How it works
You may have heard about the recent IBM DB2 10.5 announcement, which includes BLU Acceleration technology. With DB2 10.5, you have the option to insert data into the database as columnar ordered data sets that compress the data as it is being loaded. Because of the high compression rate, DB2 can work with more data in memory than before. If you have a large data warehouse and want to run analytics but don’t have a lot of space or funding for storage, you can take a traditional row based database and load it into a new BLU Accelerated database using column organized tables. Many clients are seeing a difference of 10x compression or more, which means your traditional row based database that consumed 10TB of storage will now only consume 1TB of storage. The 1TB of BLU compressed storage does not need to be decompressed to work with it, all of the compressed data is fully actionable in its current state whether it is residing on disk or loaded into memory. Now that your data is now highly compressed and actionable at the same time, more of it can reside in memory and cache, which decreases the response time of loading and querying the data. So basically, you can execute more queries in a shorter amount of time with a larger data set.
BLU Acceleration and IBM Power Systems
In addition to decreasing data loading and querying time, compression allows you to use less storage space so you can use a smaller number of faster SSDs versus a larger number of slower HDDs. SSDs will significantly enhance performance on a columnar-organized database. Since you now have a smaller server footprint, you may also see a decrease in energy requirements.
Why this is exciting
BLU Acceleration is leveraging some of the technology that differentiates Power Systems to increase database performance, such as SMT4 technology (meaning four hardware threads on a single core vs. two on x86). BLU also makes effective use of POWER’s large L3 cache, and is able to use POWER’s wide range of SIMD instructions to accelerate run-time performance, particularly arithmetic processing.
What kind of work are you doing with DB2 BLU Acceleration?
I’m getting a lot of questions from IBMers all around the world asking about what size platform and resources they need to run a DB2 10.5 columnar-organized workload. Right now, I am testing and documenting the optimal amount of storage, memory and CPU required for DB2 BLU on Power Systems. I’m very impressed with how fast this technology is processing data. I was able to load multiple terabytes into this new version of DB2 within just a few hours.
Beta clients are excited about BLU too. BNSF Railway Company, one of the largest freight rail transportation networks in North America, is using BLU to more quickly understand the vast amounts of data from their 1,700 servers that track maintenance, weather, scheduling, inventory, safety, deliveries, etc. BNSF said that because of BLU, one of their queries improved over 100-fold, and their storage consumption went down by about 10 times. Also, check out this video from Coca-Cola. They’re seeing compression rates of 90-95 percent with queries coming back in seconds.
For more information, go to the DB2 BLU Acceleration web site, which talks about its many benefits and what clients are saying.
Connect with Douglas on LinkedIn
Douglas Gibbs is an IBM Systems Optimization Competency Centre team member in IBM Software Group.
To effectively compete in today’s changing world, it is essential that companies leverage innovative technology to differentiate from competitors. Learn how you can do that and more in the Smarter Computing Analyst Paper from Hurwitz and Associates.