Normal disk read / write value for ubuntu server
How would you determine the maximum value to set an alarm for I / O activity on an ubuntu / linux server hosting up to 4 sites running apache, mysql and up to 4 cats? Or in general (if any) what is a suitable operating range for read / write I / O per second on a production server with a 7.2KB SATA hard drive?
What I am trying to understand is the impact on disks when reading / writing per second. What should I look for in the drive / drive specification to determine this?
As you can see, I am "obviously" confused. Any help / direction would be appreciated ...
a source to share
How would you determine the maximum value to set an alarm for I / O activity on a ubuntu / linux server hosting up to 4 sites running apache, mysql and up to 4 cats?
The value at which the expected problems you are worried about is more expensive than the cost of paying attention to the anxiety.
What is this number? It depends on many things, including:
What problems are you trying to avoid?
Are you worried about performance? If so, are you more worried about latency or bandwidth? How is the trade-off between interactive and batch work?
Are you worried about the wear and tear and longevity of your media? Are you worried about how often you have to restore your backups?
Are you worried about the price of discs? How important are the best discs that can lead to your job?
How many entries can you defer? How many reads can caching prevent? How weak can you relate to independence (I in ACID)?
If you really want the best drive for your situation, here are some of the questions you probably want to ask yourself. If I were in your situation, I would pick a random drive from a range of low and low prices and then see how it works. Then you have the experience to learn, so you know what to do differently next time (if any) and it won't cost you much.
a source to share
Beware of Linux, you will have some data cached in memory (which is cool because it is fast, but it will break your benchmarking).
IMHO you shouldn't check the bandwidth of hdd, as if the data were accumulated on a physical disk, you will have more fragmentation of data results too ... Use some other metrics.
At the top, you have a "0.0% wa" statistic that tells you how much the processor is spending waiting for data. If it gets high, you're in trouble (you can make a RAID array to get more bandwidth). I don't know how the top gets this information, but I'm sure you can get it too.
a source to share