How does splunk write data into Hot DB?

[yuwtennis]

Hi!

I would like to ask question regarding to Hot DB.

I understand that you can specify multiple Hot DB using the maxHotBuckets option
and whenever the maxDataSize is reached per bucket it is rolled to the warm status.
(hot bucket and warm bucket is the same size)

Does splunk first create maxHotBuckets and then writes data sequentially or
does it load balances through all the buckets?

I am bit confused with the relation with maxDataSize and maxHotBuckets.
I'd appreciate if someone can clarify with this issue.

Thanks,
Yu

[gfuente · Accepted solution]

Hello

It is related to the timestamps of the events. In a normal situation where all the events are beign received one after another in the timeline, you would have only one hot bucket, and it will be filled until it reaches the maxDataSize limit.

But, you could get events from the past that won't "fit" in the time frame of the existing hot bucket, and in that situation Splunk will create a new hot bucket for the events coming from that "past time frame", in this case you will have 2 hot buckets.

So, hot buckets groups the events by time.

Regards

[yuwtennis]

Hi gfuente.

Thank you for the reply.

I think I am beginning to understand this nature.

If a past event comes in and splunk creates a Hot bucket,
and the next incoming events fits to the time frame of the buckets,
which bucket will be used?

Thanks,
Yu