Real-Time Cloud Computing Chenyang Lu Cyber-Physical Systems - - PowerPoint PPT Presentation
Real-Time Cloud Computing Chenyang Lu Cyber-Physical Systems Laboratory http://www.cse.wustl.edu/~lu/ Internet of Things Convergence of Miniaturized hardware: integrate processor, sensors and radios. Low-power wireless: connect millions
Kollef and T.C. Bailey, Experiences with an End-To- End Wireless Clinical Monitoring System, Conference
Kollef and T.C. Bailey, Experiences with an End-To- End Wireless Clinical Monitoring System, Conference
q Smart manufacturing, transportation, infrastructure… q Closed-loop control à real-time performance q Computing at scale à cloud
q Collect data from roadside sensors and cameras q Centralized data analysis q Control city-wise traffic signals intelligently q SCATS @ Sydney [1]: controlling 3,400 signals at 1s latency
11/7/19 5 [1] http://en.wikipedia.org/wiki/Traffic_light_control_and_coordination
6
Cyber-Physical Boundary
q Latency q Delivery guarantees q Event time consistency
11/7/19 7
e.g., Emergency response e.g., Real-time monitoring
e.g., Predictive maintenance
q Infotainment on Linux or Android q Safety-critical control on AUTOSAR
11/7/19 8 Source: http://www.edn.com/design/automotive/4399434/Multicore-and-virtualization-in-automotive-environments
q Xen: credit scheduler, [credit, cap] q VMware ESXi: [reservation, share, limitation] q Microsoft Hyper-V: [reserve, weight, limit]
q Amazon, Google, Microsoft cloud services: #VCPUs
9
q Latency guarantees for tasks running in virtual machines (VMs). q Real-time performance isolation between
q Resource sharing between real-time and non-real-time
q RT
q VATC à real-time network I/O on a virtualized host. q RT
10
VATC: RT Network I/O RT
Cyber-Physical Event Processing RT Cilk Plus
q Domain-0: drivers, tool stack to control
q Guest Domain: para-virtualized or fully virtualized OS.
q Xen schedules
q Guest OS schedules threads on
q Xen credit scheduler: round-robin with proportional share.
11
PCPUs
OS Sched
Xen Scheduler
OS Sched OS Sched
VCPU Real-Time Task
12
RT-Xen
Scheduling in Xen, ACM International Conference on Embedded Software (EMSOFT'14), October 2014.
q A set of
q Hides task-specific information q Computed based on compositional scheduling analysis
13
Resource Interface Resource Interface Resource Interface
Workload Workload
Scheduler Scheduler Scheduler
q Allow
VCPU migration across cores
q Work conserving – utilize any available cores q Migration overhead and cache penalty
q Assign and bind
VCPUs to cores
q Cores may idle when others have work pending q No migration overhead or cache penalty
q Periodic server vs. deferrable server
14
4 8 12 16 t t
Task Schedule
Budget 3 4 8 12 16
VCPU Schedule Budget = 3 Period = 4
15
[1] Xi, Sisu, et al. "Real-time multi-core virtual machine scheduling in xen." 2014 International Conference on Embedded Software (EMSOFT). IEEE, 2014. [2] Kim, Hyoseung, Shige Wang, and Ragunathan Rajkumar. "vMPCP: A synchronization framework for multi-core virtual machines." 2014 IEEE Real-Time Systems Symposium. IEEE, 2014
4 8 12 16 t t
Task Schedule
Budget 3 4 8 12 16
VCPU Schedule Budget = 3 Period = 4 (0,4)
4 8 12 16 t t
Task Schedule
Budget 3 4 8 12 16
VCPU Schedule Budget = 3 Period = 4 (0,4)
4 8 12 16 t t
Task Schedule
Budget 3 4 8 12 16
VCPU Schedule Budget = 3 Period = 4 (0,4) (7,4) (13,4)
q Budget replenishes at the start of each period q The server consumes budget when executing jobs q When the budget exhausted, the server stops executing jobs
16
q CPU and network contention
Xen Hypervisor
Network
Components
Non Real-Time App
CPU Memory Storage
Real-Time App
NIC
17
q Rate-limit and shape flows q Prioritization or fair packet scheduling
q between transmissions q between transmission and reception
q Process packets in prioritized kernel threads q Dedicated packet queues per priority
NIC
Queueing Discipline
Real- Time App
Dom1
Non- Real- Time App
Dom2
18
Virtualized Hosts Based on Xen, IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS'15), April 2015.
19 10 16 32 64 128 256 512 1024 Dyn Cons Dyn 0.5 1 1.5 2 2.5 3 3.5 4 Round−trip Latency ( ms) Interrupt Interval (µs) Prio, Dom0−3.18 FQ_CoDel, Dom0−3.18 VATC
20
q Number of
q Not real-time
q Filtering (admission control)
q Ranking (VM allocation)
21
Manager
Host Host Host VM VM VM VM VM
q Support RT
q Real-time-aware
q Co-locate non-real-time
q Non-real-time
22
Time Cloud Computing, IEEE International Conference on Cloud Computing (CLOUD'15), June 2015.
q Accept real-time
q Consider only accepted real-time
q Balance CPU utilization q Consider only accepted real-time
23
Resource Interface Admission Control VM Allocation Real-Time VMs {<period, budget>} Schedulability + Memory CPU Utilization Non-Real-Time VMs Best Effort Memory Memory
24
13% 36% 31% 61% 37% 75% 30% 29% 47% 32% 73%
25
0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%
q Latency guarantees. q Real-time performance isolation. q Resource sharing between real-time and non-real-time workloads.
q RT
q VATC à real-time network I/O on a virtualized host. q RT
26
VATC: RT Network I/O RT
Cyber-Physical Event Processing RT Cilk Plus
q Latency guarantees. q Real-time performance isolation. q Resource sharing between real-time and non-real-time workloads.
q RT
q VATC à real-time network I/O on a virtualized host. q RT
27
VATC: RT Network I/O RT
Cyber-Physical Event Processing RT Cilk Plus
28