IO quantity, LGWR process is very high, this how to optimize the

We have some real time acquisition system program, so redo set up three groups, each group of three members of 4 g,
Because insert updates a lot, so the LGWR process is consistent in the execution, can optimize the function of the stop or you

CodePudding user response:

LGWR is the process of oracle background cannot little, you write the process of the quantity is according to you at the front desk business
Do you want to decrease the insert from the front

CodePudding user response:

So many members do? A written four can not slow? The underlying storage if made a raid 1, no need, one is enough,

CodePudding user response:

refer to the second floor good memory not written reply of ABC:

so many members do? A written four can not slow? The underlying storage if the raid 1, no need, one is enough,

I am afraid that a member of 4 g, soon write full, a jam

CodePudding user response:

In fact business is not so big, but writing too much every second, and illogical,

CodePudding user response:

reference leaflets 3 floor daily response:

Quote: refer to the second floor is better than good memory bad written reply of ABC:

so many members do? A written four can not slow? The underlying storage if the raid 1, no need, one is enough,

I am afraid that a member of 4 g, write full, soon appear jams

Fear is add a few more group quickly, rather than add members

CodePudding user response:

good memories than bad written references 5 floor ABC reply:

Quote: refer to the third floor of leaflets of daily response:

Quote: refer to the second floor is better than good memory bad written reply of ABC:

so many members do? A written four can not slow? The underlying storage if the raid 1, no need, one is enough,

I am afraid that a member of 4 g, write full, soon appear jams

Fear is add a few more group quickly, rather than add members

Good bosses, delete the six members, currently redolog only three groups, each of the 4 g,
Now look at IO some lower, is there any other optimization methods

CodePudding user response:

1. Storage with what? Mechanical sas 10 k/s? 15 k/s? Or SSD disk?
2. What are the business, real-time DML so much?
3. The warnings and alert log to see? Checkpoint not complete?

CodePudding user response:

refer to 7th floor good memory not written reply of ABC:

1. The store with what? Mechanical sas 10 k/s? 15 k/s? Or SSD disk?
2. What are the business, real-time DML so much?
3. The warnings and alert log to see? Checkpoint not complete?

Checkpoint not complete there are a lot of, I hear that is increase redolog, so didn't appear more than one member,,,

CodePudding user response:

I give a awr::https://linksharing.samsungcloud.com/8J1WvAG3mutj bosses behind

CodePudding user response:

Disk IO performance not, answer 1 store with what? Mechanical sata7.2 k/s? Sas 10 k/s? 15 k/s? Or SSD disk? Do the raid?

CodePudding user response:

Who runs awr in the top of the optimized SQL optimization, obvious to empty all data, delete should change to truncate

CodePudding user response:

references to the tenth floor good memory not written reply of ABC:

disk IO performance not, answer 1 store with what? Mechanical sata7.2 k/s? Sas 10 k/s? 15 k/s? Or SSD disk? Do the raid?

Sas is 10 k/s, but didn't do raid

CodePudding user response:

Is a 11 t disk

CodePudding user response:

Read and write but I think there are more than 100 m/s, should can also use

CodePudding user response:

on the 14th floor leaflets of daily reference response:

read and write but I think there are more than 100 m/s, should also can put the

The 11 floor, optimize the top SQL first

CodePudding user response:

15 floor good memories than bad written reference ABC reply:

Quote: old leaves of daily reference 14 floor response:

but I think reading and writing more than 100 m/s, should also can put the

See the 11 floor, optimize the top SQL first

Bosses, I just see there are not many delete, are generally insert

CodePudding user response:

The two delete not frequent operation

CodePudding user response:

1. Large amount of IO change the SQL have to dispose of,
2. Post your alert log, log group and 6 groups, the current size of each log is 4 g?