Monitoring Exadata Storage Cells - Oracle Exadata Recipes: A Problem-Solution Approach - page 379

Grid Disks (objectType = GRIDDISK)
Exadata grid disk metrics are broken down into two main categories: I/O requests and I/O throughput. These metrics
are further broken down into operations for small I/O and large I/O, and each of these is also classified based on a
cumulative count and rate metric.
Monitoring grid disk metric values can be a bit more complex than monitoring cell disk metrics because your
databases can be spread unevenly across your Exadata grid disks depending on your deployment. We recommend
combining a grid disk metric monitoring strategy with an IORM Database monitoring strategy, as presented in
Recipe 13-12.

Host Interconnect (objectType = HOST_INTERCONNECT)
Exadata tracks a number of InfiniBand interconnect metrics using the HOST_INTERCONNECT metric objectType :

N_MB_DROP* , N_MB_RDMA_
DROP* , N_MB_RESENT* , and N_RDMA_RETRY* metrics. Ideally, you should not see non-zero values
for these metrics.

• Watch for drops, retransmissions, and resend metrics using the

• Look for the number of MB sent and received to and from the compute nodes to be smoothly

distributed.

IORM Databse (objectType = IORM_DATABASE)
Many IORM_DATABASE metrics are collected whether you have an IORM plan enabled or not; Exadata automatically
tags the I/O request metadata instructions with the database sending the I/O request to the Exadata storage cells.

DB_IO_LOAD metric, which reports the disk I/O load per database. This is a good
way to measure which database and application are generating the highest I/O workload.

• Measure the

DB_IO_UTIL * metric, which reports the disk I/O utilization for small and large
I/Os per database. This is a good way to measure which database and application is generating
the highest disk utilization.

• Measure the

DB_IO_WT * metrics. These metrics will
show you which databases are suffering wait time based on IORM plan I/O queuing and can
be used to validate or adjust your IORM plan.

• If you have implemented an IORM plan, monitor the

IORM Consumer Group (objectType = IORM_CONSUMER_GROUP)
Oracle automatically collects IORM consumer group performance metrics on each storage server for each database
and consumer group deployed on your Exadata Database Machine when an IORM plan is enabled.

CG_IO_LOAD metric, which reports the disk I/O load per database per consumer
group. This is a good way to measure which database, application, and resource consumer are
generating the highest I/O workload.

• Measure the

CG_IO_UTIL * metrics, which report the disk I/O utilization for small and large
I/Os per database per consumer group. This is a good way to measure which database and
resource consumer group are generating the highest disk utilization.

• Measure the

CG_IO_WT * metrics. These metrics will
show you which consumer groups are suffering wait time based on IORM plan I/O queuing
and can be used to validate or adjust your IORM plan.

• If you have implemented an IORM plan, monitor the