What three changes will reduce disk I/O overheads?

Full Atomicity, Consistency, Isolation, Durability (ACID) compliance is a necessity for a new
application, which heavily reads and writes data.
This requires the following config file options:
Sync_binlog=1
Innodb_flush_log_at_trx_commit=1
Innodb_doublewrite=1
However, this configuration is expected to introduce disk I/O overhead.
What three changes will reduce disk I/O overheads?

Full Atomicity, Consistency, Isolation, Durability (ACID) compliance is a necessity for a new
application, which heavily reads and writes data.
This requires the following config file options:
Sync_binlog=1
Innodb_flush_log_at_trx_commit=1
Innodb_doublewrite=1
However, this configuration is expected to introduce disk I/O overhead.
What three changes will reduce disk I/O overheads?

A.
Use of soft links for database directories on the same physical disk

B.
Use of separate directories on the same physical disk for log files and data files

C.
Placement of InnoDB log files and datadir on separate physical disks

D.
Allocation of RAM to the buffer pool such that more of the data can fit in RAM

E.
Use of delay_key_write=ON for batch index update

Explanation:



Leave a Reply 7

Your email address will not be published. Required fields are marked *


Cristian

Cristian

The question pretty explicitly asks about reducing disk I/O overhead, input *and* output. Strictly speaking, placing og & data files on separate devices will spread the writes, not reduce them – this is why increasing the size of the buffer pool (D) is a good answer – it reduces reads to disk, and delaying key writes reduces write contention.

mutex

mutex

Spreading writes reduces contention => reduces overhead.

Rajareddy VN

Rajareddy VN

C,D,E