Primary data storage options for the cloud
A comprehensive collection of articles, videos and more, hand-picked by our editors
A RESTful API is the native access to object storage and considered the most efficient way to ramp onto object...
storage. However, it sometimes isn’t practical to modify an application to use object storage RESTful API, and it makes sense to use software options, such as conversion or gateway software, as laid out in a previous tip. There is another category of systems to consider when researching appropriate alternatives to RESTful APIs: cloud-integrated storage systems. This technical tip takes a look at how cloud-integrated storage can help facilitate object storage -- and when it’s appropriate.
Cloud-integrated storage (CIS) is really a variation of third-party ISV conversion, or gateway software, discussed in a previous tip. In simplest terms, that category of software is gateway software that is packaged, delivered and shrink-wrapped as a hardware storage system. However, these products have evolved beyond just being gateways to cloud storage -- which, in simplest terms, is object storage. Many have become just as capable as traditional SAN, NAS or unified storage systems specifically for primary or active data. These CIS systems are fundamentally similar to hybrid tiered storage systems because they use an object storage tier for secondary data, older data and snapshot data.
CIS systems make object storage completely transparent to any application accessing the CIS as NAS or SAN storage. The application is never aware that its data has been stored in the object storage. CIS delivers both advantages and disadvantages. Looking at the CIS and the object storage as a complete storage ecosystem, the CIS plays the role of the cache for the most frequently accessed data. The object storage becomes the storage for secondary, passive or infrequently accessed data.
There are a few reasons you would want CIS. For starters, CIS delivers native SAN or NAS storage performance. It deduplicates, compresses and encrypts the data before moving it to an object storage tier. This reduces the amount of capacity required for the primary active data residing on the CIS as well as the older data, secondary data and snapshot data residing on the object storage. This concept of CIS object storage tiering reduces both capital and operating expenditures.
Another advantage of CIS comes from multi-CIS system accessibility to the data stored on the object storage. It requires access via the same vendor CIS and the correct encryption key. The result is an outstanding disaster recovery (DR) tool, especially when working with cloud service providers utilizing object storage as their cloud storage offering. All data moved via CIS to that cloud object storage in addition to the server images are capable of being recovered at the cloud service provider’s data center in the event of a disaster. That’s huge. This capability facilitates much faster and lower-cost DR. For example, TwinStrata clearly emphasizes this capability for both its Cloud Array appliances and Cloud Array VM. Avere Cloud NAS and Panzura Quicksilver Cloud Storage Controller also make this a selling point of their CIS.
Another advantage is a global namespace for geographically distributed CIS NAS systems such as those from Avere and Panzura. This enables content and workflow sharing. One other CIS advantage is its ability to take advantage of multiple different vendor object storage systems and public cloud (object) storage systems. This can be critical if the object storage or cloud storage provider were to go belly-up as Nirvanix did this fall. Working with multiple object stores allows data to be easily migrated from one provider to another.
But CIS has disadvantages too. Any data moved to an object storage system via a CIS system can only be accessed and retrieved through that specific vendor’s CIS. If a different CIS is desired, the data must be recalled out of the object storage to the originating CIS, then migrated to the new CIS, and finally uploaded back to the object storage. As stored data grows into multiple petabytes, this becomes a non-trivial, massive, time-consuming undertaking.
Not all CIS can perform as well the more common SAN, NAS or unified storage systems. Some CIS systems work with a limited number of object storage systems (or, in some cases, just a single vendor’s system). It is critical to make sure any CIS system selected is certified and working with the object storage system selected, preferably in a production environment.
Then there is the data migration to the CIS. Remember, the CIS system looks, feels and acts just like any other SAN, NAS or unified storage system. This means the data residing on current installed storage systems must be migrated to the CIS system, which is again a non-trivial exercise.
CIS systems are available from Avere Cloud NAS, Ctera cloud storage gateways, Microsoft StorSimple, Nasuni (Storage Infrastructure as a Service), Panzura Quicksilver Cloud Storage Controller, Riverbed Whitewater and TwinStrata Cloud Array.
The CIS qualifies as the simplest way to move data to and from object storage. It looks, feels and acts like current storage in the data center while transparently moving data back and forth from the object storage -- just as if it’s a low-cost storage tier. There are excellent advantages as well as severe disadvantages (such as vendor lock-in) to the CIS system approach. Just remember: It is just one of several equally effective ways of on-ramping and off-ramping to object storage.