Today, enterprises across the world are crafting cloud strategies and starting to invest in private clouds to take advantage of the economic benefits that cloud models offer. In fact, Forrester found that 29% of enterprises consider building a private cloud a high or critical priority.
The more interesting finding, however, is that although enterprises greatly desire the economic benefits from a cloud model, those who have implemented a private cloud are not taking full advantage of those benefits. Why? Well, we discovered that it’s primarily due to two reasons:
1)lack of understanding about what a true cloud model consists of, and 2) underestimation of the critical components that make up a true private cloud (i.e., storage).
In fall 2010, NetApp commissioned Forrester Consulting to scientifically qualify the notion that companies lack clear understanding about clouds and hence underestimate the key components that are required to achieve maximum benefit from a cloud strategy. This paper reflects the results from a study of 150 US enterprise IT decision-makers who were surveyed about their private cloud interests and implementations. You will find that our study yielded three key findings:
Decision-makers are, in fact, confused about private clouds and traditional virtualized environments.
Private clouds are shared, highly standardized, and automated environments with self-service access to these resources. Of the 150 IT decision-makers surveyed, 27% said they were running a private cloud; however, the majority did not recognize these key characteristics and described their clouds like their traditional virtualization environments. For example, 54% anticipated providing the same tiers of service for different applications deployed within the cloud and the full breadth of storage options with little standardization or automation. In fact, only 17% of cloud users thought a private cloud would affect their traditional architecture deployments or environment options at all, showing a clear disconnect.
Storage is treated as “business as usual.” Most survey respondents (73%) expected their cloud to save them money, but only 37% expected their cloud storage to be more efficient. Why the disconnect? The respondents showed little understanding of how the storage in their private cloud would (or should) differ from traditional storage implementations. We know this because only 20% of cloud users said they offer standardized cloud storage options. We also found that 66% of cloud users planned to build their private clouds atop their existing storage assets, and for 50% of respondents, this means their existing, high-cost SAN infrastructures. Taking a “business as usual” approach to storage will limit your benefits from a cloud model and cost you more money in the long run.
Storage is not a key consideration in the decision-making process. It’s not surprising most cloud administrators (73%) have server/virtual server experience, whereas only 29% have core experience in storage administration. But what’s alarming is that storage administrators aren’t included in the upfront cloud architecture and design decisions. When it came to the decision-making, only 14% of respondents included storage administrators in private cloud platform selection, and only 21% included them in storage equipment and storage management software selections. Neglecting to involve storage administrators in key decision making processes is a red flag for cloud inefficiency.
Understanding the private cloud concept
Building a private cloud has to start with a basic understanding of how a cloud differs from a traditional virtualized environment. Forrester defines cloud computing as:
A standardized IT capability (services, software, or infrastructure) delivered in a pay-per-use, selfservice way.
In most cases, a private cloud is a shared IT infrastructure set up in your corporate data center and dedicated to just your company. Let’s break down the cloud computing definition to better understand its implications to IT operations:
Standardized capability. The value of a cloud comes in the flexibility it brings to the business — how rapidly and predictably you can deploy applications to this infrastructure. That value comes from provisioning and managing the environment in a consistent manner every time. You do that by standardizing what you deploy — which means prioritizing user options — and by automating as many processes as possible. It’s not customized or configured uniquely for each client, each time — that raises complexity, slows down processes, and is much harder to make repeatable. There may be options for consumers to choose some lightweight customizations they can apply atop the service, but the capability delivered is repeated, religiously.
Pay-per-use. A second core tenet is that the cloud is a shared environment where customers “pay” for the resources they use only when they use them. Efficiencies come from how much sharing takes place and how high you can drive the utilization of cloud resources as often as possible. A big reason clouds invoke payper-use costing is to keep the cloud efficient by incenting resource efficiency from the customer themselves.
While many enterprises won’t implement this model the same way public clouds do, they should account for and either show back or chargeback cloud consumption to maximize the overall cloud investment.
Self-service. For cloud services, self-service doesn’t mean a web page where you request the service. Selfservice is about providing self-service access through a portal to request services that have been defined in a service catalog. Thus it provides automation that requires minimal IT interaction so that the IT team can focus on more strategic business needs rather than routine tasks
The storage infrastructure for the cloud must be standardised and efficient
Nearly every successful cloud is built atop standardized, virtualized server infrastructures tightly integrated with consistent network and storage architectures that help achieve holistic economic advantage. This means standardization and automation don’t stop at the server. Storage must be provisioned with each new virtual machine and shared volumes must be easily integrated into the configuration. Standardization of the storage also means creating a service catalog of storage services attached to service-level agreements (SLAs) that satisfy the majority of infrastructure requests from your internal customers, introduce consistency, and streamline the IT
request and delivery process. This is critical to delivering the time-to-market and economic advantages of the cloud.
This study and other Forrester research show that most IT organizations focus on the server parts of their private cloud instead of optimizing each component to achieve maximum cost benefits. The storage architecture is commonly overlooked, leaving lots of efficiencies out of reach. Ideally, cloud storage should handle both latency sensitive and latency-insensitive applications at minimal cost, while taking advantage of thin provisioning, life cycle management, automated tiering, and live migration.
Cost efficiency of storage comes from reducing your total storage requirements; this translates into direct reductions in expenditures on storage and maintenance costs as well as additional savings on power, cooling, and floor space. Efficient storage allows you to offer higher SLAs to your internal customers at a lower price. Fifty-six percent of cloud users surveyed were planning to stick to the status quo with their storage options — expecting to have several tiers of service levels for different applications deployed in the cloud. Only 20% were choosing the cloud best practice of offering single-block and file-shared storage options with a single SLA.
To achieve optimum cloud efficiency with your storage architecture, you should consider the following key capabilities:
Efficient storage is the starting point. Whether a cloud is public or private, the key to success is creating an appropriate storage infrastructure in which all resources can be efficiently utilized and shared. Because data storage becomes even more crucial in a shared infrastructure model, efficiency becomes paramount.
Software efficiencies rely on built-in storage virtualization capabilities of advanced storage arrays and include capabilities such as thin provisioning, deduplication, and file cloning.
Data mobility within a shared resource pool. In a dynamic cloud environment, you need to be able to move data without interrupting running applications. Seamless data mobility helps you avoid the need for planned downtime as well as optimize operations by moving virtual storage controllers to balance load or achieve other objectives.
Standardize storage resources offered through a portal. By standardizing, customers enable their environment to become more uniform as they add new storage resources and retire aging devices. To make functions such as self-service efficient and ensure that operations of the cloud are as predictable as possible, you will want to guide your customers to a defined set of storage service procedures and repeatable processes that are defined in a service catalog. Creating a catalog of storage services that satisfy the majority of infrastructure requests from your internal customers will provide consistency and streamline the IT request and delivery process.
Rationalize the prioritization of data in order to ensure efficiency. When data is widely dispersed across a cloud-based infrastructure, it becomes more difficult to efficiently identify and retrieve “hot” data. Your customers always think all their data is the most important data in the company. But we know that there are classes of priority and that the value of data declines with time. Rather than explicitly exposing this to the customer as options that impede efficiency, solve for this on the back end using automated tiering, tierless solutions, intelligent caching, or other solutions that automate this process based on use patterns and relevancy. The method used depends on the type of workload. In some cases, reducing the number of tiers creates better efficiencies. Whichever method you choose, it’s more efficient than doing it manually. In most cases, the manual default leads to everything being placed in the highest-performing level of capacity, which is never the cost-efficient choice.
Secure multitenancy is important to consider for shared storage architecture. Only 14% of survey respondents felt multitenant security was of critical importance to their private cloud efforts. While private clouds have fewer security concerns than public clouds, enterprises will need to securely separate business units and applications from one another and from sub-administrators. For example, you’ll want to ensure financial applications can safely share your cloud without being accessible to non-financial employees. On the storage front, this requires a solution that can create multiple, separate virtual storage controllers on a single system
Importance of storage for cloud economics
Cloud economics relies upon running a large pool of resources as efficiently as possible. These benefits quickly decrease as standardization gives way to increasing customization, configuration, and options. Just as consolidating and virtualizing the server environment is key for running an efficient architecture, the same is true for storage.
Cost efficiency of storage comes from satisfying the majority of infrastructure requests as broadly and consistently as possible. Ignoring the cloud implications on storage or simply assuming it is the same as traditional storage keeps efficiencies at bay.
Your cloud storage architecture should deliver:
Increased storage utilization. For years IT departments purchased new machines to support new
applications and storage needs despite additional unused space in existing infrastructure. Consolidation, virtualization, and thin provisioning have begun the process of eliminating this unused space. Your private cloud environment relies on these concepts of higher utilization to achieve maximum cost reductions.
Reduction in provisioning time. Provisioning storage volumes is a highly repeatable task, and if it requires time from infrastructure and operations professionals, that represents cost to your cloud, time to deploy, and time away from higher-value actions that require the brainpower of these administrators. Standardization and automation are key to cloud economics, and the optimum storage architecture is one that integrates seamlessly with your cloud automation platform of choice. Storage is a key component of this architecture and should not be an afterthought, as you may end up having to retrofit your storage to your cloud to achieve predictable, efficient integration.
Consistent storage management automation. You want to provide different storage services to your cloud customers, but they shouldn’t come at the cost of consistent management or the ability to easily automate these functions. This requires a storage system that unifies the key storage functions for both assignment and protection of data. A single storage automation system that can assign volumes across different storage types, migrate data according to life cycle management policies, and centrally back up and take snapshots of data volumes will allow multiple storage tiers and SLAs to be managed as one. A storage solution that can adhere to higher-level cloud policies can dramatically improve cloud management efficiencies.