Method and system for dynamic upgrade predictions for a multi-component product
Techniques are disclosed for generating a dynamic upgrade prediction. The prediction includes generating an initial upgrade prediction for an upgrade to be performed on a subset of component nodes; performing a real time progress review, using a centralized management node, of the upgrade, wherein each of the component nodes includes a number of stages; performing an upgrade duration comparison between the initial upgrade prediction and a real time upgrade time for a subset of the stages; and generating an updated upgrade prediction for the multi-component product upgrade based on the comparison between the initial upgrade prediction and the real time upgrade time.
1. A computer implemented method of generating a dynamic upgrade prediction comprising:
generating an initial upgrade prediction for a predicted time of an upgrade to be performed on a subset of a plurality of component nodes of a multi-component product;
performing a real time progress review, using a centralized management node, of the upgrade, wherein each of the plurality of component nodes includes a plurality of stages;
performing an upgrade duration comparison between the initial upgrade prediction and a real time upgrade time for a subset of the plurality of stages; and
generating an updated upgrade prediction for the upgrade of the multi-component product based on the upgrade duration comparison between the initial upgrade prediction and the real time upgrade time, wherein generating the updated upgrade prediction varies based on whether the upgrade duration comparison indicates that the initial upgrade prediction was an overestimation or an underestimate as compared to the real time upgrade time,
wherein the initial upgrade prediction, Tinit, is generated according to:
T
init
=
∑
m
=
1
M
∑
n
=
1
N
[
m
]
T
[
m
]
[
n
]
,
wherein M is a number of component nodes of the multi-component product, each component node has N[m] stages, and T[m][n] is an estimated time expected to be spent on a stage n for a component node m, and
wherein the updated upgrade prediction is dynamically increased when the real time progress review indicates that one of the plurality of stages spends more time than the initial upgrade prediction.
2. The computer implemented method as in claim 1 , wherein only a subset of the plurality of component nodes of the multi-component product are being upgraded, and the initial upgrade prediction is generated based on an analysis of the subset of the plurality of component nodes.
3. The computer implemented method as in claim 1 , wherein the updated upgrade prediction, Testimate, is generated according to:
T
estimate
=
{
T
init
-
∑
m
=
1
x
∑
n
=
1
y
T
[
m
]
[
n
]
+
T
[
x
]
[
y
]
-
Δ
t
xy
,
Δ
t
xy
≤
T
[
x
]
[
y
]
T
init
-
∑
m
=
1
x
∑
n
=
1
y
T
[
m
]
[
n
]
+
Δ
t
xy
*
1.1
-
Δ
t
xy
,
T
[
x
]
[
y
]
<
Δ
t
xy
<
T
max
[
x
]
[
y
]
wherein x refers to a component node of the M component nodes that is being upgraded, y refers to a stage of the N[m] stages that is being upgraded, T[x][y] refers to an estimated time expected to be spent on the stage y for the component node x, and Δt xy is an elapsed time since beginning the stage y.
4. The computer implemented method as in claim 1 , wherein the initial upgrade prediction for the multi-component product is generated based on internal testing data or customer feedback.
5. A system for generating a dynamic upgrade prediction comprising:
a multi-component product including a centralized management node and a plurality of component nodes, wherein the centralized management node comprises a memory and one or more processors that are operatively coupled to the memory, and wherein the one or more processors are configured to communicate with the plurality of component nodes and to:
generate an initial upgrade prediction for a predicted time of an upgrade to be performed on a subset of the plurality of component nodes;
perform a real time progress review of the upgrade, wherein each of the plurality of component nodes includes a plurality of stages;
perform an upgrade duration comparison between the initial upgrade prediction and a real time upgrade time for a subset of the plurality of stages; and
generate an updated upgrade prediction based on the upgrade duration comparison between the initial upgrade prediction and the real time upgrade time, wherein generating the updated upgrade prediction varies based on whether the upgrade duration comparison indicates that the initial upgrade prediction was an overestimation or an underestimate as compared to the real time upgrade time,
wherein the centralized management node generates the initial upgrade prediction, Tinit, according to:
T
init
=
∑
m
=
1
M
∑
n
=
1
N
[
m
]
T
[
m
]
[
n
]
,
wherein M is a number of component nodes of the multi-component product, each component node has N[m] stages, and T[m][n] is an estimated time expected to be spent on a stage n for a component node m, and
wherein the updated upgrade prediction is dynamically increased when the real time progress review indicates that one of the plurality of stages spends more time than the initial upgrade prediction.
6. The system as in claim 5 , wherein only a subset of the plurality of component nodes of the multi-component product are being upgraded, and the initial upgrade prediction is generated based on an analysis of the subset of the plurality of component nodes.
7. The system as in claim 5 , wherein the updated upgrade prediction, Testimate, is generated according to:
T
estimate
=
{
T
init
-
∑
m
=
1
x
∑
n
=
1
y
T
[
m
]
[
n
]
+
T
[
x
]
[
y
]
-
Δ
t
xy
,
Δ
t
xy
≤
T
[
x
]
[
y
]
T
init
-
∑
m
=
1
x
∑
n
=
1
y
T
[
m
]
[
n
]
+
Δ
t
xy
*
1.1
-
Δ
t
xy
,
T
[
x
]
[
y
]
<
Δ
t
xy
<
T
max
[
x
]
[
y
]
wherein x refers to a component node of the M component nodes that is being upgraded, y refers to a stage of the N[m] stages that is being upgraded, T[x][y] refers to an estimated time expected to be spent on the stage y for the component node x, and Δt xy is an elapsed time since beginning the stage y.
8. The system as in claim 5 , wherein the initial upgrade prediction for the multi-component product is generated based on internal testing data or customer feedback.
9. A non-transitory computer-readable medium storing instructions which, when executed by one or more processors, cause the one or more processors to perform dynamic upgrade prediction operation, the operation comprising:
generating an initial upgrade prediction for a predicted time of an upgrade to be performed on a subset of a plurality of component nodes of a multi-component product;
performing a real time progress review, using a centralized management node, of the upgrade, wherein each of the plurality of component nodes includes a plurality of stages;
performing an upgrade duration comparison between the initial upgrade prediction and a real time upgrade time for a subset of the plurality of stages; and
generating an updated upgrade prediction for the upgrade of the multi-component product based on the upgrade duration comparison between the initial upgrade prediction and the real time upgrade time, wherein generating the updated upgrade prediction varies based on whether the upgrade duration comparison indicates that the initial upgrade prediction was an overestimation or an underestimate as compared to the real time upgrade time,
wherein the initial upgrade prediction, Tinit, is generated according to:
T
init
=
∑
m
=
1
M
∑
n
=
1
N
[
m
]
T
[
m
]
[
n
]
,
wherein M is a number of component nodes of the multi-component product, each component node has N[m] stages, and T[m][n] is an estimated time expected to be spent on a stage n for a component node m, and
wherein the updated upgrade prediction is dynamically increased when the real time progress review indicates that one of the plurality of stages spends more time than the initial upgrade prediction.
10. The non-transitory computer-readable medium as in claim 9 , wherein only a subset of the plurality of component nodes of the multi-component product are being upgraded, and the initial upgrade prediction is generated based on an analysis of the subset of the plurality of component nodes.
11. The non-transitory computer-readable medium as in claim 9 , wherein the updated upgrade prediction, Testimate, is generated according to:
T
estimate
=
{
T
init
-
∑
m
=
1
x
∑
n
=
1
y
T
[
m
]
[
n
]
+
T
[
x
]
[
y
]
-
Δ
t
xy
,
Δ
t
xy
≤
T
[
x
]
[
y
]
T
init
-
∑
m
=
1
x
∑
n
=
1
y
T
[
m
]
[
n
]
+
Δ
t
xy
*
1.1
-
Δ
t
xy
,
T
[
x
]
[
y
]
<
Δ
t
xy
<
T
max
[
x
]
[
y
]
wherein x refers to a component node of the M component nodes that is being upgraded, y refers to a stage of the N[m] stages that is being upgraded, T[x][y] refers to an estimated time expected to be spent on the stage y for the component node x, and Δt xy is an elapsed time since beginning the stage y.
12. The non-transitory computer-readable medium as in claim 9 , wherein the initial upgrade prediction for the multi-component product is generated based on internal testing data or customer feedback.
13. The computer implemented method as in claim 1 , wherein responsive to determining that the initial upgrade prediction is an overestimation as compared to the real time upgrade time, a modification to the initial upgrade prediction to generate the updated upgrade prediction is limited to an upgrade variation associated with a current stage of the plurality of stages being executed.
14. The non-transitory computer-readable medium as in claim 9 , wherein responsive to determining that the initial upgrade prediction is an overestimation as compared to the real time upgrade time, a modification to the initial upgrade prediction to generate the updated upgrade prediction is limited to an upgrade variation associated with a current stage of the plurality of stages being executed.