IT Focus Area: Infrastructure Optimization
May 26, 2015
How to Refresh Storage without Overspending
Storage demand is exploding in the enterprise, driven by digital content and media, virtualization, disaster recovery, big data, and retaining information for compliance purposes.
With demands growing so fast, and for some organizations, so unpredictably, overestimating need—from the application owners to the storage administrator—has seemed like the path of least resistance. The logic being that all that extra storage will probably be useful at some point. But overbuying may no longer be the safest, most viable option.
So, how did we get here?
First, many sizing processes don’t provide accurate information. Enterprise storage resource management (SRM) packages sold both by storage hardware and management software vendors aren’t perfected.
These solutions can be notoriously difficult to configure, can trigger complex change-management processes and do not provide cross-vendor management that is essential for heterogeneous storage environments.
To reduce the need for more—and more expensive—storage and avoid over-provisioning, it is important to examine the actual performance and utilization trends from within the storage arrays themselves. Within the native array tools are the secrets to a more efficient future-state storage environment, the ability to purchase less capacity and use more of it.
Some Other Things to Consider When Refreshing Storage
Knowing what to consider when determining storage needs is important in order to make accurate predictions without becoming expensive, and burdensome, IT projects. The key factor is to make an analysis that is as fact-based as possible without over-extending IT resources.
There are five things to consider for a successful storage refresh.
1. Document current storage trends
It is important to track and log information from existing storage, software and server monitoring tools over a sample period of time. This will help you discover and document current trends in utilization, performance, and allocation that will impact future storage decisions. For example, in reviewing monthly performance trends, many enterprises identify patterns and subsequently plan for activity known to take place on a certain date or time.
2. Track storage capacity
Take advantage of server monitoring tools to track storage capacity presented to the server, total file system and database utilization over a sample period of time. Examining these trends can help predict when the storage will likely be filled. For example, documenting the raw, usable and allocated storage over time for each storage device intended for refresh.
3. Examine storage workload patterns
Look at patterns over different times of day, different days of the week and different times of the month. It is important to understand usage peaks and valleys. It may make sense to consult a third party to help extract this data and to build a comprehensive picture of storage utilization and performance over time.
4. Understand future business needs
It may be helpful to bring trending data (see points 1-3 above) to meetings with application owners. This will help produce more accurate predictions of growth and application service-level requirements. Overestimating future needs is a big driver of over-purchasing. Current utilization trends and capacity data are important data points to consider when planning for the future. It is also important to consider corporate initiatives such as reorganization, mergers and acquisitions. These initiatives could affect storage use in a dramatic way down the road.
Are there any major new IT projects or data center migrations planned over the next three years?
How are these new initiatives likely to translate into storage volume and service levels?
5. Purchase emerging storage technologies wisely
It is important to consider emerging technologies, along with performance requirements. Some newer technologies may be more suitable for meeting those needs than general-purpose or high-performance disk storage. For example, enterprise solid-state drives (SSD) can be a great way to maximize performance of virtual servers, databases and other related applications. They tend to be expensive compared to disk storage, but the intelligent use of these solutions can have a dramatic impact when performance is important for the bottom line, such as with online transaction processing, high-performance computing in financial services or biotech firms. When used wisely, a small amount of SSD may eliminate the need for larger high-performance disk storage options, and may even reduce the number of servers in the data center, with all the requisite power, cooling, and real estate cost benefits.
Break the Cycle of Overspending on Storage
By considering past storage patterns and likely future developments, a fairly complete picture of what storage requirements will be down the road is possible. It is achievable without having to get involved in a complex, resource-intensive storage resource-management implementation. If started early, trends and patterns can be monitored over time to reveal opportunities of what to optimize today and what can be deferred. The information and perspective that are gathered today may end up being very useful for predicting future hardware refresh needs down the road, helping break the cycle of overspending on storage.
A Financial Services Company Case Study
Six sites and 10 mid-range arrays consolidated to two sites and two enterprise systems.
Identified concurrent peak performance activity across 10 arrays to evidence that consolidation to two could adequately meet current and future workload requirements.
Avoided acquisition of one additional array per site, resulting in approximately $800K in capital expenditures (CAPEX) savings.