By: Pierre Moncassin
There is a rather subtle balance required to make a cloud organization effective – and, as I was reminded recently, it is easy to overlook it.
One key requirement to run a private cloud infrastructure is to establish a dedicated team i.e. a Cloud Center of Excellence. As a whole this group will act as an internal service provider in charge of all technical and functional aspects of the cloud, but also deal with the user-facing aspects of the service.
But there is an important dividing line within that group: the Center of Excellence itself is separated between Tenant Operations and Infrastructure Operations. Striking a balance between these teams is critical to a well-functioning cloud. If that balance is missing, you may encounter significant inefficiencies. Let me show you how that happened to two IT organizations I talked with recently.
First, where is that balance exactly?
If we look back at that Cloud Operating Model (described in detail in ‘Organizing for the Cloud‘), we have not one, but two teams working together: Tenant Operations and Infrastructure Operations.
In a nutshell, Tenant Operations own the ‘customer-facing’ role. They want to work closely with end-users. They want to innovate and add value. They are the ‘public face’ of the Cloud Center of Excellence.
On the other side, Infrastructure Ops only have to deal one customer – Tenant Operations. In addition to this, they also have to handle hardware, vendor relationships and generally, the ‘nuts and bolts’ of the private cloud infrastructure.
But why do we need a balance between two separate teams? Let’s see what can happen when that balance is missing with two real-life IT organization I met a little while back. For simplicity I will call them A and B – both large corporate entities.
When I met Organization A, it had only a ‘shell’ Tenant Operations function. In other words, their cloud team was almost exclusively focused on infrastructure. The result? Unsurprisingly, they scored fairly high on standardization and technical service levels. End users either accepted a highly standardized offering, or had to go through loops to negotiate obtained exceptions – neither option was quite satisfactory. Overall, Organization A struggled to add recognizable value to their end-users: “we are seen as a commodity”. They lacked a well-developed Tenant Organization.
Organization B had the opposite challenge. They belonged to a global technology group that majors on large-scale software development. Application development leaders could practically set the rules about what infrastructure could be provisioned. Because each consumer group yielded so much influence, there was practically a separate Tenant Operation team for each software unit.
In contrast, there was no distinguishable Infrastructure Ops function. Each Tenant Operations team could dictate separate requirements. The overall intrastructure architecture lacked standardization – which risked defeating the purpose of a cloud approach in the first place. With a balance tilted towards Tenant Operations, Organization B probably scored highest on customer satisfaction – but only as long as customers did not have to bear the full cost of non-standard infrastructure.
******
In sum having two functionally distinct teams (Tenants and Infrastructure) is not just a convenient arrangement, but a necessity to operate a private cloud effectively. There should be ongoing discussions and even negotiation between the two teams and their leaders.
In order to foster this dual structure, I would recommend:
- Define a chart for both teams that clearly outlines their relative roles and ‘rules of engagement.’
- Make clear that each team’s overall objectives are aligned, although the roles are different. That could be reflected through management objectives for the leaders of each team. However, this also requires some governance in place to give them the means to resolve their discussions.
- To help customers realize the benefits of standardization, consider introducing service costing (if not already in place) – so that the consumer may realize the cost of customization.
Follow @VMwareCloudOps and @Moncassin on Twitter for future updates, and join the conversation by using the #CloudOps and #SDDC hashtags on Twitter.
