The following example explains you how much memory is required to calculate the deduplication percentage and generate deduplication reports:
Consider 1 GB source data and 60% data deduplication, the size of unique data is 1 GB * (1 – 60%) = 409.6 MB.
If deduplication block size is 4 KB, the number of hash keys generated are: 409.6 MB/4 KB = 102.4 KB.
The required memory is 102.4 KB * 20 bytes = 2 MB.
So, 409.6 MB unique data requires 2 MB memory.
409.6 MB: 2 MB = 204.8: 1
Every 100 GB unique data needs about 512 MB memory.
If SSD is not enabled, all the unique hash keys are stored in local memory. Every hash key corresponds to about 20 bytes (including hash key, page index and others).
In the backend, Arcserve UDP Data Store Capacity Planning Tool initially calculates the deduplication percentage for 4 KB deduplication block size. Arcserve UDP Data Store Capacity Planning Tool releases all the occupied memory once deduplication percentage for 4 KB deduplication block size is completed. Then Arcserve UDP Data Store Capacity Planning Tool calculates the deduplication percentage for 8 KB deduplication block size.
As the memory requirement of 8 KB calculation is much smaller compared to 4 KB, the peak value of memory requirement is at the end of calculating deduplication percentage for 4 KB deduplication block size.
In addition, if the memory of server is adequate enough to run the utility, the occupied memory to calculate deduplication percentage might be bigger. The extreme value of "unique data size: memory size" is "130:1".
If SSD is enabled, the required memory is greatly decreased. The "unique data size: memory" ratio is "2000:1".
|
Copyright © 2021
|
|