Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

Virtual Machine Migration

Technique in Cloud Computing

Presented by: Mita J. Bheda (170280702003) ME Sem-III, LDCE


Guided By: Prof. T. A. Raval (Comp. Dept.) LDCE
List of Topics
1. Introduction
1. Service Model
2. Deployment Model
2. VM Migration
1. Non-live Migration & Live Migration
1. Pre-Copy Approach
2. Post –Copy Approach
3. Performance Measurement Metrics
3. Comparison
4. Conclusion
5. References
1. Introduction
• Cloud computing is emerging style of computing where
application, data and resources are provided as a service to
user over web.
• Cloud computing is a service where storage and
computing resources can be accessed on subscription
basis[13].
1.1 Service Model
1. Software-as-a-Service (SaaS)
• Software as a service is software that is deployed over the
internet.
2. Platform-as-a-Service (PaaS)
• Platform as a service provide development environment as a
service.
• PAAS provide environment in which developer can create
customized solution within the context of the development
tools.
3. Infrastructure-as-a-Service (IaaS)
• Infrastructure as a service delivers a infrastructure
virtualization environment as a service.
1.2 Deployment model
1. Public cloud
• Public cloud is publically available to user over internet on
subscription basis. Example. Google app engine, Microsoft
Azure
2. Private cloud
• In the private cloud, computing resources are used and
controlled by a private enterprise. It is privately owned by
organization .
3. Hybrid cloud:
• A third type can be hybrid cloud that is typical combination of
public and private cloud.
4. Community Cloud
• It can be public or hybrid cloud shared by several organization
among specific community .
• it is used to share information and resources among
particular community such as cloud resources among
researches , government or education.
2. VM Migration
• Migration is moving one virtual machine from one physical
machine to another physical machine.
• Virtualization is also known as teleportation.
• There are two kind of Migration technique
1. Live Migration
2. Off - line Migration
2.1 Non-live(Off-line) VM Migration
• Also known as stop and copy based migration.
• Virtual machine at the source host is paused and
then transfer all states of source host to the
target or destination host and then finally resume
the working of virtual machine at the target host.
• Migration time and down time is same because
VM is not started on target host until its all pages
are sent to target.
• The major drawback of it is that it results in larger
down time.[6]
2.1 Live Migration
• Process of moving a running virtual machine or different PM
without disconnecting client or application.
• Technique where Virtual machine is migrated from one
physical machine to another while power on.
• Known as Hot, online or real time migration.
• Live migration approaches are classified into two categories

1. Pre-copy Approach
2. Post Copy Approach
2.1.1 Pre-Copy Approach
• This method copies the memory state over multiple iterations
followed by a final transfer of the memory state

Figure : Pre-copy Live migration


2.1.2 Post copy approach
• It is live migration technique where Virtual Machine
migration is performed after suspending Virtual machine at
source and transferring memory page as per requirement.

Figure : Post Copy Migration method


1. Pre-copy Virtual Machine Migration Technique

• Pre-copy VM migration is executed mainly in three phases:


(i) memory copy phase,(ii) iteration phase, and (iii) stop-and-copy phase.
• In the first phase, complete memory pages allocated to VM are transferred
from source to the target PM, while its execution at the source PM continues .
• During the transfer of memory pages in memory copy phase, some portion of
the VM memory pages gets modified. The modified memory pages are
repeatedly sent to the target PM through several iteration phases .
• When the migration iteration count reaches maximum limit, the execution of
the VM at the source PM is suspended and remaining dirtied pages and VM’s
CPU states copied to the target server in stop-and-copy phase. and then starts
its execution at the target PM.
• In this technique, complete VM memory page is transmitted instead of
modified contents of the page. This leads to increase in downtime, migration
time, and network traffic generated during the migration. So it is suitable for
low dirty pages.
2. Improving the Performance of Pre-copy
Virtual Machine Migration Technique (2018)

• In Pre-copy, complete VM memory page is transferred instead of modified


contents of the page which increase in downtime, migration time, and
network traffic generated during the migration.
• during the iterative phase of pre-copy VM migration, the VM memory
pages are transferred over the network from source to the destination
host. As the VM continues to run at the source host, VM memory pages
gets modified before its transfer to the destination host is completed.
• Proposed technique-transfers only modified contents of VM memory
page. At the source server, modified contents of VM page is computed by
applying logical operator XOR for the old and current version of VM
memory page. At the destination, the original page is reconstructed by
performing XOR operation between the received contents and page stored
in system cache.
• Result:
– minimize downtime by 69.4%, migration time by 73.5%, and reduces network
traffic generated by 74.21%.
3. OPTIMIZATION OF MIGRATION DOWNTIME OF
VIRTUAL MACHINES IN CLOUD
• The pre copy approach is not suitable for high dirty pages. This is because it would repeatedly
transmit a large number of identical dirty pages in iteration. So, it is not efficient for live
migration of VM to meet at a specific time.
• To solve this problem, a time-series based pre copy algorithms is proposed. In which, high
dirty pages are identified by their historical statistics of the TO_SEND bitmap, to avoid them
being transferred again to the target host in current iteration. In last iteration, all high dirty
pages transferred to target host.
• historical bitmap (TO_SEND_H) with size N, and Two key parameters K & N, and avoid them
being transferred again to the target host in this iteration. Until at the end of the last
iteration, will be these high dirty pages transferred to target host.
• Result:
– migration time of improved approach in high dirty rate environment increases only a
little than that in low dirty rate environment.
– down time of improved Xen is not always lower than that of the original Xen in the
environment of low dirty pages. \/
– down time of modified Xen is always lower than that of the original Xen in the
environment of high dirty pages, and the effect of decreased down time is remarkable.
Iteration of precopy-based live migration Iteration of time-series based precopy
live migration
4. Performance Measurement of Virtual Machine Migration
Using optimized Pre-copy Approach in cloud computing(2016)

• basic pre copy algorithm give better results for low-dirty


page but don’t give better result for high dirty page rate.

• optimized pre-copy algorithm can only work with high dirty


page rate environment but could not perform well for the
low dirty rate. So proposed approach further modifying
existing algorithm and then applied Characteristic Based
Compression (CBC) algorithm which reduces total migration
time and total downtime for pages with high dirty rate and
pages with low dirty rate.
Cont..
• After getting value of D and T then comparing value of T and D. If
value of D (dirty page rate) is greater than threshold value T ( high
dirty page rate), we use our optimized pre-copy approach. while D
is less than T (i.e. Low dirty page rate), we use pre-copy approach.
Here we are having two threshold values i.e. T and k.

– D(dirty page rate) = No of dirty pages / overall pages

• k is used to check that particular page is used how many times in


past history so by using technique of optimized pre-copy approach
we have decided value of k by taking different scenarios.

• Result:
– modified pre-copy algorithm which reduces migration time in both
low-dirty page rate and high-dirty page rate.
Flow chart:
5. An Improved Pre-copy Approach for Transferring the VM Data during
the Virtual Machine Migration for the Cloud Environment(2016)

• Main problem in the pre-copy approach is that it send the same


page many times if the page is modified regularly, which increase
migration time and downtime.
• To mitigate this problem a time series based live migration
approach were proposed where threshold value is used to identify
HDP. Proposed approach eliminate the requirement of threshold.
• In this method page divided in to two group: HDP and normal page.
• Page modified more than predefined threshold – it belongs to HDP
otherwise normal group. pages belongs to HDP send in last
iteration.
Cont…
• Method divided in two phase:
– Phase 1-check whether page is modified in last iteration or not using four
bitmaps, page sent if to_send & to_skip is 1 & 0 otherwise not.
– Phase-2- Instead of using any threshold, we store the history of each pages
and then check the number of times page is modified from the history data
and compare with the un-modified pages. If the page is modified more time as
compare to the unmodified pages then page is declare as a HDP otherwise
page is normal.
– Page is transferred only when value of X is less than N-X. where X –no. of
times page is modified. And N-no. of history record for each page.
• Result:
– proposed approach gives the better result. It has minimum
down time, minimum total migration time and minimum
number of page transferred.
6. A New Pre-Copy Strategy for Live Migration of Virtual
Machines

• research focus on how to reduce the total amount of transferred data and total
migration time when dirty memory pages are largely generated because of
frequent update operations.
• this novel strategy to reduce the amount of memory pages being transferred.
• The main contributions of the paper are:
– (1) A novel pre-copy strategy, which combines historical with current statistics to
judge whether a memory page is updated frequently during live migration;
– (2) author refers to the limitation on the transferring number. For a DP, if it has
been transferred many times, we can conclude that there is no need to transfer it
any more. Because a DP that has been transferred many times is more likely to be
a HDP. Transferring this page just wastes network resources. In our current
implementation, we delay the transferring of this page until the last iteration.
.Transfer Number Limitation (TNL) and Transfer Interval Limitation (TIL), which can
further reduce the duplication of data transmission and thus can reduce the total
data transferred and total migration time;
• Result:
– reduce the total migration time and total data transferred without extending the
downtime.
7. Post-Copy Based Live Virtual Machine Migration Using
Adaptive Pre-Paging and Dynamic Self-Ballooning

• Proposed method that give result in improvement in total page transferred and total
migration time by using post-copy approach with adaptive pre paging which is used to
avoid transfer of duplicate pages and to avoid the transfer of free memory pages it uses
dynamic self ballooning(DSB).this will increase speed of migration.
• One of the technique to reduce migration time is to don’t send unused memory pages
to target host.it will decrease migration time and reduce no. of iteration.

• Typical ballooning used in Xen balloons unused pages only when there is not enough
memory for new VM. In DSB, VM continues balloons the unused memory over its
execution lifetime. It saves large amount of memory.
• Post-copy has the ability to minimize network page faults, by pushing future requested
pages from the source server before they are faulted by running VM. for this active
push approach used with adaptive pre-paging.

• The results show that different migration metrics like total migration time, pages
transferred, and network overhead has improved, VM having a range of workloads. To
avoid sending of all duplicate pages, the post-copy technique is used with adaptive pre-
paging.
4. Performance measurement Metrics
1. Preparation Time :
• Time between initiating migration and transferring the VM’s processor
state to the target node.
2. Downtime:
• Time during which the migrating VM’s execution is stopped. During this
time service is interrupted for user [During the period, the CPU state and
remaining dirty pages are transferred to target host.]
3. Resume Time :
• Time between resuming the VM’s execution at the target and the end of
migration .
4.Pages Transferred :
• This is the total count of memory pages transferred, including duplicates,
across all of the above time periods.
5. Total Migration Time
• This is the sum of all the above times from start to finish.

Figure : time required for pre-copy and post-copy approach


5. Comparison
VM Migration Method Advantage Disadvantage
Non Live Stop VM at source node then - Simple Concept Downtime is more
migration transfer meory at destination - Easy to implement and noticed by
node and resume VM. user

Pre-Copy Live Transfer state of memory at - small downtime -over head due to
migration target PM and then resume - system do not crash duplicate pages
VM at destination node and because VM is running at transmission.
again transfer pages which source VM
are changed while memory -migration is not noticed by
state where transfer end user
Post copy live Virtual memory is stopped at - less network head because - downtime is
migration source and started at memory pages are more
destination after that memory transferred only once. - there is service
stage is transferred. interruption for
user
- cannot recover
from failure
6. Conclusion
• Live migration is more better than as compared to offline
migration as live migration provide seamless services to user.
• Pre-copy approach provides less downtime which may be not
noticeable by user .
• Post- copy approach has long downtime period in which
migration is noticed by user. same time pre-copy approach
increase the total migration time as compared to post copy
approach.
References
[1] Dave, Akash, Bhargesh Patel, and Gopi Bhatt. "Load balancing in cloud computing using optimization
techniques: A study." Communication and Electronics Systems (ICCES), International Conference on. IEEE,
2016.

[2] Desai, Megha R., and Hiren B. Patel. "Performance Measurement of Virtual Machine Migration Using Pre-
copy Approach in cloud computing." Proceedings of the Second International Conference on Information
and Communication Technology for Competitive Strategies. ACM, 2016.

[3] Aditya Bhardwaj and c. rama Krishna. “improving the performance of pre-copy virtual machine migration
technique" international conference on communication, computing and networking, Springer, 2018.

[4] Jain, Praveen, and Ratish Agrawal. "An Improved Pre-copy Approach for Transferring the VM Data during the
Virtual Machine Migration for the Cloud Environment." (2016).

[5] Jamshidi, Pooyan, Aakash Ahmad, and Claus Pahl. "Cloud migration research: a systematic review." IEEE
Transactions on Cloud Computing 1.2 (2013): 142-157.

[6] Singh, Gursharan, and Pooja Gupta. "A review on migration techniques and challenges in live virtual
machine migration." Reliability, Infocom Technologies and Optimization (Trends and Future
Directions)(ICRITO), 2016 5th International Conference on. IEEE, 2016.
[7] Wang, Lizhe, et al. "Cloud computing: a perspective study." New Generation Computing 28.2 (2010): 137-
146.

[8] Zhang, Shuai, et al. "Cloud computing research and development trend." Future Networks, 2010. ICFN'10.
Second International Conference on. Ieee, 2010.

[9] Yang, Jianfeng, and Zhibin Chen. "Cloud computing research and security issues." Computational
intelligence and software engineering (CiSE), 2010 international conference on. IEEE, 2010.

[10] Idrissi, H. Kamal, A. Kartit, and M. El Marram. "A taxonomy and survey of cloud computing." Security Days
(JNS3), 2013 National. IEEE, 2013.

[11] Jadeja, Yashpalsinh, and Kirit Modi. "Cloud computing-concepts, architecture and challenges." Computing,
Electronics and Electrical Technologies (ICCEET), 2012 International Conference on. IEEE, 2012.

[12] [online] https://en.wikipedia.org/wiki/Cloud_computing

[13] Leelipushpam, P. Getzi Jeba, and J. Sharmila. "Live VM migration techniques in cloud environment—a
survey." Information & Communication Technologies (ICT), 2013 IEEE Conference on. IEEE, 2013.
[14] [online] https://en.wikipedia.org/wiki/Live_migration

[15] Hines, Michael R., Umesh Deshpande, and Kartik Gopalan. "Post-copy live migration of virtual machines."
ACM SIGOPS operating systems review 43.3 (2009): 14-26.

[16] Hines, Michael R., and Kartik Gopalan. "Post-copy based live virtual machine migration using adaptive pre-
paging and dynamic self-ballooning." Proceedings of the 2009 ACM SIGPLAN/SIGOPS international
conference on Virtual execution environments. ACM, 2009.

[17] Singh, Gursharan, and Pooja Gupta. "A review on migration techniques and challenges in live virtual
machine migration." Reliability, Infocom Technologies and Optimization (Trends and Future
Directions)(ICRITO), 2016 5th International Conference on. IEEE, 2016.

[18] [online] https://www.ibm.com/blogs/cloud-computing/2015/01/choose-right-cloud-model-workload-


analysis-hierarchy/
[19] J.Arputharaj Johnson . "optimization of migration downtime of virtual machines in cloud ."
Communication and Electronics Systems (ICCES), International Conference on. IEEE, 2013.

You might also like