Private cloud storage, also called internal cloud storage, is a service delivery model for storage within a large enterprise. Internal cloud storage runs on a dedicated infrastructure in the data center, offering the same scalability benefits of public cloud storage to corporate departments and partners while addressing security and performance concerns.
Private cloud storage usually supports a single tenant in an enterprise's local data center. For larger organizations, there may be multi-tenancy features to isolate cloud storage access by office locations or department.
Scalability requirements for private cloud storage are more modest than its public cloud storage counterparts. That's because the "customer" base for private cloud storage is usually limited to groups inside the organization, while public clouds have potentially millions of customers using their services. This also means private cloud storage will more likely leverage and be built upon traditional data storage and IT infrastructure.
What it's used for
Despite the convenience and advantages of public cloud storage, the model isn't suitable for everyone and all data types. Organizations may choose or be required to keep certain data on site for legal, compliance or security reasons, for example. Private cloud storage addresses these issues by providing many of the advantages of cloud technologies while keeping data in-house in an enterprise's local data center.
While the features and technologies used for private cloud storage and provisioning remain the same as public cloud storage, the relationship between provider and customer changes. A corporate division or business unit is the customer in the case of private cloud storage, while the company's IT department is the provider rather than an off-site third party.
Building private cloud storage vs. working with a vendor
Like a public cloud, a private cloud should be delivered as a service. Users shouldn't have to be concerned with how additional data storage capacity is delivered when they request it. As with other types of cloud computing models, private cloud storage should be elastic, multi-tenant, available on demand, and provide detailed reporting and billing.
Elasticity refers to the ability to increase and reduce consumed resources as needed. Clients -- usually business users in the case of private cloud storage -- should also be able to access these resources on-demand with little or no manual intervention from storage administrators or others in IT using a delivery-as-a-service distribution model.
A delivery-as-a-service distribution model defines a set of services abstracted from the underlying physical hardware. For IT, this may mean updating service catalogs to focus more on metrics and not just hardware -- line items such as hard disk drives (HDDs) and flash, HDD speed and so on -- as they have traditionally. These metrics would include specifications like latency, data availability, throughput, resiliency and IOPS per terabyte of storage or I/O density.
Users should also be able to serve themselves, meaning they don't need to go through IT when requesting more data storage capacity. They can sign up for more storage and set up their own volumes through a dashboard with no administrator required.
Multi-tenancy enables a cloud to support multiple clients -- departments, divisions, offices and sometimes individuals in the case of private cloud storage -- securely with an assured level of performance. Users are prevented from viewing and accessing each other's data. In addition, the same level of consistent service is guaranteed irrespective of system load using features like quality of service (QoS). Delivering on security and performance is just as important with private cloud storage as it is with public cloud storage.
Detailed reporting and billing charges clients based on consumption over time by measuring storage utilization on a granular level. With private cloud storage, this means the ability to report and possibly charge against individual departments, business areas or teams. Some organizations may decide not to execute a billing component of their private cloud storage because a way to charge individual business units for usage may not exist or be necessary.
Private clouds can also charge more for premium services, such as flash storage as a performance tier instead of a base tier of HDDs.
It is important to understand how storage will be used when building a private cloud storage infrastructure. There's a significant difference between storage used for managing virtual machine (VM) instances and storage used for data, for example.
The type of storage media and storage file systems figure into a private cloud deployment strategy. Internal cloud architects weigh cost versus performance. You can, for example, save money by using commodity hardware rather than the specialized systems traditional storage uses. And because private cloud storage requires more horizontal scalability than traditional storage, it should let organizations more easily and transparently scale capacity as needed.
Marc Staimer, founder and president of Dragon Slayer Consulting, explains the difference between private clouds and storage virtualization, and the best types of physical devices for private clouds.
Highly scalable object stores are more efficient than file and block storage for application data in clouds, particularly for archive and backup. A retailer might store catalog data in object storage and VM images in block storage, for example. That's because private cloud object storage isn't suitable for transactional databases where nanoseconds count.
A compute cloud infrastructure and a private storage infrastructure are different concepts that often -- but not always -- go together. You can build a private storage cloud independent of, or in conjunction with, compute, depending on your needs. Meanwhile, IT shops that don't want to connect and test individual components for their private cloud storage can turn to pretested converged and hyper-converged bundles that integrate storage, compute and networking.
The orchestration framework that provisions resources based on customer requests is an important part of private cloud storage. Examples of these frameworks include proprietary tools such as Microsoft Azure Stack and the VMware vRealize Suite cloud management platform. There are also open source platforms like Apache CloudStack and OpenStack. Docker Swarm, Kubernetes and Mesosphere DC/OS offer open source tools that go part of the way toward enabling a private cloud implementation as well. And suppliers like Platform9 and ZeroStack optimize a few of the tools mentioned here.
Careful consideration about orchestration platform and supportability can help lead users to the right storage product or supplier for their private cloud storage. Older and legacy platforms with no native automation abilities will be more difficult to integrate.
An example of a storage platform that does offer a native API to support integration from the common provisioning platform is NetApp SolidFire.
You can also go the scale-out storage route and use open source or software-defined storage. OpenStack is a well-known open source software platform for building clouds with support for object, block and file storage.
Others include StorPool (block storage), StorageOS (for containers only), Dell EMC ScaleIO, HPE StoreVirtual and VMware vSAN. With these, you can deploy storage as part of a server instead of having dedicated storage software. VMware vSAN integration is directly supported within vSphere to enable policy-based provisioning of storage for virtual instances, for example. Hyper-converged vendors such as Nutanix bill their model of delivering software-defined storage on commodity servers as a way of building private enterprise clouds.
Other important components for a successful private cloud storage operation include:
- Automation and service. This removes most, if not all, requirements for a storage administrator to manually provision storage resources by enabling users to do it themselves through API and command-line interface frameworks.
- Scalable design. Ability to scale storage resources up or out without affecting service, performance or I/O operations.
- Multi-tenancy. Requirements here include security to prevent hosts from accessing each other's resources, maintaining consistent performance whatever the workload of each individual host or application, and QoS to guarantee steady performance regardless of system load.
- Management software. Robust management software that forms part of a wider solution integrating compute and networking. Plug-ins to virtualization platforms or CLI/API wrappers that expose operations like provisioning and reporting often deliver storage features to private cloud storage.
Private cloud storage vs. public cloud storage vs. hybrid cloud storage
By moving cloud storage into an organization on a dedicated infrastructure within its own data center, private cloud storage addresses the uneasiness many organizations have about the level of data security and performance offered by public cloud storage providers. It is more suitable to active data and data enterprises require greater control over.
Private cloud storage provides for higher levels of reliability, performance and security. Costs are also fairly predictable and lower over time than deploying and maintaining a traditional storage infrastructure. On the flipside, it requires internal resources and staff to maintain and is far more limited in terms of scalability compared to public cloud storage.
Suitable for unstructured data that's not continually changing, public cloud storage is a service owned and operated by a provider. It supports multi-tenancy by isolating and securing access to data for each of its many clients. Public cloud storage is commonly used to provide storage capacity for non-core application data, for example, as well as archiving for backup and disaster recovery purposes. Amazon Web Services, Microsoft Azure and Google Cloud Platform are the best known public cloud services.
While public cloud storage delivers a high level of scalability and requires no on-premises storage infrastructure, performance and security can vary greatly between service providers. Also, reliability, unlike with private cloud storage, is at the mercy of internet connectivity and service provider availability. It does offer a pay-as-you-go model, however, so users only pay for the storage capacity they need. While public clouds are reliable, there have been well-publicized incidents of them going down or data breaches often due to user error. And smaller public clouds run the risk of going out of business, leaving their customers responsible for moving their data off of the failed cloud provider's infrastructure.
Hybrid cloud storage allows organizations to store unstructured data for archiving and backup, for instance, and less sensitive data with a public cloud service.