IO performance failure caused by an array card battery failure

A Dell server, purchased in 2009, installed ESX 4.1, recently received the use of the department reflects the "VM running slowly."

Watch the machine on the open VM about 10, CPU utilization of about 50%, memory utilization rate of about 60%, load is not heavy.

The machine stores an average write/read latency (50MS) significantly higher than other machines (5ms), suspected of bottlenecks in IO.

At the same time, the machine RAID card battery error, in the health not good state (for several weeks, due to the purchase procedures and feel that the problem has not been replaced in time).

Then look at the array card TTY log,

Current Cache Policy:writethrough, readadaptive, Direct, No Write Cache if Bad BBU

HDD0 Media Error Count:2

Cache policy is writethrough rather than writeback.

Decisively contact the supplier immediately send a battery, replace the battery back to normal, observe performance chart, lag time of about 5ms.

The reason is that BBU is in the learning State and automatically changes the cache policy to Writethrough.

Thanks to the great God article: http://www.igigo.net/archives/251

See more Highlights: http://www.bianceng.cn/Servers/zs/