posted on Wednesday, May 19, 2010 3:18 AM
Three simple action items can help ensure your next infrastructure refresh cycle leaves your data center prepared and smelling minty fresh*.
Most rational folks agree: public cloud computing will be an integral piece of data center application deployment strategy in the future, but it will not replace IT.
Just as Web 2.0 did not make extinct the client-server model (which did not completely eradicate the mainframe model) neither will public cloud computing marginalize the corporate data center.
But it will be a part of that data center; integrated and controlled and leveraged via the new network.
As it implies, your network today is probably not the “new” network. It’s not integrated, it’s not collaborative, it’s not dynamic. Yet. It’s still a disconnected set of components covering a variety of application-related services: acceleration, optimization, security, load balancing. It’s still an unrelated set of policies spread across a variety of control mechanisms such as firewalls and data leak prevention systems. It’s still an unmanageable, untamable beast that requires a lot of spreadsheets, scripts, and manual attention to keep it reliable and running smoothly.
But it shouldn’t be for long IF there’s an efficient, collaborative infrastructure in the future of your data center. There probably is given the heavy emphasis of late on green computing (as in both cash and grass) and the more practical goal of more efficient data centers.
WHAT YOU CAN DO TO PREPARE
One of the arguments for cloud computing is the reduction in capital expenditures and the transfer of previously CAPEX related funds to OPEX. But this argument misses a fundamental fact of life in a corporate environment: employees still need access to the “public” cloud. Most organizations are playing around and testing the waters that are “cloud computing” but they aren’t moving lock, stock, and barrel. That means there is a mix of publicly and privately deployed applications that need to be managed.
Additionally, there’s a desire to reduce the operational costs associated with running a data center, which ultimately points to the same place that a holistic management solution does: a dynamic, connected infrastructure. There will still be networks (albeit possibly less complex) and there will still be a need for operational folks to manage not only the private network but the applications and (one hopes) infrastructure services that have merely shifted location, not importance to the success of the business IT is tasked with supporting.
You’ll still need to maintain a reliable network infrastructure, which means there are some opportunities to prepare for the “new” network as part of your normal infrastructure maintenance lifecycle. In preparation for a refresh cycle, get all your ducks in a row and make sure that as you look at upgrading and perhaps replacing key components in your infrastructure the solutions you consider move you toward a dynamic infrastructure rather than simply maintain the status quo. In other words, speeds and feeds upgrades are nice, but there’s more to building a dynamic data center than performance considerations. You’ll need to think more like a software architect than a network architect if you’re going to move forward and support new, more efficient data center models like cloud computing.
| INTEGRATION CAPABILITIES |
Infrastructure 2.0 is about collaboration and dynamism. Its focus is on ensuring that critical network and application delivery network infrastructure is capable of integrating into the broader data center ecosystem and collaborating (sharing data, events, alerts) either directly with other components or via a data center management framework. New infrastructure components should be be API-enabled (REST or SOAP, whichever you prefer) in order to facilitate the integration necessary to automate and ultimately orchestrate operational data center processes.
If the solution is not enabled, ask whether it will be. If this is simply not an option then carefully consider whether other options (remote scripting, etc…) will scale well within your environment. Evaluate integration options with frameworks such as Puppet and Chef in addition to traditional management vendor solutions for enabling future automation, with an eye toward extension of the frameworks to include network components or pre-existing integrations. |
|  ACTION ITEM: Make sure that new network components can be integrated into operational process automation and frameworks
|
| SUPPORT for MULTI-TENANCY |
Whether the devices themselves directly support multi-tenancy or fit nicely into an architectural multi-tenancy approach, remember that as you move toward a more dynamic infrastructure you’ll need ways in which you can isolate and segregate customers. Customers might be traditional customers or they might be business unit, departmental, or organizational in nature. Regardless of how you define customer and whether or not you’re planning to incorporate on-premise or off-premise cloud computing models, you’ll need a way to meter and charge-back/bill the customer based on usage, and that will require some form of multi-tenant support.
You’ll also want a way to ensure some level of fault-tolerance so critical business applications aren’t negatively impacted by other applications. Because most network infrastructure isn’t inherently multi-tenant, this may require an architectural approach and might involve the use of virtual network appliances to implement a hybrid physical and virtual architecture. |
|
ACTION ITEM: Look for solutions that support multi-tenancy in terms of resource (CPU, memory) and access (configuration and management) control. |
| APPLICATION (CONTEXT) AWARENESS |
| One of the hallmarks of cloud computing and similar transformative data center models is that it refocuses the role of the network as a delivery mechanism for applications. Cloud computing is a deployment model and the focus is to improve the efficiency, scalability, and ultimately availability of applications through streamlining of operational processes across the application lifecycle. That means infrastructure should perform a specific task but that it should do so with an eye toward supporting the rapid deployment, scale, and delivery of applications. Infrastructure that is application-aware can help achieve a dynamic infrastructure because its policies are generally focused on applications rather than on IP addresses or network components. Context-awareness is even more beneficial, as it can help to better deliver and scale applications based on the broader application context, encompassing end-points, application infrastructure, and network infrastructure. |
|
ACTION ITEM: Evaluate the solution with an eye toward how it will further the goal of delivering, scaling, and securing applications |
It should be noted that these action items are in addition to your existing infrastructure acquisition requirements. Budgetary constraints, speeds and feeds, and other features should certainly also be considered (as they always have) as part of any acquisition evaluation process. But including these three items in addition to existing requirements will help ensure that your data center has a strong foundation of dynamic infrastructure that will better enable the entire data center to be prepared to handle the coming pressures and challenges associated with emerging but disruptive deployment models.
* YOUR MINTY FRESHNESS MAY VARY DEPENDING ON WHETHER THE VENDOR REPRESENTATIVES VISITING YOUR DATA CENTER HAVE MINTY GUM OR NOT.
