December 2010

 

Inside This Issue

When Does Hybrid Cloud Storage Make Sense?

By Joel Christner

Many organizations are considering use of hybrid cloud storage solutions, cloud storage gateways, or cloud storage on-ramps to help address cost associated with deploying storage to satisfy the never-ending hunger of today's users and applications. However, most do not know when a hybrid cloud storage solution makes sense. "Which applications should be used with cloud storage?" some may ask, or "Which should I avoid?"

Cloud storage is provided as a service that is accessible over the Internet, generally using HTTP-based RESTful APIs, SOAP APIs, or in some cases file protocols. This presents a few challenges in terms of applicability of cloud storage, which can be addressed by use of a hybrid cloud storage solution, specifically:

• Integration—hybrid cloud storage solutions speak standard protocols to your servers and manage communications to cloud storage services using provider APIs so that you don't have to. Further, they provide volume management, thin provisioning, snapshots, and other data management and protection capabilities.

• Performance—with deduplication and compression, hybrid cloud storage solutions will minimize bandwidth consumption while also minimizing cloud transfer and storage chargers. Additionally, integrated tiering that identifies working set data and hotspots can optimize performance by keeping working set data in local SSDs.

• Security—through integrated encryption, hybrid cloud storage solutions improve security and minimize data confidentiality and data privacy concerns.

The capabilities and features of hybrid cloud storage solutions overcome many challenges with using cloud storage and, therefore, make them a very good fit for many applications. Particularly, hybrid cloud storage solutions are a good fit when:

• The application has a discernable working set—that is, the application works with a slowly changing subset of all application data. This is common for applications where the data itself is unstructured (even if it's stored in a structured database).

• The application data is able to be deduplicated and/or compressed; in general, performance will be directly related to the ability of the hybrid storage solution to deduplicate that data (less data transferred over the Internet, faster response times, and so on).
These characteristics are common for applications that are high-growth and centered around user productivity, collaboration, archival, data protection, and long-term storage. Specifically:

• Email, including Microsoft Exchange 2010—email is able to be deduplicated, and the working set is centered around the emails, attachments, calendar entries, and other items that users interact with (send and receive) on a daily or weekly basis. People tend to work with email at the top of their inbox more frequently.

• Collaboration and document sharing, including SharePoint 2010 and 2007 —web-centric document libraries are experiencing tremendous growth in terms of adoption which is driving massive storage capacity increases. With version control (each version stored individually as a new file) and recycle bins (allowing users to do their own restores) more information is being kept around longer.

• User home directories—typically users share the latest content they've been working on, and the IO requirement for this data is very low.

• Libraries and archives—business-critical or project-centric information is stored for long-term retention, which is generally not accessed very frequently, and does not require the performance levels provided by today's traditional storage.

• Backup data—backups are generally only accessed when a restore needs to happen and are generally very redundant, especially if full backups are taken frequently, or the content being protected is user-centric where multiple versions of the objects are stored and then protected.

Many companies have started testing and deploying hybrid storage solutions including public cloud storage services for the above use cases and are finding that they are able to address capacity growth requirements without compromising on performance, security, or data protection, while also being able to dramatically lower their costs.

Joel Christner is the chief scientist of StorSimple. Prior to StorSimple, he held technical leadership positions at Cisco Systems, solidifying the company's end-to-end strategy for application performance management and WAN optimization. He also published three books on the topics of WAN optimization and application acceleration. He holds a number of industry certifications (CCIE, MCSE, others) and has a Master of Science in Computer Science from Columbia University.

Print Article

Three Keys to Showing Business Value from Real-Time IT Monitoring

There are three main parts to the solution of how to demonstrate concrete business value from real-time IT monitoring. The challenge is to clearly convey how IT departments can show real value.

Read more

The 3 Biggest Mistakes In Data Center Operations

How can you avoid making major mistakes when operating and maintaining your data center? The key lies in the methodology behind your operations and maintenance program.

Read more