[Note: This post is the second in a series about the maturity curve of IT as it moves toward cloud readiness. Read the first post here about standardizing and virtualizing.]
I’ve met with many clients over the last several months that have reaped the rewards of standardizing and virtualizing their data center infrastructure. Footprints have shrunk from rows to racks. Power and cooling costs have been significantly reduced, while increasing capacity, uptime and availability.
Organizations that made these improvements made concerted efforts to standardize, as this is the first step toward IT optimization. It’s far easier to provision VMs, manage storage, and network from a single platform and the hypervisor is an awesome tool that creates the ability to do more with less hardware.
So now that you are standardized and highly virtualized, what’s next?
My thought on the topic is that after you’ve virtualized your Tier 1 applications like e-mail, ERP, and databases, the next step is to work toward building out a converged infrastructure. Much like cloud, convergence is a hyped technology term that means something different to every person who talks about it.
So to me, a converged infrastructure is defined as a technology system where compute, storage, and network resources are provisioned and managed as a single entity.
Sounds obvious and easy, right?! Well, there are real benefits that can be gained; yet, there are also some issues to be aware of. The benefits I see companies achieving include:
→ Reducing time to market to deploy new applications
- Improves business unit satisfaction with IT, with the department now proactively serving the business’s leaders, instead of reacting to their needs
- IT is seen as improving organizational profitability
→ Increased agility to handle mergers, acquisitions, and divestitures
- Adding capacity for growth can be done in a scalable, modular fashion within the framework of a converged infrastructure
- When workloads are no longer required (as in a divestiture), the previously required capacity is easily repurposed into a general pool that can be re-provisioned for a new workload
→ Better ability to perform ongoing capacity planning
- With trending and analytics to understand resource consumption, it’s possible to get ahead of capacity shortfalls by understanding when it will occur several months in advance
- Modular upgrades (no forklift required) afford the ability to add capacity on demand, with little to no downtime
Those are strong advantages when considering convergence as the next step beyond standardizing and virtualizing. However, there are definite issues that can quickly derail a convergence project. Watch out for the following:
→ New thinking is required about traditional roles of server, storage and network systems admins
- If you’re managing your infrastructure as a holistic system, it’s overly redundant to have admins with a singular focus on a particular infrastructure silo
- Typically means cross training of sys admins to understand additional technologies beyond their current scope
→ Managing compute, storage, and network together adds new complexities
- Firmware and patches/updates must be tested for inter-operability across the stack
- Investment required either in a true converged infrastructure platform (like Vblock or Exadata) or a tool to provide software defined Data Center functionality (vCloud Director)
In part three of IT Optimization and Cloud Readiness, we will examine the OEM and software players in the infrastructure space and explore the benefits and shortcomings of converged infrastructure products, reference architectures, and build-your-own type solutions.
Photo credit: loxea on Flickr