2010 Prediction: Rich Corley, Akorri

By Rich Corley
published: Thursday, December 31 2009

Virtualization in 2010: The Importance of Storage

I’m a long time storage guy. When I started in information technology, disk drives were 14” and weighed hundreds of pounds. Back in the mainframe era, never losing data was the utmost requirement. Data storage was critical to any IT organization and the lifeblood of any company stored on disk drives called DASD (Direct Access Storage Device). Because of this, storage suppliers focused on ensuring applications always had access to their data…multiple paths to the data in case a path failed, enabling multiple copies of data, local and remote, in case of drive or site failures.  It was critical that application performance be highly optimized and also, given the high costs of computing in the mainframe era, the same was true for utilization of the infrastructure (Storage, Compute, Memory, Network).

I’ve seen much evolution of storage technology and how it is purchased and deployed. In the past when dollars were cheap, the ongoing development of storage technology enabled a great variety of low cost solutions, which was wonderful for IT organizations needing to deploy new applications at reduced costs. Over-provisioning of storage became the norm in solving performance problems because of low cost, even if it gave IT more infrastructure to manage.

What’s interesting is that if you compare the industry and economic dynamics of the mainframe era with what will happen in 2010, I believe you will see a striking similarity – data storage is a critical business asset, and economic drivers are making optimization of that asset a top priority. What is different are the tools needed to achieve this goal given the advent of virtualization and the new levels of data center complexity it has created.

In the mainframe days, data center administrators used System Management software to understand all aspects of the system to ensure the maximum possible performance and utilization was achieved. Today, a similar end to end solution is needed for virtual infrastructure. 

Early phase one virtualization projects for most organizations included finding all of the lightly used servers and consolidating less than critical applications.  Performance may not have been what was expected, but could be managed with the virtualization technology element manager.  As organizations embark upon the next phase of virtualization, which involves virtualizing servers running business critical applications and at higher utilization rates, the tools used in phase one no longer work to ensure and demonstrate to the business that the virtualized environment is meeting required IT service levels at lower costs.
In 2010, there will be very little tolerance for risk, and Infrastructure Performance Management tools that provide metrics and analysis to guarantee performance within a virtualized environment will fill that void.  These tools are focused on the entire system operating together to support the applications.  A key capability of these tools will be to identify issues of resource contention in the storage array created by virtualization.

Storage array data is created and modified by applications and is read and written onto them, but arrays have no visibility into what’s going on with a given application, so what happens when applications start becoming more dynamic or virtualized? Welcome to 2010. Managing storage isn’t easy and many mid size organizations that are heavily deploying virtualization don’t have storage specialists.
Dollars are expensive now. And while server virtualization is wonderful, the application demands on the shared storage infrastructure become more complex. Applications that were once static now dynamically migrate, and the importance of managing storage as part of the complete IT infrastructure is as important as it was in the mainframe days, if not more so. For IT administrators, being able to troubleshoot, manage service levels, optimize, and plan IT infrastructure easily and efficiently has become a key business requirement. They are being asked to spend less time building infrastructure and more time on managing and optimizing it to deliver application performance and efficient infrastructure utilization…in other words, do more with less AND reduce costs.

To recap, what was true in the mainframe era will be as true in 2010, even as storage has evolved:  1) Cost in the data center matters… now more than ever and 2) Storage is a critical business asset and it is difficult to manage and optimize.

What has changed is the complexity. How will 2010 data centers address this problem? 1) Take a complete system view of infrastructure 2) Understand storage workloads at a system level to further optimize storage by allocating storage type (Flash/SSD, 15K FC drives, lower-cost SATA drives) by the type of workload best suited for it. 3) Use Infrastructure Performance Management software to enable monitoring and troubleshooting, management of service levels, optimization and planning for both virtual and physical infrastructures in order to make IT administration easier and more effective.