Support replicate /batch resources

[xi-yang]

We need a way to model a group of resources with exact same properties. For example, in AWS we can create a number of EC2 instances in a batch.

This should not change the existing relationships. We should also use a common notion instead of creating multiple batch relationships.

[xi-yang]

Solution:

1. nml:hasService <BatchServiceURI>
2. <BatchServiceURI>
   mrs:providesBatch
   <BatchInstanceURI> .
3. <ResourceURI> .
   mrs:hasBatch
   <BatchInstanceURI>
4. <BatchInstanceURI>
   mrs:batchRule <batchRule1> .

<batchRule1>
mrs:type "number" ;
mrs:value "10" .

or

<batchRule2>
mrs:type "name:regex" ; // or "name:list"
mrs:value "regex or comma-separated-list" .

[xi-yang]

Alternatively, do not use BatchService but simply let Resource mrs:hasBatch.
Then the batch has the following properties:
mrs:batch_number "10" ;
mrs:batch_rule "name:regex:^abc[0-9]$" . // or "name:list:abc0,abc1,abc2".

[muzcategui1106]

I have added a new class called BatchResourcesTool and implemented the following method addBatchToModel which will create a copy of a delta model and do the following:

1. the code will look at resources with the property mrs:hasBatch and it will find the number of batches to be created

2. it will create new resources with the same properties as the parent, the name of these new resources will be resource name + "batch" + n where n is just a number. These new resources might be linked to resources that are batch resources as well, meaning that new statements might be one to one or one to many.

3. After batch resources and their statements are added to the copy of the delta model, I removed the old resource which had the original mrs:hasBatch statement to avoid interference while pushing.

This was tested with creating multiple VMs at once where root volumes and network interfaces have to be batched as well.

[muzcategui1106]

While the code works for push(add) it has not been tested for push(delete) since I need to first come up with a solution on how to pull multiple resources. The current solution works at a model level, so it should work for any driver and any type of resources.

Current problems:

1. Figuring out a way to model batch resources when pulling
2. Push non-batch resources related with batch resources

[xi-yang]

Miguel,
Can you create an issue in VersaStack repo and move your comments there?

[muzcategui1106]

Mr Yang
How are you?

I opened the issue on the versaStack an include you in it. Tomorrow, I have
to meet with a professor and classmates for a presentation during the
morning. I was wondering if it is ok if I work from the university
tomorrow. I would come to the office Tuesday Wednesday and Thursday. Please
let me know if it is ok.

I hope you have a good Sunday.

Sincerely,
Miguel Uzcategui

On Sun, Sep 20, 2015 at 10:54 AM, xi-yang notifications@github.com wrote:

Miguel,
Can you create an issue in VersaStack repo and move your comments there?

—
Reply to this email directly or view it on GitHub
#2 (comment)
.