How are Dedupe Ratios Calculated in Element Software?
Applies to
Element Software
Answer
Deduplication (dedupe) refers to the elimination of redundant data by Element Software. This occurs at the individual block level -- if a given block's data is identical to another's, only one block will be retained, and the metadata will point to the existing block rather than to two identical blocks.
The dedupe ratio is the ratio of total data to unique data in a given context (for instance, globally across the cluster or within a given account). This ratio can be a little hard to understand, in that a small change in the ratio corresponds to a larger percentage change in the amount of shared data.
For example, take a cluster with 4 VMs each 20GB in size.
If they share 5GB of data each (25%) the dedupe ratio would be (80 GB Total / 65 GB Unique) = 1.23x Dedupe
If they share 10GB of data each (50%) the dedupe ratio would be (80 GB Total / 50 GB Unique) = 1.6x Dedupe
If they share 15GB of data each (75%) the dedupe ratio would be (80 GB Total / 35 GB Unique) = 2.3x Dedupe
In the example, the Dedupe ratio between the first two changes by less than four tenths, but the percentage changes from 25% to 50% of shared data. The ratios change more drastically beyond 50% of shared data.
Additional Information
Sample data based on 1000 volumes and 250TB of used space.
Percent Shared | Unique Data | Total Data | Dedupe Ratio |
---|---|---|---|
0% | 250000000 | 250000000 | 1.00 |
5% | 237512500 | 250000000 | 1.05 |
10% | 225025000 | 250000000 | 1.11 |
15% | 212537500 | 250000000 | 1.18 |
20% | 200050000 | 250000000 | 1.25 |
25% | 187562500 | 250000000 | 1.33 |
30% | 175075000 | 250000000 | 1.43 |
35% | 162587500 | 250000000 | 1.54 |
40% | 150100000 | 250000000 | 1.67 |
45% | 137612500 | 250000000 | 1.82 |
50% | 125125000 | 250000000 | 2.00 |
55% | 112637500 | 250000000 | 2.22 |
60% | 100150000 | 250000000 | 2.50 |
65% | 87662500 | 250000000 | 2.85 |
70% | 75175000 | 250000000 | 3.33 |
75% | 62687500 | 250000000 | 3.99 |
80% | 50200000 | 250000000 | 4.98 |
85% | 37712500 | 250000000 | 6.63 |
90% | 25225000 | 250000000 | 9.91 |
95% | 12737500 | 250000000 | 19.63 |
100% | 250000 | 250000000 | 1000.00 |