Performance Tuning a U2 Database - Memory

In computing, memory refers to a device that is used to store information for immediate use in a computer or related computer hardware device. It typically refers to semiconductor memory, specifically metal–oxide–semiconductor (MOS) memory, where data is stored within MOS memory cells on a silicon integrated circuit chip. The term "memory" is often synonymous with the term "primary storage" or “main memory”. There is also memory that is associated with the CPU known as cache memory. Both memory types are extremely fast to keep up with today’s high-speed processors.

When using the U2 Databases, a good rule of thumb for sizing memory is as follows:

1. The memory should be sized at 25% of the “Dynamic Data Set”. The “Dynamic Data Set” is defined as the data files that are consistently being touched by an application such as a customer file or orders file. 
2. The U2 database memory requirement is approximately 7% of the total size of the database size and that will give you memory per user.
3. Finally, the typical UNIX environment like AIX, Solaris and HP/UX usually requires 8 – 10 GB.

An example of this would be as follows:

I have to size a system to capacitate 500 users for a UniVerse system running an Order Entry System. Here are the givens I have to work with:

1. The total database size is 250GB
2. The total “Dynamic Data Set” is 190GB
3. The OS is AIX 7.1

• With this information, I first calculate the base memory size which would be 190GB X .25 (25%) = 47.5GB
• Next I calculate the UniVerse Database total memory per user. 250GB X .07 (7%) = 17.5GB. That works out to 17.5GB / 500 = 35MB per user
• The ideal OS size for AIX 7.1 is 8GB
• Now I add up everything- 47.5GB + 17.5GB + 8GB = 73GB. I would then round up to 80GB.

These formulas should be part an IT department’s “Best Practices”. Also, whenever possible make sure some of your larger files stay cached in main memory / primary storage. UNIX flavors like AIX and HP/UX try to utilize maximum memory for that purpose. Also maximizing cache for the processors is highly recommended based on the U2 Databases performing better with a larger number of processors and less cores. There are tuning considerations involved with main memory and cache having to do paging, choice of algorithms and keeping certain data files “memory resident”

In the next blog, we will continue to cover the three major resources of the computer: CPU, Memory and Disk.