buchachon - Fotolia
Many small to medium-sized businesses have moved large parts of their IT infrastructures to the cloud, but larger organizations have been slower to adopt the strategy other than the use of a few software-as-a-service products. The cloud is less compelling to these data centers because they already have infrastructure and personnel in place, as well as a much larger footprint. An enterprise hybrid cloud model that lets these organizations leverage their on-premises infrastructure and the cloud would be a better approach. While a hybrid cloud is not perfect, the following steps should help data storage administrators get the most out of an enterprise hybrid cloud initiative.
Step 1: Know your infrastructure
Before leveraging the cloud, an enterprise must understand its existing infrastructure, specifically how much data it has and the amount of data it should place in the cloud. There are two competing factors when making this assessment:
- The time it takes for a file to be transferred from/to the cloud. The impact of this will vary by the size of the file accessed and the speed of the cloud connection. If occasional access to relatively small files is all that is needed, the cloud can be an ideal storage area. If constant or occasional access to large files is required, the cloud may not be an ideal storage area.
- Where the cloud access occurs. For example, cloud latency is more significant if the cloud is accessed from an on-premises device. Latency may not be an issue if access is from within the cloud itself (a cloud compute instance accessing data already in the cloud).
The infrastructure assessment boils down to understanding what the organization has and how that data may be accessed in an enterprise hybrid cloud storage scenario.
Step 2: Leverage cloud strengths
You need to leverage what the cloud is good at versus what the organization does well. The cloud is particularly good at temporal things, for example, instantiating a dozen virtual machines (VMs) to process a specific data set and then spinning those VMs down. The cloud is weakest at more permanent things, such as storage. A need to store petabytes of information in the cloud for a long period of time becomes expensive, as the same amount of capacity is re-purchased every billing cycle. For smaller organizations with lower capacity requirements this is less of an issue, but for large enterprises it may be cheaper for them to build their own private cloud than to use the public cloud.
Step 3: Know your hybrid
The most critical part of an enterprise hybrid cloud strategy is the component that will interface between the data center and cloud. Most often this component will come in the form of an appliance or VM, but in some cases it may be an agent that is installed on every server in the environment. The value of an appliance is that it offloads the translation from on-premises to cloud, but it can also be a bottleneck because all data needs to go through this single device. The upside of an agent is that the design is more scalable because all the servers can access the cloud at the same time; the downside is that the servers need to have the agent installed on them.
Step 4: What will you hybrid?
The final step is deciding what to put into the cloud. For many organizations, backup is their first initiative. But this contradicts the challenges related with long-term, high-capacity data storage in the cloud. The cloud is ideal for backup, but for enterprises, a best practice is to keep only the most recent backups in the cloud. This data is what is most needed in the event of a disaster.
Long-term backup data and archives can also be stored in an enterprise cloud. The hybrid appliance acts as a local mount point and cache for the cloud. But the problem of long-term, high-capacity storage again becomes an issue. The organization needs to do the math on what the five-year to 10-year cost of that storage will be. The organization's current capacity plus its growth rate also need to be factored in. The amount of capacity paid for in Year 10 may dwarf what is paid for in Year 1.
In my experience, most enterprises with more than a few petabytes of storage find that using cloud-like technologies (object storage systems) internally is a more cost-effective approach than putting data in the cloud for long-term storage.
Another potential use case is cloud-based application development and testing in the cloud, with production on-premises. In this use case, the hybrid appliance is put in the cloud and data from the data center is cached in the cloud so application development and Q&A teams can test on near-production copies of data. When code is found to be stable, it can then be migrated into the data center for full-scale production.
A hybrid approach to the cloud makes the most sense, but due to the typical size of a large-scale data center, there are considerations around total cloud cost. Most organizations should be able to use the cloud in some way, but it will likely be more temporal in nature where a job is scaled up and then scaled down quickly. That said, most large organizations should be able to borrow cloud concepts such as object storage and apply them in their internal data centers.
Backup, cache top list of enterprise hybrid cloud use cases
Successful hybrid cloud implementation rests on automation, monitoring
Appetite for hybrid cloud in the enterprise to grow significantly
Hybrid cloud scenarios get support from SQL Server 2014