too great to form any basis for reasonable comparison. Thankfully, most people will see straight through this shallow attempt by Oracle to make themselves look better than they are. I cannot imagine any reasonable person claiming that Oracle’s x86 solutions offer 3x the performance of IBM’s Power Systems solutions, when comparing today’s technology. I’m sure most people will agree that this first comparison is simply meaningless.

Okay, now let’s look at the second claim: “nearly 60 percent faster than the best DB2 result running on IBM’s x86 server“. Oracle now compare their new leading x86 result with IBM’s leading x86 result. However, if you look at the benchmark details, you will see that IBM’s result uses half the number of CPU processors, CPU cores, and CPU threads. If you look at performance per core, the Oracle result achieves 60,046 tpmC per CPU core, while the IBM result achieves 75,367 tpmC per core. While Oracle claims to be 60% faster, if you take into account relevant system size and determine the performance per core, IBM is actually 25% faster than Oracle.

Finally, let’s not forget the price/performance metric from these benchmark results. This new Oracle result achieved US$.98/tpmC, whereas the leading IBM x86 result achieved US$.59/tpmC. That’s correct, when you determine the cost of processing each transaction for these two benchmark results IBM is 39% less expensive than Oracle. (BTW, I haven’t had a chance yet to determine if Oracle Used their Usual TPC Price/Performance Tactics for this benchmark result, as the result details are not yet available to me; but if they have, the IBM system will prove to be even less expensive again than the Oracle system.)

Benchmark results are as of January 17, 2012: Source: Transaction Processing Performance Council (TPC), www.tpc.org.
Oracle result: Oracle Sun Fire X4800 M2 server (8 chips/80 cores/160 threads) – 4,803,718 tpmC, US$.98/tpmC, available 06/26/12.
IBM results: IBM System p 570 server (8 chips/16 cores/32 threads) -1,616,162 tpmC, US$3.54 /tpmC, available 11/21/2007. IBM System x3850 X5 (4 chips/40 cores/80 threads) – 3,014,684 tpmC, US$.59/tpmC, available 09/22/11.

Deploying DB2 and InfoSphere Warehouse on Private Clouds

Conor O'Mahony — Mon, 12 Dec 2011 21:40:12 +0000

Cloud computing is certainly a hot topic these days. If an organization is not already using cloud computing, it has plans to do so. The economics, agility, and value offered by cloud computing is just too persuasive for IT organizations ignore.

Even the high-profile Amazon outage couldn’t slow cloud computing’s relentless march towards mainstream adoption. If anything, that outage helped make cloud computing more robust by highlighting the need for hardened policies and procedures around provisioning in the cloud.

IBM recently announced updates to a set of products that make it easy to deploy DB2 and InfoSphere Warehouse on private clouds:

IBM Workload Deployer (previously know as WebSphere CloudBurst), which is a hardware/software appliance that streamlines the deployment and management of software on private clouds.
IBM Transactional Database Pattern, which works with the IBM Workload Deployer to generate DB2 instances that are suitable for transactional workloads.
IBM Data Mart Pattern, which generates InfoSphere Warehouse instances for data mart workloads.

These patterns consist of more than just deploying virtual images with pre-configured software. You should instead think of them as being like mini-applications for configuring and deploying a cloud-based database instances. Users specify information about the database, and then the pattern builds and deploys the database instance.

The Transactional Database Pattern is for OLTP deployments. It includes templates for sizing the virtual machine, database backup scheduling, database deployment cloning capabilities, and tooling (including Data Studio). The Data Mart Pattern incorporates the features to the OLTP pattern, together with deep compression and data movement tools. But, of course, it is configured and optimized for data mart workloads in a virtual environment.

Need Help Determining Hadoop Split Sizes? Use Adaptive MapReduce Instead!

Conor O'Mahony — Wed, 07 Dec 2011 17:07:10 +0000

IBM is actively working on adaptive features for the Map and Reduce phases of its InfoSphere BigInsights product (which is based on Apache Hadoop). In some cases, this involves applying techniques commonly found in mature data management products, and in some cases it involves developing new techniques. While a number of these adaptive features are still under development, there are some features in the product today. For instance, BigInsights currently includes an Adaptive Mapper capability that allows Mappers to successively process multiple splits for a job, and avoid the start-up costs for subsequent splits.

When a MapReduce job begins, Hadoop divides the data into multiple splits. It then creates Mapper tasks for each split. Hadoop deploys the first wave of Mapper tasks to the available processors. Then, as Mapper tasks complete, Hadoop deploys the next Mapper tasks in the queue to the available processors. However, each Mapper task has a start-up cost, and that start-up cost is repeated each time a Mapper task starts.

With BigInsights, there is not a separate Mapper task for each split. Instead, BigInsights creates Mapper tasks on each available processor, and those Mapper tasks successively process the splits. This means that BigInsights significantly reduces the Mapper start-up cost. You can see the results of a benchmark for a set-similarity join workload in the following chart. In this case, the tasks have a high start-up cost. The AM bar (Adaptive Mapper) in the chart is based on a 32MB split size. You can see that by avoiding the recurring start-up costs, you can significantly improve performance.

Of course, if you chose the largest split size (2GB), you would achieve similar results to the Adaptive Mapper. However, the you might potentially expose yourself to the imbalanced workloads that sometimes accompany very large splits.

The following chart shows the results of a benchmark for a join query on TERASORT records. Again the AM bar (Adaptive Mapper) in the chart is based on a 32MB split size.

In this case, the Adaptive Mapper results in a more modest performance improvement. Although, it is still an improvement. The key benefit of these Adaptive MapReduce features is that they eliminate some of the hassles associated with determining the split sizes, while also improving performance.

As I mentioned earlier in this post, a number of additional Adaptive MapReduce features are currently in development for future versions of BigInsights. I look forward to telling you about them when they are released…

In the mean time, make sure to check out the free online Hadoop courses at Big Data University. I previous blogged about my experiences with these courses in Hadoop Fundamentals Course on BigDataUniversity.com.

Comparing HDFS and GPFS for Hadoop

Conor O'Mahony — Wed, 30 Nov 2011 17:07:48 +0000

Here is a chart that compares the performance of Hadoop Distributed File System (HDFS) with General Parallel File System-Shared Nothing Cluster (GPFS-SNC) for certain Hadoop-based workloads (it comes from the Understanding Big Data book). As you can see, GPFS-SNC easily out-performs HDFS. In fact, the book claims that a 10-node GPFS-SNC-based Hadoop cluster can match the performance of a 16-node HDFS-based Hadoop cluster.

GPFS was developed by IBM in the 1990s for high-performance computing applications. It has been used in many of the world’s fastest computers (including Blue Gene and Watson). Recently, IBM extended GPFS to develop GPFS-SNC, which is suitable for Hadoop environments. A key difference between GPFS-SNC and HDFS is that GPFS-SNC is a kernel-level file system, whereas HDFS runs on top of the operating system. This means that GPFS-SNC offers several advantages over HDFS, including:

Better performance
Storage flexibility
Concurrent read/write
Improved security

If you are interested in seeing how GPFS-SNC performs in your Hadoop cluster, please contact IBM. Although GPFS-SNC is not in the current release of InfoSphere BigInsights (IBM’s Hadoop-based product), GPFS-SNC is currently available to select clients as a technology preview.