PB-LRU: A Self-Tuning Power Aware Storage Cache Replacement Algorithm for Conserving Disk Energy

1
PB-LRU A Self-Tuning Power Aware Storage Cache
Replacement Algorithm for Conserving Disk Energy

• Qingbo Zhu, Asim Shankar and Yuanyuan Zhou

Presented Hang Zhao Chiu Tan
2
PB-LRU Partition-Based LRU

• Storage is a major energy consumer, 27 of power
  budget in a data center.
• PB-LRU is a power aware, on-line cache
  management algorithm.
• PB-LRU dynamically partitions cache at run time
  for energy optimal cache size per disk.
• Practical algorithm that dynamically adapts to
  workload changes with little tuning.

3
Outline

• Motivation
• Background
• Why need PB-LRU?
• Main Idea
• Energy estimation at Run Time
• Solving MCKP
• Evaluation Simulation
• Conclusion

4
Motivation

• Why is power conservation important?
• Data centers are an important component of the
  Internet infrastructure.
• Power needs for a data center are increasing at
  25 a year, with storage taking up 27.
• How to reduce power in storage?
• Simple. Spin down disk when not in use.

5
Motivation (II)

• But
• Performance and energy penalty when disk moving
  from low to high mode.
• Data center volume is high. Idle periods small.
  Makes spinning up and down impractical.
• Solution Multi-speed disk architecture.
• PB-LRU targets multi-speed disk.

6
Background

• Break-even time Minimum length of idle time
  needed justify spinning up/down.
• Oracle DPM Knows length of next idle period.
  Uses this to regulate power modes.
• Practical DPM Use thresholds to regulate
  powering up or down.

7
Why need PB-LRU?

• Earlier work PA-LRU.
• Idea Keep blocks from less active disks in
  cache. Thus extends idle period.
• Cost More misses to active disks.
• Justification Since active disks are already
  spinning, cheaper in terms of power consumption.

8
However

• PA-LRU requires complicated parameter tuning. 4
  parameters needed.
• No intuition between parameters and disk power
  consumption or IO times.
• Thus difficult to adopt simple extensions or
  heuristics for real world implementation.
• PB-LRU is a practical implementation !

9
PB-LRU Main Idea

• Divide cache into partitions, one for each disk.
• Each partitioned managed individually.
• Resize partitions periodically.
• Workloads are not equally distributed through
  different disks.

10
Main Idea (II)

• So what do we need?
• Estimate, for each disk, the energy consumed for
  a particular cache size. (estimation problem)
• Use these estimates to find partitioning that
  minimize total energy consumption for all disks.
  (MCKP problem)

11
Estimation Problem

• Q How to estimate energy consumption per disk
  for different cache sizes at run time?
• Use simulators. One (multi-disk) simulator for
  every cache size.
• Requires (NumCacheSizes X NumDisks) simulators.
  Impractical!

12
Estimation Problem (II)

• Mattsons Stack
• Take advantage of inclusion property.
• A cache of k blocks is a subset of k1 blocks.
  Accessing a stack at position i means a miss at
  caches smaller than size i.
• PB-LRU uses Mattsons Stack to predict hit or a
  miss for different partition sizes.

13
Estimation Problem (III)

• In addition, PB-LRU keeps track of previous
  access time and previous energy consumption.
• With these pieces of information, energy
  consumption of various cache is estimated.

14
Cache Accesses
Time T1 T2 T3 T4 T5
Access 5 4 3 2 1
5 possible Cache sizes
Stack
1
2
3
4
5
Mattson Stack
Cache Size Pre_miss Energy
1 T5 E5
2 T5 E5
3 T5 E5
4 T5 E5
5 T5 E5
RCache
1
2
3
Before
Existing Cache (real)
15
Time T1 T2 T3 T4 T5 T6
Access 5 4 3 2 1 4
LRU
4th element of stack. Miss for cache size lt 4
Stack
4 (1)
1 (2)
2 (3)
3 (4)
5 (5)
T6 Access Block 4
Cache Size Pre_miss Energy
1 T6 E6
2 T6 E6
3 T6 E6
4 T5 E5
5 T5 E5
RCache
4
2
3
E6 E5 E(T6-T5) 10ms ActivePower
LRU
16
Solving MCKP

• MCKP is NP-hard. But modified problem solvable
  using dynamic programming.
• General result Increase cache size for less
  active disks, decrease cache size for active
  disks.
• Why? Penalty for reducing cache size of an active
  disk is small, while the energy saved for
  increasing cache size for inactive disk is large

17
Evaluation Methodology

• The integrated simulator
• Disk power model
• CacheSim
• DiskSim
• Multi-speed disks model
• Similar to IBM Ultrastar 36Z15
• Add 4 lower-speed modes 12k, 9k, 6k and 3k RPM
• Power model 2-competitive thresholds

18
Evaluation Methodology cont.

• The traces
• Real system traces
• OLTP database storage system, (21 disks, 128MB
  cache)
• Cello96 Cello file server from HP, (19 disks,
  32MB cache)
• Synthetic traces
• generated based on storage system workloads
• zipf distribution to distribute requests among 24
  disks and blocks in each disk
• hill shape to reflect temporal locality
• Inter-request arrival distribution exponential,
  Pareto

19
Simulation results
Limited save due to high cold misses rate 64

• Algorithms
• Infinite cache
• LRU
• PA-LRU
• PB-LRU

PB-LRU saves 9
Outperform LRU 22
20
Simulation results cont.
PB-LRU has 5 better response time
saves 40 response time

• Oracle DPM does not slow down the average
  response time for it always spin disk in time for
  a request
• All PB-LRU results are insensitive to the epoch
  length

21
Accuracy of Energy Estimation

• OLTP, 21 disks with Practical DPM
• Largest deviation of estimated energy from real
  energy is 1.8

22
Cache partition sizes
11-12MB
1MB

• MCKP partition tendency
• gives small sizes to disks which remain active
• increase the sizes assigned to relatively
  inactive disks

23
Effects of spin-up cost
Disks stay longer at low-power mode
Break-even time increases
24
Sensitivity Analysis on Epoch Length

• The epoch length just needs to be large enough to
  accommodate the warm-up period after
  re-partitioning.

25
Conclusion

• PB-LRU online storage cache replacement
  algorithm partitioning the total system cache
  amongst individual disks
• It focuses on multiple disks with data center
  workloads
• Achieving similar or better energy saving and
  response time improvement with significant less
  parameter tuning

26
Future work

• Taking pre-fetching into consideration to
  investigate the role of cache management in
  energy conservation
• Optimally divide the total cache between the
  cache and pre-fetching buffers
• Implement the disk power modeling component into
  the real storage system

27
Impact of PB-LRU

• 5 citations found at Google Scholar
• Energy conservation techniques for disk
  array-based servers (ICS04)
• Performance Directed Energy Management for Main
  Memory and Disks (ASPLOS04)
• Power Aware Storage Cache Management
• Power and Energy Management for Server Systems
• Management Issues