Planning for uninterrupted service

Planning for load balancing in the event of disk failure

In the event of an appliance disk failure (in an acceleration deployment), the appliance might not be able to keep up with requests and will proxy them to the servers it is accelerating. This can be problematic if your back-end servers cannot handle the increased load. If you are concerned that your back-end servers are not powerful enough to handle the load if an appliance disk fails, you can configure your load balancers to query the appliance for disk health and redirect traffic accordingly.

Querying for disk health

To query the appliance for disk health, send a request to the following URL:

http://cache_hostname:3132/cache_manager/disk_status

The appliance will send one of the following disk status messages in response to the preceding HTTP request:

If the appliance is handling requests normally, it will issue the following status message:
```
HTTP/1.0 200 OK
```
```
X-Cache-Status: DISK OK
```
If the appliance cannot keep up with incoming requests, it will send the following status message:
```
HTTP/1.0 500 ERROR
```
```
X-Cache-Status: DISK BUSY
```

Changing the disk busy threshold

The default value of the disk busy threshold is 10. The default value of 10 indicates that 10 percent of misses were proxied to back-end servers in the last five minutes because the disk was too busy. Use the config.http.disk_busy_thr option to change the disk busy threshold. For example:

set cofig.http.disk_busy_thr //

//