ptnphotof - Fotolia
For many organizations, building a hybrid cloud is at the top of the to-do list for 2016. And while deploying a hybrid cloud model is an achievement in itself, it isn't enough to just have interoperability and stability between a private and public cloud. IT teams also need to apply data management techniques to address bottlenecks and maintain hybrid cloud performance.
Hybrid clouds need a way to access data storage systems in both private and public clouds. If, for example, an organization uses the public cloud for cloud bursting, the public cloud platform needs visibility into the data objects required to run an application. This is difficult to achieve due to security and governance constraints and the limitations of current wide area network (WAN) technology. However, without addressing these bottlenecks, hybrid cloud performance will fall flat.
The first bottleneck in hybrid cloud performance is the WAN itself. Telcos, especially in the U.S., have been reluctant to acknowledge the cloud's impact on WAN. The higher bandwidth and low latencies delivered by 1 Gbps fiber are years away for many users. And, when they do arrive, they will interface to 100 Gbps or 200 Gbps backbones inside the data center, resulting in a major speed disconnect.
Apps written for local storage, such as a storage area network or even private cloud, will be throttled by public cloud access back to the "single copy" of the data in the private cloud. To avoid bottlenecks, use data management techniques, such as determining what data to keep in both clouds, and what to asynchronously update. Also determine which in-house, primary data copies, such as inventory counts and order numbers, have to be synchronously updated, with locking mechanisms for database entries.
If your data can be sharded, this reduces traffic even further, since the primary copy can be split between the clouds and localized to where compute will reside. If the shard splits are planned in advance, the outstanding transactions on the shards moving to the public cloud can be processed to provide a clean switchover and improve hybrid cloud performance.
Getting bulk data to the public cloud is also a challenge. The best approaches involve transporting old-fashioned tape or disk into the public cloud. Encrypting that data is crucial, since data will travel across a number of sites and vehicles.
Database synchronization is another challenge due to the 100x latencies of WANs compared to local area networks. Sharding the database is a good approach, as is changing the operating procedure to make the two segments of the app more independent. Examples include preallocating transaction ID numbers to each application segment, reducing the need to tap a single record in one segment for all transactions.
Proper data management techniques are key to maintain hybrid cloud performance. Aim to make public and private segments of an app independent from other segment's storage systems. While this usually involves modifying the app, the benefits of not actually moving or accessing data in real-time will pay back quickly in more satisfied users and customers.
Build a hybrid cloud migration strategy
Extend private WAN into the cloud
How to get your hybrid cloud model off the ground
Why management needs to get serious about hybrid clouds
Dig Deeper on Building and maintaining a hybrid cloud
Related Q&A from Jim O'Reilly
OpenStack Cinder has added a revert-to-snapshot function, enabling enterprises to recover from corrupted data sets. However, if the feature falls ... Continue Reading
Don't let backup data encryption fall through the cracks. When encrypting backups, key management and compression are just two of the best practices ... Continue Reading
While tape is notably offline and thus protected from cyberattacks, the cloud could comprehensively surpass it for backup if service providers figure... Continue Reading