Consuming and realizing the benefits of a public cloud compute resource is straightforward for many organizations. Even organizations with major security concerns have found consuming public cloud virtual machines (VMs) advantageous.
For instance, enterprise data center managers can offload services such as test and development to cloud services without too much concern. The cloud consumer realizes the agility and cost advantages of ephemeral cloud instances.
It’s a more daunting challenge to replicate the success of cloud compute services in enterprise storage services. Cloud storage providers hope to change that perception amongst customers.
Data isn’t fungible
Amazon recently introduced a data analytics product. On the surface, the solution is logical; Amazon has the compute capacity and services to analyze big data. The challenge is getting the data to the AWS data center. Amazon’s solution is to ship a physical storage device and have customers copy data to the device and ship it back to Amazon. Some joke it’s the AWS equivalent to Amazon Prime.
The very idea of putting data on a physical device and shipping it to a cloud provider highlights the concerns of many security-sensitive organizations. Consultant and IT analyst Justin Warren describes the challenge this way: “Data isn’t fungible like compute.”
Warren implies that instantiating multiple copies of data off-premises has impacts and concerns that don’t exist for ephemeral compute instances. From a technical perspective, it’s physically prohibitive to move large amounts of data offsite. From a security perspective, having multiple copies of enterprise data in the mail creates a long list of concerns.
For these reasons, it’s been difficult to adopt cloud storage as a primary option for enterprise workloads. For customers with a large enterprise data center footprint, a basic requirement is that data is local to the majority of resources. Generally, the industry has answered the challenge with two approaches: a cloud storage gateway, and local storage that is managed by a cloud provider. Here’s my overview of a solution from each category.
Zadara Storage: Managed local storage
Zadara Storage provides a scale-out storage solution owned and managed by Zadara within customer data centers. I consider Zadara a private cloud service.
The pricing structure and management are the major differences between Zadara and solutions from enterprise storage vendors. Traditional storage vendors require a long-term license commitment and customers to manage the hardware. The model is similar to most public cloud offerings with customers paying for the services consumed vs. the size of the cluster. The commitment cycle for the amount of storage consumed is shorter as well.
Zadara manages all aspects of the hardware. If the appliances begin to run low on space, Zadara manages the expansion of the storage cluster. The interface for provisioning storage is web based, and Zadara provides a multi-tenancy capability.
If a service provider or a customer wants to carve up a physical cluster for multiple service offerings, Zadara abstracts the overhead from the end user. For example, a customer could create three Virtual Private Storage Arrays (VPSAs) and offer completely separate SLA and administrative domains to internal customers.
An additional capability is the option to create copies of the data in AWS. The AWS backup allows for AWS-based disaster recovery. The base solution is hosted within a customer’s data center.
Zadara may appeal to security-conscious organizations, because the data never leaves customer premises.
ClearSky Data: Cloud gateway
ClearSky Data is a gateway provider that leverages public cloud and local gateway devices that offer caching. ClearSky stands out because it uses point of presences (POPs) to reduce the impact of latency to retrieve stale data. Most solutions on the market offer gateway devices, which provide access to AWS hosted data. These solutions cache data locally on the gateway devices, improving performance for frequently accessed data.
ClearSky reduces latency by providing and managing high-speed links between customer data centers and their POPs. When stale data needs to be retrieved, the request is serviced via high-speed WAN links to POPs residing in the same metro area. The end user sees a huge increase in performance vs. a solution that relies on just cloud gateways and remote cloud data centers. As expected with any cloud service, ClearSky charges by usage vs. available capacity.
Security may be a concern for some customers, because the data does leave the customer’s premises. ClearSky does offer encryption as a controlling mitigation.
Ready to move to cloud-based storage?
ClearSky and Zadara offer solutions based on the technical challenges and security concerns of their customers. Both companies hope the technical workarounds for performance and security lure customers into moving data to cloud services.
Do either of these approaches entice you to go with cloud-based storage? Share your thoughts in the comments.