Scenarios for custom alerts
Create custom alerts with one or more conditions that trigger when all the conditions are met for a given resource. By creating a custom alert, you can detect multiple configuration, capacity, and performance conditions together to determine whether an urgent situation occurred in your storage or SAN fabric.
- Receive alerts when the response time of storage on a specific tier is too high
- Receive alerts when the response time of volumes is too high during times of active I/O
- Receive alerts when the response time of volumes is too high, but do not generate these alerts when batch and backup jobs are running
- Receive alerts if a port is being used for both inter-node communication and host I/O exchanges
- Receive alerts for link resets that are not associated with link initialization
- Receive alerts for invalid word transmissions that are not associated with link initialization
Receive alerts for Sustainability values
To monitor the power and temperature Service Level Agreements (SLAs) of your storage system, you want to get notified when the power of your storage system exceeds a certain defined value or SLA.
- Solution
- Define a custom alert by specifying the Total Power Consumed greater than or equal to (>=) a value (for example, 70) along with the level of Severity.
Receive alerts when the response time of storage on a specific tier is too high
Your storage systems run both critical production applications and noncritical test applications. The production applications use tier 1 storage, while the test applications use storage on tiers 2 and 3.
To ensure consistent, top performance for tier 1 storage, you want to be notified when its response time is higher than 6 ms/op so that you can resolve the bottleneck. However, to avoid too many alerts, you do not want to receive notifications when the response time of tier 2 or 3 storage exceeds 6 ms/opop.
- Solution
- Define a custom alert that checks if the volumes used by an application are in Tier 1 pools and if their Overall Response Time is higher than 6 ms/.
Receive alerts when the response time of volumes is too high during times of active I/O
You care about high read response times on your volumes, but they can be caused by cache misses when there is only a trickle of I/O.
- Solution
- Define a custom alert with volume-level thresholds that combines checks for response times and I/O.
Receive alerts when the response time of volumes is too high, but do not generate these alerts when batch and backup jobs are running
To ensure the consistent, top performance of your volumes, you want to be notified when their response times are becoming too high. However, to avoid too many alerts, you do not want to receive notifications when batch and backup jobs are running on your storage. You understand that these jobs can cause an expected spike in response times and do not require action on your part.
- Solution
- Define a custom alert that checks if the Read Response Time of volumes exceeds an amount that is more than expected in your environment and the Read Transfer Size is less than 256 KiB/op. Typically, read transfer sizes greater than 256KiB/op indicate that batch or backup jobs are running in the background.
Receive alerts if a port is being used for both inter-node communication and host I/O exchanges
You want to avoid potential bottlenecks by ensuring that storage system ports aren't being used for both inter-node communication in the local cluster and for I/O exchanges to host computers. You can also use this custom alert to check for adherence to best practices that are related to configuring ports for nodes with 8 or more ports. It does not apply to nodes that contain only 4 ports.
- Solution
- Define a custom alert that checks if the I/O rate for ports indicates exchanges between local nodes and hosts. For a storage system, set up a custom alert with the following attributes and conditions:
Receive alerts for link resets that are not associated with link initialization
You want to identify link resets that are generated in response to hardware failures or link congestion. Link Resets generated by link initialization are ignored.
- Solution
- Define a custom alert that checks if link resets occur and if those resets are not associated with a link initialization. For a switch, set up a custom alert with the following attributes and conditions:
Receive alerts for invalid word transmissions that are not associated with link initialization
You want to identify invalid transmission words that are generated because of poor link quality. Poor or marginal link quality can be caused by a bad SFP, HBA, or cable. Invalid transmission words that are generated by link initialization are ignored.
- Solution
- Define a custom alert that checks if invalid word transmissions occur and if those resets are not associated with a link initialization. For a switch, set up a custom alert with the following attributes and conditions: