Difference between revisions of "Talk:Containers/Guarantees for resources"
| (One intermediate revision by the same user not shown) | |||
| Line 1: | Line 1: | ||
Hi, | Hi, | ||
| − | I guess limiting is only useful when there's enough room for all groups. Eg.: the calculation | + | I guess limiting is only useful when there's enough room for all groups. Eg.: the calculation will make sense only if: |
<code>(1) ∀ i: Li >= Gi</code> | <code>(1) ∀ i: Li >= Gi</code> | ||
| Line 10: | Line 10: | ||
, and the above condition is enough: | , and the above condition is enough: | ||
| − | <code>(3) L1 = (1/N-1) * ((R-G2)+(R-G3)+...+(R-Gn) - (N-2)*(R-G1)) > G1 </code> | + | <code>(3) L1 = (1/N-1) * ((R-G2)+(R-G3)+...+(R-Gn) - (N-2)*(R-G1)) >= G1 </code> |
⇔ | ⇔ | ||
| − | <code>(4) ((R-G2)+(R-G3)+...+(R-Gn) - (N-2)*(R-G1)) > (N-1)*G1</code> | + | <code>(4) ((R-G2)+(R-G3)+...+(R-Gn) - (N-2)*(R-G1)) >= (N-1)*G1</code> |
⇔ | ⇔ | ||
| − | <code>(5) (N-1)*R-(G2+G3+...+Gn)-(N-2)*R+(N-2)*G1 > (N-1)*G1</code> | + | <code>(5) (N-1)*R-(G2+G3+...+Gn)-(N-2)*R+(N-2)*G1 >= (N-1)*G1</code> |
⇔ | ⇔ | ||
| − | <code>(6) R - (G2+G3+...+Gn) + (N-2)*G1 > (N-1)*G1</code> | + | <code>(6) R - (G2+G3+...+Gn) + (N-2)*G1 >= (N-1)*G1</code> |
⇔ | ⇔ | ||
| − | <code>(7) R > G2+G3+...+Gn+G1</code> | + | <code>(7) R >= G2+G3+...+Gn+G1</code> |
Thus limiting is only useful if the natural condition under (2) holds. | Thus limiting is only useful if the natural condition under (2) holds. | ||
Latest revision as of 05:25, 7 May 2010
Hi, I guess limiting is only useful when there's enough room for all groups. Eg.: the calculation will make sense only if:
(1) ∀ i: Li >= Gi
Otherwise, the group limit is too strict.Obviously it's required that:
(2) ∑ Gi <= R
, and the above condition is enough:
(3) L1 = (1/N-1) * ((R-G2)+(R-G3)+...+(R-Gn) - (N-2)*(R-G1)) >= G1
⇔
(4) ((R-G2)+(R-G3)+...+(R-Gn) - (N-2)*(R-G1)) >= (N-1)*G1
⇔
(5) (N-1)*R-(G2+G3+...+Gn)-(N-2)*R+(N-2)*G1 >= (N-1)*G1
⇔
(6) R - (G2+G3+...+Gn) + (N-2)*G1 >= (N-1)*G1
⇔
(7) R >= G2+G3+...+Gn+G1
Thus limiting is only useful if the natural condition under (2) holds.
Cheers,
Gyula
