ScalableDigitalHealth (SDH): An IoT-Based Scalable Framework for Remote Patient Monitoring

. 2024 Feb 19;24(4):1346. doi: 10.3390/s24041346

Algorithm 1 Latency-based Horizontal Pod Autoscaling with Custom Metrics API

Require:

L_{target}

: Target latency value

1:
ActivePods: Set of currently active pods
2:
MinPods: Minimum number of pods
3:
ScalingThreshold: Threshold for scaling (e.g., percentage increase in latency)
4:
CustomMetrics: Custom Metrics API for collecting and measuring latency values
5:
6:
Initialize Custom Metrics API: CustomMetrics.initialize()
7:
MaxPods = None
8:
while True:
9:
$L = {}$
10:
for pod in ActivePods:
11:
$L_{i} = getThresholdLatencyValue (p o d)$
12:
$L . append (L_{i})$
13:
end for
14:
$t o t a l L a t e n c y = \sum (L)$
15:
$a v e r a g e L a t e n c y = \frac{t o t a l L a t e n c y}{ActivePods . length}$
16:
if MaxPods is None:
17:
MaxPods = determineMaxPods()
18:
$l a t e n c y C h a n g e = \frac{(a v e r a g e L a t e n c y - L_{t a r g e t})}{L_{t a r g e t}} \times 100$
19:
call AggregateLatenciesAcrossNodes()
20:
call CalculateSeparateLatencyThresholdsForEachNodeOrPod()
21:
call TakeIntoAccountDifferencesInCapabilitiesAcrossNodes()
22:
if $l a t e n c y C h a n g e > S c a l i n g T h r e s h o l d$ and len(ActivePods) < MaxPods:
23:
$P = ⌈ ActivePods . length \times \frac{l a t e n c y C h a n g e}{S c a l i n g T h r e s h o l d} ⌉$
24:
$P = min (P, M a x P o d s - ActivePods . length)$
25:
elif $l a t e n c y C h a n g e < - S c a l i n g T h r e s h o l d$ and len(ActivePods) > MinPods:
26:
$P = ⌈ ActivePods . length \times \frac{- l a t e n c y C h a n g e}{S c a l i n g T h r e s h o l d} ⌉$
27:
$P = min (P, ActivePods . length - M i n P o d s)$
28:
else:
29:
$P = 0$
30:
end if
31:
if $P > 0$ :
32:
call ScaleUp(P)
33:
elif $P < 0$ :
34:
call ScaleDown( $- P$ )
35:
end if
36:
call wait( $τ$ )