Xclone presentation final
Upcoming SlideShare
Loading in...5
×
 

Like this? Share it with your network

Share

Xclone presentation final

on

  • 238 views

 

Statistics

Views

Total Views
238
Views on SlideShare
238
Embed Views
0

Actions

Likes
0
Downloads
0
Comments
0

0 Embeds 0

No embeds

Accessibility

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

Xclone presentation final Presentation Transcript

  • 1. Xclone Fastest VM Provisioning Xclone
  • 2. What we do? • Environment o 6 physical hosts(CPU:12, Memory:16GB) • Requirement o Deploy 100 serengeti VMs in 30 minutes o Make it as fast as possible
  • 3. Why we do it? • VMware use it everyday • Hybrid Cloud • Distributed App • Only Automation ? • Fast
  • 4. Naive Solution Host Disk Mem Host Disk Mem Host Disk Mem Host Disk Mem Host Disk Mem Host Disk Mem VM VMVMVMVMVM
  • 5. Problem Analysis • Full Clone o Resource Usage o Time Cost: ~3 minutes o Naïve Solution:3 *(5 + 100/6) = 66 minutes Metrics Value cpu <2% memory 200M disk 40MB/s network 80-100MB/s
  • 6. Problem Analysis • Linked Clone o Time Cost: ~10 seconds o Naive Solution:  3*5 + 10*(100/6)/60 + delta = 18~22 minutes • Shortcomings o Must in one host
  • 7. Problem Analysis • Hardware Performance Test • Metrics Value Disk read 25MB/s Disk write 50MB/s Network ~100MB/s
  • 8. Problem Analysis • Key Decision o Full clone in multi-host, linked clone in inner host  linked clone much faster o Use tree clone  disk bottleneck  single srcVM lock
  • 9. Our Solution(Algorithm) VM VM VM VMVM VM
  • 10. Our solution(Architecture) UI Deploy Algorithm Information collector Infrastructure Ops VC API
  • 11. Test Result 0 10 20 30 40 50 60 70 time(minutes)
  • 12. Future work • Ramdisk • Infrastructure Aware Deployment(Disk & Network) • Full Clone Optimization
  • 13. Our code & docs Codes on: Git: https://github.com/linvictor88/Xclone Docs: Plan Edit Doc:https://docs.google.com/a/vmware.com/document/d/1eoD5X2B4hBI69eFlkV671n1QFz- Pggy14UZTyeG0YjI/edit#heading=h.84hu40wudn8o Plan pub: https://docs.google.com/document/d/1eoD5X2B4hBI69eFlkV671n1QFz-Pggy14UZTyeG0YjI/pub Status Edit Doc:https://docs.google.com/a/vmware.com/document/d/1gg_7zw9ZJwiqR_wQSnwOMqn1Zk1aTxeZgEne 0AwQUvc/edit#heading=h.vlkvaq5eerus Status pub:https://docs.google.com/a/vmware.com/document/d/1gg_7zw9ZJwiqR_wQSnwOMqn1Zk1aTxeZgEne 0AwQUvc/pub Cloning research Doc:https://docs.google.com/a/vmware.com/document/d/1rqD8pmG6So3ARA4YD9Hd6Ut0q9ROAZ7K_qG V4Xr8k4U/edit# Cloning research pub: https://docs.google.com/document/d/1rqD8pmG6So3ARA4YD9Hd6Ut0q9ROAZ7K_qGV4Xr8k4U/pub Xclone Code Design https://docs.google.com/a/vmware.com/document/d/17zMxuN4EjpyRkAMarahvlEyTjsb3hQmG- O4NMQmAk40/edit
  • 14. Learned Lessons • Teamworking • Source code management
  • 15. Thanks • Junping Du • Xinhui Li • Yang Liu • Terry Li