Storage Infrastructure for the Cloud
With total IT spending on cloud computing forecast to grow at least three times by 2012, you’ve probably heard a lot about the potential benefits of cloud computing. Perhaps your company has already begun purchasing some cloud services rather than adding to its existing IT infrastructure every time a new requirement arises.
Right now there is unprecedented economic pressure for companies to find new models of working, and technological maturity is making workable solutions both possible and affordable. You may already be considering ways to make your own data center more “cloud-like” to boost efficiency, cut capital costs, and provide the elastic scaling you need to adapt rapidly to changing business requirements. How to actually go about doing this—especially where storage is concerned—may still be a bit hazy.
In this article, I begin by defining what cloud computing is and then discuss the broad requirements for storage that meet the needs of cloud infrastructure—whether that infrastructure is a “private” cloud running within your data center to meet your company’s needs or a “public” cloud providing for-fee services to a broad community.
What Is Cloud Computing?
In the broadest sense, when we talk about the “cloud” we are referring to the ability to deliver IT as a service (ITaaS). Some people define the cloud from a technology standpoint, where exact architectural models and development protocols are specified, but at the outset it is simpler to think of cloud computing as a business model for delivering IT as a service. Cloud services are the end deliverable of cloud computing and can be broken down into four categories, as illustrated in Figure 1.
A cloud can be either private: limited to the ecosystem of end users, partners, and/or customers directly associated with a company; or public: available to more or less anyone with Internet access.
Storage Infrastructure for Cloud Computing
Whether a cloud is public or private, the key to success is to create an appropriate infrastructure to deliver each cloud service, whatever it might be, as efficiently as possible. In a private cloud you may need to support a broad variety of applications, so your goal is to create an infrastructure that can flexibly allocate resources to each application as required.
It’s probably already clear that, on the compute side, server virtualization provides an appropriate infrastructure for cloud services because it allows compute resources to be efficiently partitioned and quickly allocated, increased, decreased, or deallocated as needs change. A rapidly maturing set of management services also provides speed and flexibility as well as increased availability.
Much less has been written about how to create efficient and effective storage infrastructure for cloud computing. In fact, the first Storage Networking Industry Association (SNIA) work group focusing on cloud storage was just announced at Storage Networking World in April 2009, with the objective “to identify, develop and coordinate system standards and interfaces for cloud storage.”
Since there are no approved or de facto standards in place for cloud storage, there are a few questions you should consider when evaluating new or existing storage solutions for suitability:
- Can your storage scale elastically? Similar to what you do with virtualized servers, you need to be able to allocate, increase, decrease, and deallocate storage rapidly and with a minimum of overhead.
- Can you automate storage management processes? The more you can automate regular practices such as provisioning, backup, and replication, the better your environment will scale.
- Can you meter and report on usage? In order to implement cloud services, you must understand resources a user of your service needs, have the ability to report back on actual usage, and, now or in the future, be able to bill based on resource usage.
- Can you move data freely? If your data is tied to inflexible storage, efficiency and availability will suffer.
- Can you establish multi-tenancy while guaranteeing that resources are sufficiently secure? Allowing multiple business units or separate entities to share the same storage hardware is a necessity for efficient cloud storage.
- Can you boost storage efficiency? The first step is to increase utilization; beyond that, reducing overhead, thin provisioning, and eliminating redundancy are all opportunities to increase efficiency.
- Can you efficiently protect your data? A key to cloud success is integrating all your processes such that they are simple, repeatable, and efficient. Consistent data protection and disaster recovery processes that cover every service you provide, at the appropriate policy level, are essential.
- Can you do everything on a single network fabric? The arrival of Fibre Channel over Ethernet (FCoE) makes it possible to consolidate your SANs and LANs on a single Ethernet fabric for decreased cost and increased flexibility.
- Does your storage environment support server virtualization? Assuming that server virtualization will be a key component of your cloud infrastructure, you will want storage that integrates closely with any virtualization solutions you now use or are likely to adopt in the future.
What Should You Do Now?
If you are working to evolve your existing data center toward a cloud model, you can get started by:
- Rethinking your data center design to accommodate the density and power requirements of the latest IT hardware; this topic is discussed in more detail in a companion article in this series
- Rearchitecting racks, cabling plants, and network infrastructures so that applications can be moved around dynamically to better accommodate virtualization; in most existing data centers, key business applications are boxed in for security reasons
- Virtualizing everything: servers (and possibly desktops), networks, and storage
As you move toward a virtualized storage environment that offers the same advantages as server virtualization, make sure you evaluate storage with the guidelines from the previous section in mind.
NetApp Storage for the Cloud
NetApp® data management solutions have already been deployed in a wide variety of public and private cloud environments, such as Telstra and Sensis.
Based on this experience, we continue to refine our product offerings to meet emerging cloud needs and offer a full feature set to meet the guidelines set out above.
Elastic Scaling. NetApp flexible volume technology (FlexVol®) abstracts storage volumes from underlying disks. Any storage container (LUN or volume) regardless of size is automatically spread across a large number of disks for optimal performance and can grow or shrink nondisruptively.
NetApp has already deployed tens of petabytes of storage in shared storage environments in which customers routinely scale volumes up and down as needs change.
Automated Storage Processes. For storage automation, NetApp provides a suite of management products that simplify storage operations including data protection.
- NetApp Provisioning Manager allows you to create repeatable, automated provisioning processes based on policies you define. An at-a-glance dashboard lets you monitor a variety of metrics, including capacity utilization, policy compliance, and space management statistics.
- NetApp Protection Manager provides similar capabilities to Provisioning Manager, but they are focused specifically on automation of data protection and replication.
- NetApp SANscreen® provides a real-time, multivendor, and multiprotocol service-level view of your entire storage environment.
- The NetApp SnapManager® suite of products extends the capabilities of NetApp storage to integrate with popular applications including Oracle®, SAP®, Exchange, Microsoft® SQL Server®, and Microsoft SharePoint®.
Usage Monitoring and Chargeback. By automatically correlating your end-to-end storage infrastructure against business services, SANscreen allows you to monitor service-level agreements, meter usage for cost awareness and chargeback, and proactively manage capacity to maintain optimal utilization of all resources. Provisioning Manager also includes chargeback capabilities.
Moving Data Freely. NetApp has a long history of moving data efficiently during data protection operations by using NetApp SnapMirror® or SnapVault® software, NDMP, and so on. We continue to investigate more efficient ways to facilitate data movement in conjunction with VMware and other virtualization partners wherever possible. Because data movement has been identified as a critical element for cloud infrastructure, we are focusing on this area.
Multi-tenancy. Traditionally, ensuring the highest level of storage isolation and security has meant independent hardware. Both private and public clouds need to be certain that security is as tight as it can be without sacrificing efficiency. NetApp MultiStore® software lets you create multiple, separate, and completely private logical partitions on a single storage system, so you can share storage without compromising privacy and security.
With MultiStore, storage can be provisioned, moved, and protected based on the boundaries that you define; virtual storage containers allow you to apply policies appropriate to each container that could correspond to a particular application or client of a cloud service.
Storage Efficiency. By boosting storage efficiency, you can significantly reduce your total storage requirements, which translates into savings in power, cooling, and space. Because of the flexible provisioning enabled by NetApp FlexVol technology, NetApp customers routinely achieve storage utilization rates of 60% or higher, whereas the industry average is closer to 30%. NetApp offers a range of additional technologies that further boost storage utilization, as summarized in Table 1.
Table 1) NetApp efficiency technologies.
|Dual-Parity Raid ||The double-parity protection of RAID-DP® protects you in the event that two disks fail at once, saving 46% versus data mirroring. |
|NetApp Deduplication ||NetApp deduplication identifies and eliminates redundancy at the block level. Space savings range from 25% to 55% for most data sets, up to 95% for full backups stored on disk, from 70% to 95% for virtual server and desktop environments, and up to 70% for engineering environments. |
|Thin Provisioning |
|With NetApp thin provisioning, storage is treated as a shared resource, and capacity is consumed only as it is needed. Thin provisioning can reduce your storage capacity requirement by 20% to 30%. NetApp has been working closely with VMware to provide enhanced thin provisioning functionality for virtual server environments. |
|Snapshot Technology ||NetApp provides space-efficient, nondisruptive Snapshot™ technology. NetApp Snapshot technology delivers up to 80% space savings over competing products. |
|Thin Replication ||Replication is an effective way to ensure business continuity. NetApp SnapMirror and SnapVault software perform only incremental block transfers—thin transfers—after an initial baseline, saving bandwidth and reducing the storage required for disk-based backups. If your source volume is deduplicated, the target volume will inherit the same space savings, and source and target storage systems need not be the same model and configuration. |
|Cloning ||NetApp FlexClone® technology lets you create a “virtual copy” of a data set in seconds and only consumes additional storage space as changes are made to the clone—making it an ideal technology for virtual environments that maintain many identical copies of the same operating systems. Space savings can be as high as 80%. |
Integrated Data Protection and DR. NetApp provides fully integrated and comprehensive data protection and disaster recovery based on the technologies discussed above, including Snapshot, SnapVault, SnapMirror, the SnapManager suite, and Protection Manager. These technologies offload the burden of data protection from servers to storage and provide a consistent approach to data protection across all your application services.
Unified Fabric. The emergence of FCoE makes it possible to move your data centers toward a single Ethernet fabric for all your storage and networking needs. In reality, though, you’ll probably be in transition for some time, supporting native Fibre Channel, FCoE, and Ethernet storage.
NetApp has been an Ethernet storage leader—first as a NAS pioneer and then as an early proponent of iSCSI. NetApp is the only vendor with native support for FCoE, which is a logical progression for our unified storage approach that can simultaneously deliver Fibre Channel, iSCSI, and NAS on all our platforms. By supporting FCoE we provide a simple evolutionary path for Fibre Channel SAN users to migrate to a unified fabric.
Server Virtualization Support. The items discussed in this section up to this point combine to make NetApp storage a compelling option in virtual server environments. NetApp adds to that close integration with the full range of virtualization solutions, including VMware®, Citrix XenServer, and Microsoft Hyper-V™. You can use either NetApp deduplication or NetApp FlexClone technology to eliminate the storage burden that results from storing dozens of copies of the same operating software.
Cloud computing is real and it’s happening now. Smart companies are already taking advantage of cloud services to meet many noncore IT functions. They are also evolving their internal IT infrastructures to become more cloud-like and to focus on service delivery to increase efficiency and flexibility while cutting costs.
Storage capable of meeting the needs of cloud computing infrastructures presents a number of unique requirements. To create an effective cloud infrastructure now you should focus on storage scaling, automation, data movement, multi-tenancy, space efficiency, and support for virtualization.
| || |
Senior Director of Data Center Solutions
Jeff and his team are responsible for market strategies and alliances for all NetApp data center solutions, including core systems, data center networking, manageability and data center automation software, and dynamic data center and cloud computing solutions.