higher sla satisfaction in datacenters with continuous vm placement constraints
TRANSCRIPT
![Page 1: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/1.jpg)
Higher SLA Satisfaction in Datacenters with Continuous Placement Constraints
Huynh Tu Dang [email protected]
Fabien Hermenier [email protected]
![Page 2: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/2.jpg)
SLA for a virtualised application
![Page 3: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/3.jpg)
SLA for a virtualised application
spread the replicas
![Page 4: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/4.jpg)
SLA for a virtualised application
performance guarantee
![Page 5: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/5.jpg)
SLA for a virtualised application
low latency
![Page 6: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/6.jpg)
N3
N2
time
SLA: spread(VM1, VM2)
VM1
VM2
reconfiguration algorithm
N1
![Page 7: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/7.jpg)
N3
N2
time
SLA: spread(VM1, VM2)sys-admin query: offline(N1)
VM2VM1
VM2
reconfiguration algorithm
VM1
N1
Rec
onfig
urat
ion
proc
ess
![Page 8: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/8.jpg)
N3
N2
time
SLA: spread(VM1, VM2)sys-admin query: offline(N1)
VM2VM1
VM2
reconfiguration algorithm
VM1
N1
with discrete restrictions
![Page 9: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/9.jpg)
Discrete restriction is not enough
not an unpredictable situation,an algorithmic issue
![Page 10: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/10.jpg)
Evaluating the reliability of discrete placement constraints
• simulate a 256-server datacenter
• running 350 HA webapp (5,200 VMs)
• BtrPlace as the reconfiguration algorithm
• 4 reconfiguration scenarios that mimic industrial use case
• 100 instances per scenario
![Page 11: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/11.jpg)
Studied constraints
among
singleResourceCapacity
maxOnline
splitAmong
spread
DBs on a same edge-switch for a fast synchronisation.
keep resource for hypervisor management operations
webapp split over 2 clusters for disaster recovery
240 nodes online at maximum to fit licensing policy
replicas on distinct servers for fault tolerance
![Page 12: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/12.jpg)
verti
cal e
lastic
ity
Tiers 1
Tiers 2
Tiers 3
scenario
![Page 13: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/13.jpg)
verti
cal e
lastic
ity
Tiers 1
Tiers 2
Tiers 3
scenario
![Page 14: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/14.jpg)
scenariohorizontal elasticity
Tiers 1
Tiers 2
Tiers 3
![Page 15: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/15.jpg)
scenariohorizontal elasticity
Tiers 1
Tiers 2
Tiers 3
![Page 16: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/16.jpg)
scenarioboot storm x 400
![Page 17: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/17.jpg)
scenarioserver failure
![Page 18: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/18.jpg)
Scenario Violated SLAs
ActionsVM Boot Migrate Node Boot Node Shutdown
Vertical Elasticity 40.72 0% 99.99% 0.005% 0.005%
Horizontal Elasticity 0.19 99.82% 0.18% 2.82% 0%
Server Failure 29.56 61.29% 35.89% 2.82% 0%
Boot Storm 0.35 98.57% 1.43% 0% 0%
Migrations lead to unanticipated placements
![Page 19: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/19.jpg)
0
25
50
75
100
VerticalElasticity
HorizontalElasticity
ServerFailure
BootStorm
Violations
spread among splitAmong maxOnline
performance lossspof
failure
Migrations tend to violate relative placement constraints
![Page 20: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/20.jpg)
spread(VM[1,2])
Trading unreliable discrete constraints …
we addressed an assignment problem
![Page 21: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/21.jpg)
spread(VM[1,2])
… for safe continuous constraints
we must address a scheduling problem
![Page 22: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/22.jpg)
Continuous placement constraints with
![Page 23: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/23.jpg)
from discrete to continuous among|simpleAmong
N1N2 N3
N4 N5N6
stay on a same partition by the end of the reconfiguration process
![Page 24: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/24.jpg)
from discrete to continuous among|simpleAmong
N1N2 N3
N4 N5N6
stay on a same partition by the end of the reconfiguration process
![Page 25: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/25.jpg)
from discrete to continuous among|simpleAmong
N1N2 N3
N4 N5N6
stay on a same partition by the end of the reconfiguration process
![Page 26: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/26.jpg)
Disallow movements between partitions • basic knowledge of a reconfiguration process • still an assignment problem
N1N2 N3
N4 N5N6
from discrete to continuous among|simpleAmong
![Page 27: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/27.jpg)
Disallow movements between partitions • basic knowledge of a reconfiguration process • still an assignment problem
N1N2 N3
N4 N5N6
from discrete to continuous among|simpleAmong
![Page 28: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/28.jpg)
allDi↵erent(dhost
1 , d
host
2 )
discrete spread(VM[1,2]) ::= continuous spread(VM[1,2]) ::=
allDi↵erent(dhost
1 , d
host
2 ) ^d
host
1 = c
host
2 =) a
start
1 � a
end
2 ^d
host
2 = c
host
1 =) a
start
2 � a
end
1
continuous spread
Disallow temporary overlapping • require to know this may happen • scheduling 101
![Page 29: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/29.jpg)
continuous maxOnlinediscrete maxOnline(N[1..10], 7)::=
continuous maxOnline(N[1..10], 7)::=
10X
i=1
nqi 7
8i 2 [1, 10], n
on
i
=
⇢0 if n
q
i
= 1
a
start
i
otherwise
n
off
i
=
⇢max (T ) if n
q
i
= 0
a
end
i
otherwise
8t 2 T, card({i|non
i
� t ^ n
off
i
}) 7
scheduling 201
detailed knowledge of a reconfiguration process
harder to imagine, model & implement
![Page 30: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/30.jpg)
5 10 20 50 100Duration (sec.)
Solve
d in
stan
ces
discretecontinuous
020
4060
8010
0
10 20 50 100Duration (sec.)
Solve
d in
stan
ces
discretecontinuous
020
4060
8010
0
5 10 20 50 100Duration (sec.)
Solve
d in
stan
ces
discretecontinuous
020
4060
8010
0
boot storm
horitzontal elasticity server failure
Performance overhead
5 10 20 50 100 200Duration (sec.)
Solve
d in
stan
ces
discretecontinuous
020
4060
8010
0
vertical elasticity
![Page 31: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/31.jpg)
5 10 20 50 100Duration (sec.)
Solve
d in
stan
ces
discretecontinuous
020
4060
8010
0
10 20 50 100Duration (sec.)
Solve
d in
stan
ces
discretecontinuous
020
4060
8010
0
5 10 20 50 100Duration (sec.)
Solve
d in
stan
ces
discretecontinuous
020
4060
8010
0
boot storm
horitzontal elasticity server failure
Performance overhead
5 10 20 50 100 200Duration (sec.)
Solve
d in
stan
ces
discretecontinuous
020
4060
8010
0
vertical elasticity
![Page 32: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/32.jpg)
• discrete restriction is not enough
• continuous restriction is a solution
• a different view on the problem
• challenging, but still possible to implement
Conclusions
![Page 33: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/33.jpg)
Future Work
• a broader range of constraints and objectives
• reducing performance overhead
• static analysis to detect un-necessary continuous constraints
• controlled relaxation to handle hard situations
![Page 34: Higher SLA Satisfaction in Datacenters with Continuous VM Placement Constraints](https://reader034.vdocuments.us/reader034/viewer/2022042818/55b54384bb61eb4b3a8b46e0/html5/thumbnails/34.jpg)
open source, 20+ placement constraints, demo, tutorials, everything for reproducibility
http://btrp.inria.fr