Sam Walker

UCP Director API - Full infrastructure automation

Blog Post created by Sam Walker on Oct 1, 2015

Inspired by a conversation yesterday with Paul Meehan, I thought I'd spent the time to post about what the UCP Director API is and why it should be interesting to people looking to use a cloud management platform or automation suite to move to the nirvana of an IT infrastructure that manages itself.  In reality, there still needs to be someone to manage an automated IT infrastructure, so this is more about having the ability for service providers, big VM consumers, customers who have already bought in to the idea of IT automation and a cloud management platform or portal, etc. giving customers the ability to spend time doing what they need rather than unnecessary time consuming BAU tasks.

 

First of all; what is UCP.  Hitachi are modest when it comes to selling its virtues of a UCP solution; I am being careful for this to fully remain a technical post rather than a marketing one.  I'm a techie so will try and tone down the Marketing lingo as much as possible!  In short, UCP is Hitachi's converged infrastructure offering that comes in a variety of flavours.  The flavour I am talking about here is 4000 / 4000e (the 'E' is for entry).  Full details can be found here:  https://www.hds.com/products/converged-infrastructure/hitachi-unified-compute-platform/

 

A UCP solution is a pre-validated infrastructure running on Hitachi storage, Hitachi Compute Blades and a variety of Cisco or Brocade networking.  For the UCP 4000 / 4000e solution specifically, this is a converged VMware or MS Hyper V platform that has a product (UCP Director) which orchestrates the management of the whole stack; through the vCenter or SCVMM, the provisioning of networking, storage and compute blades is run through the UCP Director plugins.  To say 'plugin' really does undersell the integration work we have done with VMware vCenter; I can't talk with full authority on the Hyper V solution but it's a similar level of integration.

 

What this means is that the time and manual tasks associated with tasks (such as provisioning a LUN and presenting it to a cluster, configuring an additional VLAN, deploying a cluster of blades with Auto Deploy, etc) is all run through the UCP Director; everything is orchestrated.  Additionally, there's also a set of PowerCLI cmdlets (over 250) and a REST API interface which mean all tasks that can be executed through the GUI can be done through a cloud management platform too.

 

An example below using the REST API:

 

Just using the Chrome REST Client, using one line I have created and attached a LUN to an existing cluster - the details included are below (cluster name (in the URL), pool ID, size in bytes, whether the volume should be formatted or not, storage system ID, ports (if you want) and volume name). 

 

1.PNG

After posting this one API call, I can see from within vCenter the tasks that are taking place; creating volume in pool 0, attaching that to the hosts, making sure there's the correct FC zoning and formatting a volume on the LUN.

2.PNG

Which we can see has been added to the cluster as a datastore (seen here on only one host, but it was easier to screen shot!)

3.png

If using this API within vRA, a threshold action could be to create a new datastore using the UCP API, then add the datastore to a datastore cluster using the VMware API - so when your datastore cluster reaches a certain threshold, it's automatically expanded with another LUN.


This is a simple example which scratches the surface of what is available; for example, through a single line API call, you are able to create clusters, deploy bare metal hosts (Windows or Linux) servers, apply VLANs to the physical switch infrastructure, upload firmware to UCP, etc.  all through the single API.


For info on the full functionality available, have a read through https://www.hds.com/assets/pdf/hitachi-ucp-director-api-reference-v-4-0.pdf


Outcomes