Some would consider it as difficult challenge
I think was the last minute changes in the final report that might save us from getting a fail grade.
But then it was fun.
So here I am .. Introducing Project CloudMine, an open source, auto-scaling (Ah! Rather Elastic) and a FUN application implemented on Open Nebula.
The challenge was to built a automated,elastic,load balancing application with a monitoring service on cloud services. We implemented on Open Nebula on DAS4 systems at TU delft.
Installation steps are pretty simple. If you have Open Nebula cloud installed at your end. Then simply download it from Github and then
export PROJECT_CLOUD=<dir-to-ProjectCloud>
export PATH=$PATH:$PROJECT_CLOUD/bin
now there are two aspect left before it automatically starts acquiring Virtual Machines
in $PROJECT_CLOUD/conf you will two configuration files - cloud_env and image_conf
In image_conf you need to paste the .one file which consists of you image details
whereas in cloud_env you need put your
user:cloudpass
RPC-XML service
POOL_SIZE
So the CloudMine reads this file for Open Nebula especially Pool size it automatically scales up or scales down based in this size.
Once you have set your configurations, you need to start the automation script.
auto user1 password
user1 and password are default user and credentials for CloudMine (yes we did implement ACL). It starts a service so in the mean while if you chance the pool size, this will pick up the change and start or stop VMs.
The Load balancing on the other hand is kept simple and is implemented in a Round Robin fashion.
If you want to submit a task.Do as follows
submit user1 password <file> <word-to-be-mined>
We have a benchmark script that runs multiple tasks and collects the output and vm stats.
There are two kinds of monitoring implemented in CloudMine. Well one evolved from another. The one which is implmented by default is as follows. The moment a VM is acquired it starts a monitoring service in that VM. We have scripts that pull these logs and display.
Overall it is a fun application. We have identified we shortcoming and will improve it (i.e if we get an opportunity).
Keep following us for more exciting stuff.
In image_conf you need to paste the .one file which consists of you image details
whereas in cloud_env you need put your
user:cloudpass
RPC-XML service
POOL_SIZE
So the CloudMine reads this file for Open Nebula especially Pool size it automatically scales up or scales down based in this size.
Once you have set your configurations, you need to start the automation script.
auto user1 password
user1 and password are default user and credentials for CloudMine (yes we did implement ACL). It starts a service so in the mean while if you chance the pool size, this will pick up the change and start or stop VMs.
The Load balancing on the other hand is kept simple and is implemented in a Round Robin fashion.
If you want to submit a task.Do as follows
submit user1 password <file> <word-to-be-mined>
We have a benchmark script that runs multiple tasks and collects the output and vm stats.
There are two kinds of monitoring implemented in CloudMine. Well one evolved from another. The one which is implmented by default is as follows. The moment a VM is acquired it starts a monitoring service in that VM. We have scripts that pull these logs and display.
Overall it is a fun application. We have identified we shortcoming and will improve it (i.e if we get an opportunity).
Keep following us for more exciting stuff.
No comments:
Post a Comment