Exadata Storage Panel
Note: Spotlight on Oracle RAC may include monitoring of Exadata systems depending on the license purchased.
Cell Averages
Measure how much the Exadata Storage Servers (Cells) are used.
| Metric | Description |
|---|---|
| Cell IO |
The rate of read and write requests over all Exadata Storage Servers (Cells), Hard Disk and Flash Disk. |
| Avg. CPU Load | Average CPU load across all Exadata Storage Servers (Cells). The metric is calculated as User + Kernel time expressed as percentages. |
| Single Block Read | The time taken to serve a single block read from a datafile on the Exadata Storage Servers (Hard Disk). |
| Flash Disk Latency | The time taken to serve a single block read from a datafile on the Exadata Flash Disk Space. |
Smart Scans
Exadata's Smart Scans optimize SQL statements executed in Exadata's Storage Servers. Smart Scans are typically for scan operations like Full Table Scans.
| Metric | Description |
|---|---|
| % IO Saved |
An indicator of the number of bytes saved in I/O by using Smart Scans. The proportion of:
For example: If a table is 100MB and about to be smart scanned, then Exadata returns 50MB, Exadata Smart Scan saved 50MB of IO (50%). |
| % Success |
An indicator of how many of the table scans eligible for Smart Scan, were executed with Smart Scan. Table scans using Direct Path Read are eligible for Smart Scan. Calculated as a proportion of:
|
| Storage Index Saved |
An indicator of the amount of I/O saved in a Smart Scan due to the Oracle Storage Index. This is calculated as a proportion of:
|
Smart Flash Cache
Speed up I/O performance for frequently accessed data. Exadata Smart Flash Cache acts like a large, database-aware disk cache for the storage cell.
| Metric | Description |
|---|---|
| FC Hit % |
The proportion of Exadata Smart Flash Cache (ESFC) hits that are successful. Hits are I/O requests made through the Exadata Smart Flash Cache. Hits are successful when the requested data is found in the Smart Flash Cache. Hits are unsuccessful (Misses) if the requested data needs to be retrieved from the Exadata Storage Servers (Cells) because it cannot be found in the Smart Flash Cache. |
| % IO Time Saved |
An indicator of the time saved by the Exadata Smart Flash Cache (ESFC). The proportion of:
over
|
Exadata HHC
Statistics for Exadata Hybrid Columnar Compression (HCC).
| Metric | Description |
|---|---|
| % Compression units decompressed on cells |
This is calculated as a simple ratio of Cell CUs sent uncompressed over Cell CUs sent. A value of 100% is good. It indicates compressed cells are being uncompressed by the Exadata Storage Servers. Values under 100% indicate compressed cells are being sent to the Exadata Database Servers, for the Database Servers to uncompress them. You want to avoid this as this leads to performance issues. |
Cell Summary
Summary statistics for Exadata's Storage Servers (Cells).
| Metric | Description |
|---|---|
| Number of Cells | The number of cells (Exadata Storage Servers) as recorded in v$cell. |
| Cell Balance |
An indicator of the balance of resource usage across the cells. A high value indicates balance, that the cells are doing equal amounts of work. A low value indicates one cell is doing all the work. The balance measure is based on total reads and writes for each cell. |
| Total Cell Storage |
The total raw disk space in Exadata's Storage Servers (Cells): flash space and hard disk space. Note: Not all of this space is usable due to Exadata's mirroring requirements. |
| Flash Disk Space |
The flash-based storage capacity in Exadata's Storage Servers. Note: This is the total raw storage value. Not all of this space is usable due to Exadata's mirroring requirements. |
| Hard Disk Space |
The hard disk space in Exadata's Storage Servers. Note: This is the total raw storage value. Not all of this space is usable due to Exadata's mirroring requirements. |
System Information
The System Information area contains details of the Oracle version and operating system version.
The Uptime for the Oracle RAC cluster is that of the database instance with the longest uptime. This may underestimate the actual uptime of the cluster.