Getting all the pieces of a data center infrastructure puzzle to fit together is a difficult task. Interoperability problems are common, and vendors often point to other manufacturers’ products as problem sources. This complicates heterogeneous systems management and delays critical product support. Over the last few years, we’ve seen more data centers take an integrated approach–where one vendor supplies an entire data center’s hardware–in an effort to help solve some of these problems.
Download for later:
- Internet Explorer: Right Click > Save Target As
- Firefox: Right Click > Save Link As
In this podcast, Nick Martin, assistant site editor for SearchDataCenter.com, talks with Bill Kleyman, virtualization architect for MTM Technologies Inc., to find out more about integrated infrastructure platforms and what factors IT pros should consider before adopting a single-vendor approach.
Nick Martin: What are the tradeoffs for integrated infrastructure platforms like blade versus rack hardware?
Bill Kleyman: There are always going to be pros and cons with whichever data center infrastructure platform an organization decides to go with. The key differentiators are the clear business drivers of that company. Both a rackmount hardware environment or a blade-chassis solution could work very well.
Look at it this way–HP ProLiant rackmount servers come in packing a punch. They’re capable of handling hundreds of gigabytes of RAM with numerous cores, depending on your processor. These machines can handle virtualization, application hosting and general server requirements. The trick is understanding how this hardware plays a role in your organization’s long-term and short-term goals.
The best way to understand this is to look at an example, such as a company that is rapidly expanding. Say they’ve adopted a virtualized platform on Citrix Systems Inc.’s XenServer or VMware and will be in need of hosting servers. From there, they plan to stream full corporate desktops as a part of their “bring-your-own-device-to-work” initiative. This environment will be ever expanding and growing. What is the best solution for this scenario?
If they buy standalone servers, administrators will have to manage each box individually or through some third-party management tool. They’ll also need to manage workloads, hardware profiles and networks to make sure all of their independent servers work well together.
On the other hand, we have an integrated infrastructure. Cisco System Inc.’s Unified Computing System (UCS), is a great example of how far these technologies have come. In using this solution, the company in our scenario will have all of its necessary data center infrastructure components under one roof. Things such as networking and switching, blade management and chassis control are all under one graphical user interface (GUI). Administrators can copy entire hardware profiles and load them onto dormant servers for quick configurations. Management and growth becomes an easier data center function.
Now, if that same company still wanted to virtualize, but knew that its computing demands were limited, then going with some rackmount servers might be the best option.
Martin: What are the pros and cons of integrated infrastructure management?
Kleyman: Ease of manageability and growth potential are easily some of the biggest benefits of going with an integrated infrastructure. The management tools built around controlling and modifying blade chassis has become extremely granular.
Administrators are able to sign in and immediately view everything they need to know about their environment. Using the UCS GUI as an example, an engineer can sign in to see if there are any critical errors in the environment. If they see an issue, they are able to drill down to the exact blade that is experiencing the problem and troubleshoot it all the way down to the DIMM slot level. Managing hardware profiles becomes easy as well. Simply slide a new blade into the chassis and you’re able to copy a hardware profile from one blade to the new blade and have it ready in no time. That means deploying an entire chassis full of blades can take minutes instead of days.
The cons are fairly straightforward. One of the biggest benefits of using an integrated infrastructure is the “one-throat-to-choke” theory. However, this can be a downside as well. Remember, integrated infrastructure management is designed to specifically analyze and look at that given environment. That means anything you place on top of that environment must be monitored in a different manner. Don’t forget about workload management at the hypervisor level, backup and snapshots, and any required enterprise application management. These will still be standalone, and they will still need monitoring and management.
Martin: What are the issues associated with integrating separate diverse platforms into an integrated data center infrastructure?
Kleyman: Compatibility always becomes the question when unifying a heterogeneous platform with an integrated one. That said, there is nothing that states that these two environments can’t live and play well together.
Before moving in the direction of an integrated solution, make sure all of the major components are compatible. For example, if you have a storage area network (SAN) capable of only iSCSI connectivity, you may run into issues when it becomes necessary to connect fiber into the backbone. Also, make sure that the switching environment can handle the new intensity of a blade environment.
Martin: What are the main considerations when migrating from a heterogeneous data center infrastructure to an integrated infrastructure platform?
Kleyman: Before moving to an integrated infrastructure, make sure there is a good business case for it. Conduct research to make sure that your environment, applications and workloads can all handle the change.
This means verifying that your SAN is capable of handling new connections to a blade chassis and making sure the performance will be there to back it up. Also, make sure your switching infrastructure is set up as well. Older wiring and old switches can be detrimental to a new integrated infrastructure solution.
One of the best pieces of advice is to run a proof-of-concept analysis with a chassis of your choice. One of the greatest benefits of deploying a solution like UCS is that it’s always expandable. Start out with two or three blades to gauge performance under normal workload stress. By conducting a proof-of-concept analysis inside the environment, an IT team can quickly see how the chassis interacts with the existing infrastructure.
Martin: So far we’ve seen slow, but notable, growth in the adoption of integrated infrastructure. Do you expect this same trend to continue?
Kleyman: In 2010, SearchDataCenter.com awarded Cisco's UCS B230 M1 Blade Server the bronze medal in their annual Products of the Year Awards. There were key reasons behind this decision. Blade and chassis environments are becoming more prevalent in data centers. As demands for consolidation, virtualization and performance come into play, an integrated infrastructure environment becomes a very viable solution. This has become even more common since deploying an integrated infrastructure no longer blows an IT budget out of the water. Blade and chassis solutions have become more affordable and a much more logical next step for many organizations.
There has been a lot of industry excitement and push toward these types of environments. In fact, alliances and partnerships have been developed around integrated infrastructure. Recently, a fully validated data center infrastructure design was released by combining three top-tier solutions from Citrix Systems, NetApp Inc. and Cisco. That is, using a Cisco UCS chassis as the hardware platform, virtualizing workloads and streaming desktops with Citrix and storing everything on a NetApp FAS storage system.
Moving forward, as more organizations begin to see the benefits of an integrated infrastructure and how it can help them grow, we’ll see the demand rise proportionately.
This was first published in October 2011