Cloud Data Center/Storage Power Efficiency Solutions Junyao Zhang 1.

Cloud Data Center/Storage Power Efficiency Solutions

Junyao Zhang

1

Current Storage Energy Efficiency Solutions: Tradeoff Energy/Performance

Multi-speed Disk: DRPM CPU has Dynamic Voltage and Frequency Scaling

(DVFS): can we use this idea on storage systems?

2

VM Consolidation: SRCMap

Request Consolidation (Replication-based): EERAID, Diverted Access Redirect requests to some replicas to spin down

the others

Data Consolidation: MAID, PDC Skew data into partial disks/cache disks so

that others can be shut down

In what degree should we tradeoff energy/performance?

Power Proportional

A standard metric proposed by Google [1]: Computer components should consume energy in

proportion to the system utilization. Observation:

3

Robust and Flexible Power-Proportional Storage

Strictly satisfy power proportional :

6

7

Solution

9

Fine-grained power proportionality for one data-set

More

11

Read Performance

12

Write Performance

13

Handling Recovery

Bounded wake-up

Rebuild is power-proportional

14

Near power-proportional(cnt.)

15

Multi-data set: Fair Scheduling

16

Degradation

17

Sierra: Practical Power-proportionalilty for Data Center Storage

Power proportional layout with the concern of the following factors: Fault-tolerance, Loading balance, Consistency, Good

performance.

Three challenges: Layout that allows significant power savings Maintain read and write availability at the original

levels Predict the number of servers required at anytime

18

C1: Power-aware Layout

19

Gear 1 (g=1): need 2 nodes (Gear group 0) to keep 1 the copy of all nodes

Gear 2 (g=2): 4 nodes

C1: Power-aware Layout

Extending to three replicas and more: two options Rack-aligned Rotated

20

C2: Distributed virtual log (DVL)

Aim: maintain read/write availability Write: if secondaries not available, entering

“logging mode”(write primary replicas to DVL and replicate DVL r-1 times )

21

C3: Gear Scheduler

Aim: predict system load and schedules servers to power down or up accordingly. Observation: predict hourly behavior based on

historical records of this hour.

22

Power Savings

23

Performance

24

Response Time

Conclusion

Power proportional is becoming an important metric for power/energy tradeoff

Rabbit proposed a idea-power proportional layout

Sierra considered factors such as: power, reliability, load balancing, consistency and etc.

25

[1] L. A. Barroso and U. H¨olzle. The case for energy-proportional computing. Computer, 40(12):33–37, 2007.

[2] E. Thereska, A. Donnelly, and D. Narayanan. Sierra: a powerproportional, distributed storage system. MSR-TR-2009-153, November 2009.

[3] H. Amur, J. Cipar, V. Gupta, G. R. Ganger, M. A. Kozuch, and K. Schwan. Robust and flexible power-proportional storage. In SoCC, 2010.

32

Thank you

33