How many is good - arch logs?

Recently i have created a database and went live. When i created the db from dbca, redo logs files were created as default 100mb. I further created members in the group.

I have set fast start mttr target as 1800, log checkpoint as zero and log checkpoint interval 1800.

When observing over a few weeks i could see that only one file getting created as archive log per day. The db is small and not much activities. In case if ineed to restore db then i may lose a days data as the current online log would not been archived.

Under the above circumstances:

> should i create a small size redo logs say 20mb and see more arch log files created.

> any other valid suggestions

The question is "how much data can I afford to lose?".

If it's 10 minutes, then issue an ALTER SYSTEM SWITCH LOGFILE every 10 minutes.

If it's NONE then investigate Data Guard.

you can use archive_lag_traget

Thanks Davey :)

You need to guarantee the online logs are not lost.
Do this by creating multiple members per log group on seperate disks.
That way if an individual drive fails you have not "lost" the online data.

Recovery would involve restoring the current online log as well as any archive logs that are needed.

As i said earlier, i have log members located on separate disks and they are secure.

Also i have only one arch log per day and is this healthy? :confused: ?. This one lil worries me.

I have read that the logs be switched every 30mins to have arch logs in sufficient numbers per day. Does this needs to be done manually? or this is implicitly done by setting init parameter.

I have not configured Dataguard for this db.

Conventional Wisdom (to which I don't subscribe) says to have your logs switch about every 30 minutes. However, most people have the opposite problem, switching too often. You dont need to switch more than once a day as long as your checkpoint interval is configured correctly, but I'm a big believer in getting it in the archived redo log.

Personally, I force my logs to switch every 15 minutes, but I run standby dbs.

Quote:

---

Originally Posted by ixion

Recovery would involve restoring the current online log as well as any archive logs that are needed.

---

WHAT? From where would you restore the current online redo?

SCN and time based recovery. :)

Quote:

---

Originally Posted by DaPi

WHAT? From where would you restore the current online redo?

---

My mistake in wording: "Restoring".. (I meant restoring in a general sense not to literally restore from tape or backup set) You wouldn't need or want to restore, just use a valid one or make sure that a valid online log from one of the members exists. ;)

No, you would need online redo logs.

1) Any changes in the online redo logs are applied in a time based recovery.

2) applied to make the datafiles current.

Thanks to all Guru's for your time and views.

Quote:

---

Originally Posted by dbasan

No, you would need online redo logs.

1) Any changes in the online redo logs are applied in a time based recovery.

2) applied to make the datafiles current.

---

1. Time based recovery, only will apply up to a point in time. Where the online logs may or may not include redo data from that time frame.

2. yes, in a complete media recovery or instance recovery.

before i execute my backup for archive logs, I make sure that all online redo logs are archive. :)