Future job of the senior infrastructure manager

Hybrid cloud architectures will need IT departments to move from core server skills to managing the capacity to let developers code

Hybrid cloud architectures will need IT departments to move from core server skills to managing the capacity to let developers code.

It is inevitable that enterprise IT in 2020 will comprise a hybrid mix of on- and off-premises services. While your particular combination of cloud services will vary, it is unlikely any enterprise IT department will still primarily focus on configuring server, storage and network devices as a core competency.

The shift to business technology and IT as a service is well underway, so you can either ignore it, try to contain it or embrace it.

In the cloud era, your new role is to establish guardrails to guide developers to the best cloud services; get out of the way while they deploy, test and release them; and then take over ongoing operational support, so they can get back to coding. But you have to earn the right to be their cloud manager.

The operational skills you developed to manage traditional datacentre infrastructure need to be refocused on the application tier, since this is where you will add the most value.

Build your cloud management capabilities with two primary objectives in mind. First, consider how you can make developers more productive. Second, assess how you can optimise the runtime performance, availability and cost-efficiency of cloud applications, wherever they are deployed.
Start preparing now to manage an IT portfolio based on services automatically deployed on demand and from an elastic pool of infrastructure, most of which you will not own. That means you need to start thinking about applications first: how your developers can build and deploy them faster, and link them more easily to a range of existing and new business services and data sources, in-house and outside your enterprise walls.

Cloud security

Cloud security starts with identity verification, access controls and permissions. Extend your existing identity services into the cloud by taking control of existing public cloud user accounts, security keys and credentials. Define and enforce role-based access controls across clouds to restrict access to specific pools or types of resources by team or business unit. Acknowledge these limitations clearly in your service catalogue.

Review current regulatory and corporate compliance constraints. These determine where you will let applications run and where data can reside (public or private cloud, or a combination); you are ultimately responsible for compliance, whatever the mix. In addition, determine your tolerance for shared multi-tenant environments, either on or off-premises. Integrate cloud security with your existing lightweight directory access protocol (LDAP) and/or Active Directory infrastructure and include spending caps by role to keep cloud costs under control. Finally, make sure you actively track, log and report all security and compliance events, from configuration changes to placement of sensitive data.

Cloud automation

You won’t own the entire infrastructure as your cloud portfolio grows. You will probably have a mix of on-premises private clouds built on virtualised or converged infrastructures as well as off-premises public cloud infrastructure that, for the most part, is managed for you. In either case, leave infrastructure management to the cloud providers (private or public) – they are responsible for multi-tenancy, resource pooling and scale-out in each cloud domain, for instance. The providers expose the building blocks; your job is to integrate and standardise them to simplify and accelerate consumption.

Standardise and automate everything. Traditional IT provisioning is a slow and manual process, while cloud provisioning is on-demand, automated and application programming interface (API)-driven. Developers do not want to fill out a help desk request, nor will they tolerate a lengthy approvals process to get access to cloud resources.

Beyond provisioning, cloud consumers also expect automated compliance, availability, scalability and performance management features. If you have built your problem diagnosis and remediation processes for specific infrastructure or applications you own, they will need to be generalised to handle new cloud-specific metrics and to enable automated remediation. As your cloud application life cycle compresses, there will be no time to craft custom management processes for each new application.

Cloud management

Understand why developers use new cloud management tools. They are likely to rely on newer cloud management tools already – understand why. Cloud developers turn to Dell’s Enstratius, RightScale Cloud Management, Scalr, ServiceMesh’s Agility Platform and others and rely on Opscode’s Chef and Puppet Labs’ Puppet for automation because these tools aim squarely at simplifying their lives.

From unified dashboards, they simplify application design via re-usable templates and blueprints and offer push-button deployment to multiple clouds. They automate configuration, scaling and recovery operations and consolidate multiple cloud accounts, users and roles in a common framework that hides infrastructure complexities. In short, they bring order to the chaos of multi-cloud management without limiting developer productivity.

If developers continue to manage cloud themselves and do not see value in central IT’s cloud management capabilities, they will go around the IT infrastucture staff. Not only will this cut into your developers’ productivity, it will foster the notion that cloud is competition for the infrastructure team. Don’t let that idea take hold. Embrace the hybrid cloud opportunity by proving you can lower the burden on business cloud users wherever they use cloud services and that you have the right skills and tools to make sure their apps deliver the right user experiences.

Find out what your developers are doing that keeps them from coding. You will learn the most from your early cloud adopters. Spend time with them to understand how much infrastructure control they need and how much they are doing just because they have to. Ask them where they need more visibility and focus your monitoring efforts there. Find out why they choose a particular public cloud, so you can establish a baseline for your own cloud operations — what do they like and where are the gaps?

Catalogue your existing cloud services. Before you can define your cloud management requirements, you need to pick a starting set of deployment models and determine what level of abstraction is available in each. Which integration APIs are available, what types of infrastructure, which development tools? How much operational management is provided for you by the cloud platform itself, and what will you have to build or acquire to fill in the gaps?

It is likely IT infrastructure managers will spend much less time configuring servers and installing management software in the future. These people will spend more time extracting meaningful insight from performance metrics and negotiating service provider agreements.

This is an extract of the Forrester report: Cloud Management In A Hybrid Cloud World (July, 2013) by Dave Bartoletti.

Next Steps

How to land a converged infrastructure job

Read more on Software development tools