Hybrid cloud offers benefits of hosted applications and offloading manual tasks but beware latency and outages in mission-critical functions.
Hybrid cloud storage has evolved in recent years to become a deployment methodology that blends on-premises storage infrastructure with public and hosted cloud services.
The road to hybrid cloud will not be short or easy for many organisations, and many will have to make detours along the way to fully exploit the rapid innovation of hyperscale cloud storage environments.
The workload and data mobility of hybrid cloud storage strategies will allow infrastructure and operations (I&O) professionals to hedge their cloud bets and switch providers when terms or capabilities no longer meet expectations, or when cloud storage providers offer a new “can’t miss” service.
Enterprises have treated cloud storage resources as complementary storage silos to support next-generation analytics and systems of engagement cloud applications, and offload menial unstructured storage burdens.
But cloud storage has much more to offer. Next-generation hybrid cloud storage technologies are already available and ready to bridge the gap between your on-premises applications and resources and cloud storage services by simplifying workload mobility and cloud federation between service providers.
Drawbacks to mass cloud storage migration
The rapidly falling price of cloud storage services is pushing organisations to migrate away from expensive conventional storage systems, toward cloud storage services. While cloud storage should absolutely be a part of your future storage infrastructure, there are a few limitations that should be kept in mind.
The untimely collapse of Nirvanix forced its clients to hastily move away from its cloud storage services, shocked early adopters and highlighted the need for cloud federation and data migration services. Long-distance data migration is still a struggle for enterprises, particularly those with large workloads on the scale of hundreds of terabytes to petabytes.
The inconsistent network connectivity commonly found in remote sites exacerbates this. Data transfer across internet protocol (IP) networks can take a significant amount of time when data payloads go beyond 100 terabytes. The cost of outbound bandwidth and time associated with migrations make cloud migrations difficult and effectively lock in clients.
Cloud security is still the top impediment to full-scale cloud service adoption, although the steady stream of improvements cloud providers are making go some way in addressing this objection. Enterprise storage teams may be hesitant in handing over control of their data to cloud providers, using security and compliance concerns to slow down cloud deployments.
Strategies to mitigate data latency
While the physics limitations of rapid data movement cannot change, progressive enterprises and service providers have come up with clever strategies to put computing and storage resources closer together. A handful of colocation and datacentre providers – such as Equinix, CoreSite, and TelecityGroup – have facilities either physically close to a cloud service provider or house an instance in their datacentre.
This proximity allows customers to bypass the internet and connect directly to their cloud service providers, (often for a flat monthly rate, via fibre-cross-connections at port speeds of 1Gb/s to 10Gb/s), which can provide sub-millisecond latency.
Amazon’s Direct Connect emerged in 2011 as a means of accelerating data movement, and its partner Equinix claims that Direct Connect can transfer files to Amazon 138% faster than public internet connections.
Clouds, and the workloads they house, are getting larger and more business-critical on a daily basis. Beyond raw scale, you must leverage new strategies and technologies and super charge your hybrid cloud storage strategy to keep pace with innovation.
Henry Baltazar is a senior analyst at Forrester Research.
Register now to receive ComputerWeekly.com IT-related news, guides and more, delivered to your inbox.
Related content from ComputerWeekly.com
RELATED CONTENT FROM THE TECHTARGET NETWORK
This was first published in March 2015