You wouldn't race down the motorway the day after you've passed your driving exam. You need to mature: learn the basics, build up your experience, and plan ahead. Virtualisation is no different. Almost every organisation is rushing headlong to deploy the technology, create the building blocks for cloud computing, and capitalize on the advantages of agility, cost, quality, and risk. But rushing with the speed of a Formula 1 driver into a virtualisation strategy - whether it is a stand-alone virtual or hybrid physical and virtual environment - is fraught with risk. Don't expect to deliver a dynamic, self-service virtualised datacentre overnight. Like driving, you need to adopt a maturity approach - building your virtual cloud future one step at a time. Try and do it all at once, and you'll either stall or crash.
The virtualisation journey from straightforward server consolidation to dynamic datacentre can be long and hard. Even the best virtualisation plans struggle, sooner or later, with added complexity, staffing requirements, service level agreement (SLA) management, departmental politics, or something else.
The abstract and increasingly dynamic nature of virtualisation makes it easy for rogue deployments to go undetected as they circumvent security and compliance processes, overwhelm networks, and obscure the root causes of business service problems. Most organisations struggle to use virtual resources in more complex, mission-critical applications, as they must be assured and managed accordingly to their role in business services - something not directly the expertise or interest of virtualisation platform specialist.
Don't be floored by VM sprawl
During the initial server consolidation phase, when the organisation is migrating from the physical to the virtual, you'll want to avoid the common pitfall of VM sprawl. This is where the number of virtual machines running in a virtualised infrastructure increases over time, simply because of the ease of creating new VMs, not because those VMs are absolutely necessary for the business. The concern here is the overuse of the infrastructure if it is not needed and the cost of licenses for virtual machines that may not have been required.
To prevent sprawl, you need to carefully analyze whether you really need all those shiny new VMs and be certain you can justify the cost of the infrastructure and the licenses for all VMs created. Virtualisation management technology is the answer here: allowing you to better provision, control, assure, secure, and optimize stand-alone virtual or hybrid physical and virtual environments. Management software can, for example, ease the physical-to-virtual (P2V) migration process with system-independent image backup and P2V restore directly to VMware or Hyper-V. And it can provide continuity through consolidated backup and onsite/ offsite recovery to ensure more uptime.
In fact, most organisations face one or more clear 'tipping points', where virtualisation deployment stalls as they stop to deal with these new challenges. This VM stall, a concept first introduced by Andi Mann from CA Technologies, tends to coincide with different stages in the virtualisation maturity lifecycle - such as the transition from tier 2/3 server consolidation to mission-critical tier 1 applications; or from basic provisioning automation to a private/hybrid cloud approach.
If your deployment is close to stalling, there's an accelerator pedal close by - one that will help you to drive further efficiency and performance. Forget about provisioning virtual servers one by one: when the business user knocks on your door and asks where to put a new departmental system with 12 users, you know you don't have the resources to do it manually. You need to automate - industrialize - the process as though it were a factory, based on a standard best practice procedure.
Automation has many advantages here. It offers the ability to identify, track and standardize your virtual infrastructure, integrate and orchestrate processes across IT silos, and accelerate the provisioning of virtual services. Once the industrialized provisioning process is live, automation technologies can then be used to monitor application traffic response time or perform root-cause analytics to help isolate and remediate virtual environment faults quickly. After that, you can move onto the next step in VM management: automated, self-service VM provisioning, management of resource pools, and individual VM tracking with chargeback accounting and reporting.
Avoid hardware vendor lock-in
A word of caution at this point regarding a speed camera ready to catch the unwary: Beware lock-in. Many hardware vendors now offer virtualisation management software as part of their bundle, sometimes free, often even with the hypervisor embedded into the machines themselves. That's fine if you want just to manage a single server, or a small group of servers. But in today's dynamic datacentre environment, you should consider the flexibility to quickly and efficiently move a virtual image seamlessly from platform to platform.
Consider the lock-in implications carefully with regards to your management software as it relates to your current and future virtualisation platforms such as VMware, Microsoft, Citrix (Xen) and Red Hat. Plan for change. Your tools should work with your current and future processes.
Flexibility to rapidly provision cloud capacity
As more organisations move to the cloud, they will tell you that elasticity is the centerpiece of a successful virtualisation strategy. As demand for a service rises and falls, you'll need the flexibility to rapidly provision capacity using your private or hybrid cloud. Imagine doing that manually! Automated, template-based self-service virtual server provisioning tools can help you provide resource pooling, provisioning additional capacity from Amazon EC2 and Amazon VPC solutions or other public pools, integrated self-service (with chargeback), together with capacity and utilization tracking and reporting.
Most organisations aren't betting solely on evolving their existing investments into a private cloud. Many are placing parallel bets where at least new applications and services can take advantage of new more revolutionary approaches, designed from the ground up for dynamic cloud environments. This means that services must be designed from the outset for dynamic provisioning and scaling. To date, however, application development and operations have been largely disconnected, lacking an integrated Dev Ops approach to managing the business logic layer, down to load balancers and network-attached storage.
The latest automation solutions overcome this challenge, allowing organisations to rapidly define complete, complex composite applications, including components like load balancers, firewalls and even storage from a single platform; to many this is the 'holy grail' of the dynamic datacentre. The technology allows you to take virtual appliances and assemble them into composite services through a simple drag and drop interface. These services are then deployed and scaled as needed using commodity virtual machines from a private pool or public cloud service. One interface manages the cloud, providing access to high availability features, resource metering, and operations monitoring.
Reaching the dynamic datacentre
Without robust, insightful management tools, virtualisation will stay stuck in the slow lane, limited to basic deployments. You need new and better ways to harness the full IT and business potential of virtualisation by breaking down the significant barriers standing in their way. Virtualisation management solutions provision, control, secure, assure and optimise your virtual environments. Robust management, together with advanced automation, enable you to progress from an entry-level virtualisation project through a mature dynamic datacentre and private cloud strategy.
You'll be in the fast lane overtaking competitors quicker than you think.
Gregor Petri is Advisor, Lean IT and Cloud Computing, for CA Technologies.
Join the CIO New Zealand group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.