How to chose the appropriate IOPS limits in Virtual Machines?


i need configure iops limits on hyper-v vms

how can measure iops of machines make sure configure appropriate , "safe" level of iops?

i´ve used iometer before enter production , tests was:

100% read 512 b =~13.500 iops

75% write 64 kb =~ 700 iops

1000gb test file, in 8 x 1.2tb sas 10k rpm disks in raid-5 in dell vrtx , m630 dual-xeon 256 gb of ram blades

i need make sure configure iops limit make sure proper operariont of vms , low risk of 1 vm "eating" available iops , generating slowliness on other vms


1. here nice description how it...

http://blog.synology.com/?p=2086

2. raid 5 known fo read properties , worse writes. use if rely more on reading. better writing characteristics prefere raid 10

3. optimize disk cluster size according files , tune nic properties accordingly.

4. rely heavily on best practices tools ( mean native ones).

5. consider roles of vms in load , in time. model , test various load situations correspond real conditions.

6. overlooked baseline. create baseline comparison in tuning.

7. technet , msdn give advices tuning vm. here 1 msdn

https://msdn.microsoft.com/en-us/library/dn567657.aspx

regards

milos

ps: next time consider ssd disks :-)



Windows Server  >  File Services and Storage



Comments

Popular posts from this blog

Generating Event ID 91, 44 with CA (In Event Viewer with source as Certsvc)

Group policy default policy cannot be edited because network name not found

Problem when changing update source