My VMworld Breakout Session: Key Lessons Learned from Deploying a Private Cloud Service Catalog

By John Dixon, Consulting Architect, LogicsOne


Last month, I had the special privilege of co-presenting a breakout session at VMworld with our CTO Chris Ward. The session’s title was “Key Lessons Learned from Deploying a Private Cloud Service Catalog,” and we had a full house for it. Overall, the session went great and we had a lot of good questions. In fact, due to demand, we ended up giving the presentation twice.

In the session, Chris and I discussed a recent project we did for a financial services firm where we built a private cloud, front-ended by a service catalogue. A service catalog really enables self-service – it is one component of corporate IT’s opportunity to partner with the business. In a service catalog, the IT department can publish the menu of services that it is willing to provide and (sometimes) the price that it charges for those services. For example, we published a “deploy VM” service in the catalog, and the base offering was priced at $8.00 per day. Additional storage or memory from the basic spec was available at an additional charge. When the customer requests “deploy VM,” the following happens:

  1. The system checks to see if there is capacity available on the system to accommodate the request
  2. The request is forwarded to the individual’s manager for approval
  3. The manager approves or denies the request
  4. The requestor is notified of the approval status
  5. The system fulfills the request – a new VM is deployed
  6. A change record and a new configuration item is created to document the new VM
  7. The system emails the requestor with the hostname, IP address, and login credentials for the new VM

This sounds fairly straightforward, and it is. Implementation is another matter however. It turns out that we had to integrate with vCenter, Active Directory, the client’s ticketing system, and client’s CMDB, an approval system, and the provisioned OS in order to automate the fulfillment of this simple request. As you might guess, documenting this workflow upfront was incredibly important to the project’s success. We documented the workflow and assessed it against the request-approval-fulfillment theoretical paradigm to identify the systems we needed to integrate. One of the main points that Chris and I made at VMworld was to build this automation incrementally instead of tackling it all at once. That is, just get automation suite to talk to vCenter before tying in AD, the ticketing system, and all the rest.

Download this on-demand webinar to learn more about how you can securely enable BYOD with VMware’s Horizon Suite

Self-service, automation, and orchestration all drove real value during this deployment. We were able to eliminate or reduce at least three manual handoffs via this single workflow. Previously, these handoffs were made either by phone or through the client’s ticketing system.

During the presentation we also addressed which systems we integrated, which procedures we selected to automate, and what we plan to have the client automate next. You can check out the actual VMworld presentation here. (If you’re looking for more information around VMworld in general, Chris wrote a recap blog of Pat Gelsinger’s opening keynote as well as one on Carl Eschenbach’s General Session.)

Below are some of the questions we got from the audience:

Q: Did the organization have ITSM knowledge beforehand?

A:The group had very limited knowledge of ITSM but left our project with real-world perspective on ITIL and ITSM

Q: What did we do if we needed a certain system in place to automate something

A: We did encounter this and either labeled it as a risk or used “biomation” (self-service is available, fulfillment is manual, customer doesn’t know the difference) until the necessary systems were made available

Q: Were there any knowledge gaps at the client? If so, what were they?

A: Yes, the developer mentality and service management mentality are needed to complete a service catalog project effectively. Traditional IT engineering and operations do not typically have a developer mentality or experience with languages like Javascript.

Q: Who was the primary group at the client driving the project forward?

A: IT engineering and operations were involved with IT engineering driving most of the requirements.

Q: At which level was the project sponsored?

A: VP of IT Engineering with support from the CIO

All in all, it was a very cool experience to get the chance to present a breakout session at VMworld. If you have any other questions about key takeaways we got from this project, leave them in the comment section. As always, if you’d like more information you can contact us. I also just finished an ebook on “The Evolution of the Corporate IT Department” so be sure to check that out as well!