Kbase P50004: How does the parameter -spin effect load balancing on multi processor machines?
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
15/12/2005 |
|
Status: Verified
GOAL:
How does the parameter -spin effect load balancing on multi processor machines?
FIX:
The -spin parameter is a database startup parameter which sets the spin lock algorithm on as opposed to a system of semaphores and queues which control shared memory locks when -spin = 0. When it's set, instead of 'napping' between attempts to own a latch, we keep retrying -spin times between 'naps'. Finding the "sweet point" for -spin on any environment is particular to that environment. It is essentially a performance tuning exercise for which we offer consultancy.
Load balancing on the other hand is an OS responsibility. Distributing processing and communications activity evenly across a computer network so that no single device is overloaded.
So essentially, load balancing, has to do with multiple servers that are 'load balanced', whereas -spin has to do with optimising database lock waits on a multi-processor machine.
The solutions referenced below discusses -spin in more detail and more particularly, the effect that having -spin set on one database and not on the other on the same server, which may be of further interest in this discussion.