Cloud providers like Microsoft Azure, Google Cloud Platform and Amazon AWS built their storage infrastructures on an object storage foundation that offered almost unlimited scale at extremely competitive prices. Object storage vendors adopted these architectures and made them more enterprise suitable, which gives the data center a choice between an on-premises object store and a public cloud storage provider. The decision between on-premises and cloud is not actually an either-or decision; IT professionals look to leverage the best attributes of both, but creating this hybrid cloud storage model presents challenges of its own.
The Advantages of a Hybrid Cloud Storage Strategy
Having object or cloud storage available, both on-premises and at an external cloud provider, has numerous advantages. For DR, as an example, object storage systems can automatically replicate data between on-prem and cloud site, making DR management simple and cost effective.
The cloud also has data analytics and data management capabilities that many organizations are looking to leverage, often comparing multiple external providers as each has different strengths. Sometimes the strength may be as simple as a pricing advantage, other times it may be a compute capability like video transcription or context indexing. Ideally, organizations want to move data sets seamlessly between providers to take advantage of those strengths without having to update applications to the new location of the data.
On-premises object storage has its strengths: it’s fast, fully in your control, and incurs no access charges. Many data centers will find on-premises object storage less expensive for long-term retention of large data sets. Others will want to, or even be required, to keep certain data sets out of the public cloud. Object storage systems can support legacy protocols like NFS and SMB so those applications and data can move into the object storage model.
The Problems of a Hybrid Cloud Storage Strategy
There are numerous problems with creating this seamless hybrid cloud vision. The first is that each cloud has to be managed as its own separate silo of storage and data has to be manually moved to that silo when its attributes are required. This manual movement also means that applications have to be manually updated to point to the new location of the data. There are also API limitations, while most providers today claim some S3 compatibility; there are nuances between the implementations, limiting mobility. Since most organizations also have multiple data centers, each can justify its own object storage system. Integrating these systems together is also important.
Most providers try to overcome these challenges by replicating data between the various organizational locations and external locations. But, replication, while needed for disaster recovery, is not what the multi-cloud use case requires.
The Hybrid Cloud Storage Requirement
Hybrid cloud storage requires that data be moved seamlessly between locations. There are data management solutions that attempt to lay a global file system across these different storage types and locations. The problem is these file systems create a lowest common denominator effect, and they add expense to the solution. Additionally, each location is still a separate silo of storage; it’s just managed centrally. An alternative is to extend the object storage model to support these differing providers.
A multi-cloud object storage system, instead of requiring the implementation of a separate file system, extends the existing file system. With a multi-cloud object storage system, there are no separate silos of storage; it is all one system regardless of location. Instead of the separate systems at each location, each location offers nodes within the same global cluster. The organization sees a global storage cluster and can move data between these nodes based on policy.
A single object layer also ensures API compatibility since each location is actually running the same object storage technology. It also means that if the multi-cloud object storage system supports NFS and SMB, then those protocols are now fundamentally multi-cloud as well.
The use cases and justification for on-premises object storage and external cloud storage will vary from organization to organization. But, almost all organizations will need to leverage the multiple locations. For example, some organizations will want to move data to the cloud temporarily for processing and then bring it back on-premises for long-term storage. Others will want to control the growth of on-premises storage and use the cloud for long-term retention. The advantage of a multi-cloud object storage system is that it fulfills the needs of both use cases.
To learn more about using multi-cloud storage register for our on demand webinar, “Cloud Storage vs. On-Premises Storage“, and learn how the new world of multi-cloud storage coupled with scalable on-premises storage can:
- Achieve limitless scalability and seamless capacity expansion
- Enable unified data management across clouds and on-premises
- Consolidate object and file data to a single storage environment
- Reduce costs and eliminate complexity