LIQUID-A Scalable Deduplication File System For Virtual Machine Images

CONTENTS
 INTRODUCTION
 ADVANTAGE OF CLOUD COMPUTING
 VIRTUAL MACHINE
 ADVANTAGES OF VIRTUAL MACHINE
 DISADVANTAGES OF VIRTUAL MACHINE
 DEDUPLICATION
 BENEFITSOFDEDUPLICATION
 EXISTING SYSTEM
 ISSUES IN VM STORAGE
 LIQUID SYSTEM ARCHITECTURE
 DEDUPLICATION IN LIQUID

CONTENTS(cont)
 OPTIMIZATION ON FINGER PRINT CALCULATION
 FILE SYSTEM LAYOUT
 COMMUNICATION AMONG COMPONENTS
 HEART BEAT PROTOCOL
 FAST CLONING FOR VM IMAGE
 FAULT TOLERANCE
 GARBAGE COLLECTIONS
 ADVANTAGES OF LIQUID
 CONCLUSION
 REFERENCES

INTRODUCTION
Cloud computing means storing and accessing
data programs over internet instead of yours
computers hard drive.
Figure 1: A Sample Cloud Computing Network[1]

ADVANTAGE CLOUD COMPUTING
Lower computer cost.
Improved performance.
Reduced software cost.
Instant software cost.
Unlimited storage capacity.
Increased data reliability.
Device independence and the “always on!,anywhere
and any place”.
Free from maintenance and the “no-need-to-know”.

VIRTUAL MACHINE
A virtual machine is a software that creates a
virtualized environment between the computer
platform and the end user in which the end user can
operate software.
Vitualization deals with extending or replacing an
existing interface so as to mimic the behavior of
another system.
Crucial component in cloud computing.

VIRTUAL MACHINE(cont)
Virtual machine – Hypothetical computer.
Execute programs like a physical machine.
Initial state of a virtual machine is stored in a file
called virtual machine image.

VIRTUAL MACHINE(cont)
Figure 2 :Virtual Machine Representation[1]

ADVANTAGES OF VIRTUAL MACHINE
Familiar interfaces
Isolation
-Each OS run seperately with its own virtual
resources.
High Availability
-If one VM server is failed then easily access data
from another one.
Scalability
-Add Or remove resources easily.

ADVANTAGES OF VIRTUAL
MACHINE(cont)
Back up with fast recovery
-Using VMDK data recovery tool.
Reduction of cost
-it save cost by running multi OS on single
machine.
-sharing of hardware.

DISADVANTAGES OF VIRTUAL
MACHINE
Difficulty in direct access to hardware.
Great RAM consumption since each virtual machine
will occupy a separate area of the same.
Greate use of disk space , since it takes all the files for
each operating system installed on each virtual
machine.
A virtual machine is less efficient than an actual
machine when it access the host hard drive indirectly.

DEDUPLICATION
Data Deduplication – data compression technology.
Eliminate duplicate copies of repeating data.
A redundant data block is replaced instead of storing
multiple times.
Improves storage utilization.

DEDUPLICATION(cont)
Figure 3: Deduplicated File System[2]

DEDUPLICATION(cont)
14
Traditional
File system
copycopy
Deduplication
copy
Idea. Eliminate
storage of data with
same content No new
blocks!
User’s point of
view

BENEFITS OF DEDUPLICATION
 Lower storage space requirements.
 Minimize the additional storage cost.
Performance increased.
Increase Network efficiency.
 Efficient Volume replication.

EXISTING SYSTEM
Hypervisors such as xen ,KVM etc.
Network Attached Storage(NAS)
Storage Area Network(SAN)
Direct Attached storage(DAS)

ISSUES IN VM STORAGE
High demand on VM storage remains a challenging
problem.
Existing systems have made efforts to reduce storage
consuptions.
Uses SAN cluster.
Cannot satisfy increasing demand due to cost
limitatations.
Hence we propose LIQUID.

LIQUID SYSTEM ARCHITECTURE
Three compononts – single meta server with hot back
up , multiple data server and multiple clients.
Runs on user – level service process.
VM images are split into fixed size data blocks.
Meta server – namespace , finger print , reference
count.
Meta server – mirrored to hot back up shadow meta
server.

LIQUID SYSTEM ARCHITECTURE(cont)
Data servers – charge of managing data blocks in VM
images.
Organized in a distributed hash table.
A liquid client provides a POSIX compatible file
system.
Client – critical component (provides deduplication).
Fault tolerance – mirroring the meta server.
Replicas of data blocks are stored.

LIQUID SYSTEM ARCHITECTURE(cont)
Figure 4: Liquid Architecture[3]

DEDUPLICATION IN LIQUID
Liquid choose fixed size chunking instead of variable
size chunking.
Better since all files stored in vm images will be aligned
on disk block boundaries.
Advantage – simplicity.
Block size choice.
Block size – balancing factor which is hard to choose.
Great impact on both deduplication and IO
performance.

DEDUPLICATION IN LIQUID(cont)
Smaller block size – more random seeks when
accessing a VM image.
Not tolerable.
A block size also not preferable , it will reduce
deduplication ratio.
Liquid choose different block size under different
situation.
Advised to use a multiplication of 4 kb between 256 kb
and 1 MB to achieve good balance between IO
performance and deduplication ratio.

OPTIMIZATION ON FINGER PRINT
CALCULATION
Rely on comparison of data block finger prints for
redundancy.
Finger print – collision resistant hash value calculated
from data block contents.
MD5(26) and SHA-1[12] are frequently used for this
purpose.
Finger print collision – very small , orders of
magnitude smaller than hardware error rates.

CALCULATION(cont)
So we could safely assume that two data blocks are
identical.
Finger print calculation – expensive.
Delays finger print calculation for recently modified
data blocks.
Runs deduplication lazily only when it s necessary.
Client side maintains a shared cache which contains
recently accessed data blocks.

CALCULATION(cont)
A portion of memory is used by the client side of
liquid as private cache.
Private cache hold – modified data blocks and delay
finger print calculation on them.
Modified data block ejected from shared cache and
added to private cache.
Modified data will be ejected if private cache becomes
full.

CALCULATION(cont)
And ejected based on LRU policy.
Only then will the modified data block’s finger print
be calculated.
Liquid uses multiple threads for finger print
calculation.
Multiple threads will process different data blocks
currently.
Provides good IO performance.

FILE SYSTEM LAYOUT
All file system meta data are stored on the meta server.
Organized in a file system tree.
Client side could cache portions of file systems meta
data for fast accesses.
When a VM is stopped , modified meta data and data
blocks.
Will be pushed back to meta server.
Data servers ensures modification on VM image is
visible to other client nodes.

FILE SYSTEM LAYOUT(cont)
Figure 5: Process Of Look-up by Fingerprint[4]

COMMUNICATION AMONG
COMPONENTS
 Heartbeatprotocol
 P2P Data Block Sharing
 On-Demand Data Block Fetching

HEART BEAT PROTOCOL
META SERVER – manages all data servers.
Exchange regular heart beat message with each data
server in a ROUND ROBIN FASHION.
Detect failed data servers when there are many data
servers.
To speed up failure detection data servers send an
error signal to meta server.

FAST CLONING FOR VM IMAGE
Copying large images may be time consuming.
Liquid provide efficient solution by means of fast
cloning.
VM images represented by meta data files having
reference to data blocks.
By copying meta data file and updating reference
count a clone VM image is achieved.
Modification on cloned images will not effect the
original image.

FAULT TOLERANCE
Data replication
Data migration
Hot backup of meta server

GARBAGE COLLECTIONS
Removes unused garbage data blocks when running
out of space.
Reference counting of all data blocks are maintained
by meta servers.
Garbage collection request is issued periodically to
data server.
Garbage collection is executed based on the data block
membership in the bloom filter.

ADVANTAGES OF LIQUID
Fast virtual machine deployment with peer to peer
data transfer.
Low storage consumption by means of deduplication.
Instant cloning for virtual machine images.
On demand fetching through a network caching with
local disks.
LIQUID files has no specific limit.

CONCLUSION
Presented LIQUID which is a deduplication file system
with good IO performance.
Achieve by caching frequently accessed data blocks in
memory cache.
Avoids additional disk operations.
Deduplication of VM images proved to be effective.

REFERENCES
[1] www.slideshare.net
[2]www.IJTRD1303.pdf
[3] www.cs.newpaltz.edu/~lik/publications/Xun-
Zhao-IEEE-TPDS-2014.pd.
[4]www.internationaljournalssrg.org/IJCSE/2015/Speci
al.../IJCSE-ICETSH-P118.pdf

LIQUID-A Scalable Deduplication File System For Virtual Machine Images

More Related Content

What's hot (20)

Similar to LIQUID-A Scalable Deduplication File System For Virtual Machine Images (20)

Recently uploaded (20)

LIQUID-A Scalable Deduplication File System For Virtual Machine Images