Kenneth Li Kai, Director, Enterprise Marketing & Solution Sales Dept, Huawei Malaysia
Data&StorageAsean: In your own words - what is Object Storage?
Kenneth: People might not realize it, but object storage is something we use in our day to day lives. When you post updates and share photos on social networking sites, all that information or data needs to be stored somewhere. Object storage refers to the way in which we manage and work with units of storage known as objects.
In essence, an object is a piece of data accompanied by metadata. The parameters that cover metadata can include information like how old the piece of data is, when it was last accessed and what size it is. Fundamentally those two pieces, the metadata and the data together makes an “object”.
Data&StorageAsean: What are the major use cases for Object Storage solutions?
Kenneth: Enterprises are realizing that their file systems in the long run will no longer keep up with the growing needs of performance and capacity. File systems are a good fit for smaller installations but with the millions of files beings stored, the system will begin to slow down. There are several reasons why an enterprise should move towards an object storage solution.
Scale
Large volumes of unstructured data is a prime use case for object storage, because it can handle huge volumes of data and its associated metadata. The bigger and more unpredictable the volume, the greater the use case for object storage is for you. With an object storage system in place, this will drastically reduce your costs of storing such large amounts of data and can increase your abilities to search the data for multiple purposes.
Management
As the amount of data continues to grow, it makes sense that a new approach is required to effectively manage all this data with the available resources. An object storage system offers different, simplified manageability. Object storage manages a namespace, not just a rack of storage. In this scenario, a namespace could be a rack of storage, or multiple racks. It could also be local, or it could be geographically dispersed. This is real productivity and provides a broader view of your storage resources.
Consistency
With the use of object storage, data is less likely to get lost or misplaced. Object stores offers a well-structured safe semantics which guarantees for a wholesome user experience. This in turn increases data integrity and durability at scale.
Data&StorageAsean: Should Object Storage be on premise or in the cloud?
Kenneth: Every enterprise should have choices.
Between on premises and cloud based solutions it comes down to the specific needs within each enterprise. The decision to go with a cloud, on-premises or hybrid, or hosted solution should be made only after reviewing the unique requirements of your business.
It is important for enterprises to learn the advantages and limitations of the architecture in order to get the most value for businesses.
Data&StorageAsean: What are some of the key features people should look for when choosing an Object Storage solution?
Kenneth: There are many key feature requirements people look for based on different scenarios. From our experience, some of these features may include but are not limited to:
Flat Architecture
Performance and capacity expanded linearly with the hardware nodes increased.
High Reliability
There is an assurance of reliability in both the system level and at the data center level. For the system level, object storage should provide the flexible data redundancy protection strategy, such as erasure coding, and rapid data recovery capability; for the date center level, the A-A disaster recovery features based on storage is latest requirement.
Strong Consistency
This is especially useful in big data scenarios. It is necessary to support the strong consistency rather than eventual consistency for more accurate big data analysis.
Data&StorageAsean: Is there anything unique about your company’s object storage offering(s)?
Kenneth: Huawei’s distributed object storage is one kind of storage resource of FusionStorage. It supports Amazon S3 and OpenStack Swift protocols, using HTTP as data transmission protocol which is very suitable for the new applications, such as web mobile, IoT and big data scenarios. It is a fully distributed cloud storage platform, which is built on the Huawei COTS X86 server architecture to meet the block storage, file storage, and object storage requirements, meeting structured, unstructured, and semi-structured data access requirements in terms of IOPS, bandwidth, and mass expansion.
FusionStorage is designed in a way that the performance increases linearly with the capacity. FusionStorage is OpenStack-native storage, and its control plane is based on OpenStack, which can interoperate with other open architecture.
Customers can also use Huawei’s distributed object storage to construct cross-DC active-active disaster recovery solution for their public cloud, and can also provide DR Service to their end users. This is the largest active-active cluster in the industry, which can be scale-out up to 4096 storage nodes, also RPO=0 and RTO=0.
New storage architecture designed for the new applications, such as big data analysis. Big data applications requires the newest data for the accurate analysis. Huawei’s product can be integrated with Hadoop with S3 interface, and optimizes S3a adapters for high read performance and ensure that the big data system can get the newest data stored in the Huawei distributed object storage. This is the result of the innovative architecture which uses strong consistency rather than eventual consistency strategy for data store. In the Hadoop scenarios, some old public cloud providers often use the latter strategy for data store.
Our other unique product features include:
Tiering and lifecycle management: data moves from hot tier to warm and cold tier automatically based on policy.
Flexible data redundancy strategy with high capacity usage and the first real active-active DR solution cross two data centers.
Huawei distributed object storage uses online Erasure Coding for data redundancy protection. Compared with the 33%-50% capacity usage of other products using multiple copies as strategy, the capacity usage of Huawei distributed object storage can be up to 94%, which thanks to using online Erasure Coding redundancy protection.
0 Comment Log in or register to post comments