内部监控项

1. System-level:

2. JVM-level:

3. IoTDB-level:

Workload: read and write related workloads 

Metadata:

name: # of SGs

tags: user=<user_name>

name: # of Devices

tags:  sg=<name>, user=<user_name>

name: # of Series 

tags: sg=<name>, datatype=<int32 | int64 | float | double | string | bool>, user=<user_name>

Write:

name: write latency ( histogram of time cost)

tags: sg=<name>, api=<sql | record | records | tablet | tablets>, user=<name>

name: request per second (TPS) (Rate)

tags: sg=<name>, user=<name>

name: point per second (TPS) (Rate)

tags: sg=<name>, user=<name>

name: income bandwith (KB/s) (Rate)

tags: sg=<name>, user=<name>

Query:

name: query latency ( histogram of time cost)

tags: sg=<name>, query type=<last | aggregation | group by | raw | udf> (need to confirm)?, user=<name>

name: time range per query (histogram of time range, or add a tag (whose name is time_range))

tags: sg=<name>, query type=<last | aggregation | group by | raw | udf> (need to confirm)?, user=<name>

name: request per second (TPS) (Rate)

tags: sg=<name>, query type=<last | aggregation | group by | raw | udf> (need to confirm)?, user=<name>

name: point per raw query (TPS) (Rate)

tags: sg=<name>, user=<name>

name: outcome bandwith (KB per raw query) (Rate)

tags: sg=<name>, user=<name>

Modules:

WAL

name: throughput (KB/s)

tags: sg=<name>

Memtable:

name: number of memtable

tags: sg=<name>

name: the size of memtables (histogram, points or MB), only update when memtable will be flushed

tags: sg=<name>

name: the size of chunks (histogram, points or MB), only update when memtable will be flushed

tags: sg=<name>

name: the duration of memtables (histogram, seconds), only update when memtable will be flushed

tags: sg=<name>

Flush: 

name: time cost of flushing a memtable

tags: sg=<name>

name: time cost of ordering data

tags: sg=<name>

name: time cost of encoding

tags: sg=<name>

name: time cost of IO

tags: sg=<name>

name: size of encoded memtable

tags: sg=<name>

name: number of pending tasks

tags: sg=<name>

name: number of flushing tasks

tags: sg=<name>

Compaction

• TTL
• Unseq
• ...

Lingzhe Zhang

as detailed as possible for diagnosing.

name: number of pending files

tags: sg=<name>

Cluster

Xinyu Tan

as detailed as possible for diagnosing.

QueryExecutor

Xiangwei Wei

Cache

Size and Hit Ratio  Yuan Tian

MManager

Zesong Sun

Last 

UDF

MQTT

Sync