Industry Trends Analysis
Business Challenges
-
High Costs on IT Infrastructure and Maintenance
Conventional big data service operations require huge initial investment on equipment room construction, storage servers, application servers, and utilities. Meanwhile, the service scale and platform architecture will require upgrading to match business growth, resulting in more heterogeneous resources and difficulties in O&M.
-
Long Analysis Cycles, Poor Performance
For traffic bursts during hotspot events or promotional activities, a conventional data center cannot expand storage and computing resources quickly enough to handle service surges. As a result, big data and analysis tasks accumulate, and processing cannot be completed before the maximum tolerable time, resulting in business losses.
-
Waste of Capacity and Efficiency by Offsite Data Analysis
Big data analytics require excellent capabilities to deal with concurrent data reads and writes, which is unachievable by conventional storage servers. Therefore, data is usually copied to local computers and then analyzed. Such conventional practices demand extra local storage space and degrade analysis efficiency.
-
Waste of Resources Due to Inflexible Provisions
Data is the critical production material for businesses. Companies often spend heavily to purchase extra storage space that is several times larger than the data volume to store data in redundancy (more than three copies) and prevent data loss or damages.
Typical Scenarios
-
Real-Time Big Data Analysis
-
Offline Big Data Analysis
Introduction
BigData Pro provides high-performance computing resources with flexible scalability, low-latency, and high throughput, compatible with the industry's mainstream platforms for real-time analysis services. Together with the large-bandwidth network and multi-protocol OBS, BigData Pro greatly improves your efficiency in resource utilization for real-time analysis services.
Advantages
Decoupled deployment of computing and storage resources allow for maximized utilization efficiency. Auto scaling of computing resources meets ever-changing service needs.
BMSs are connected to OBS over the network with ultra-high bandwidth (25 Gbit/s), guaranteed to meet your requirements for low-latency and high bandwidth. The solution achieves 2× higher performance than that of on-premises big data analysis platforms.
Introduction
Computing resources are decoupled from storage resources to reduce waste. Scalable storage eliminates the waste of reserved storage space. Multi-protocol storage capabilities greatly reduce data replication operations.
Advantages
Computing and storage resources are auto scalable separately. Therefore, you do not need to reserve storage space or bind computing resources to storage resources.
Object, file, and HDFS protocols are supported, facilitating data ingestion and access, and eliminating ineffective replications.
BMSs can use the InfiniBand technology to achieve high throughput and low latency, which ensures ultra-high performance of offline big data computing.
Cross-AZ disaster recovery of distributed storage allows data durability to reach 99.9999999999% (12 nines).
Solution Advantages
-
Efficient Resource Utilization
By deploying computing resources separately from storage resources, BigData Pro delivers 75% and 50% higher efficiency in cluster and storage utilization, respectively. Auto scaling is also available to dynamically match computing resources with changing service loads.
-
Active Data Reuse
Object Storage Service (OBS), integrated in the BigData Pro solution, supports various protocols, allowing you to access your data stored in OBS from various terminals. You can also perform analysis tasks online. With active data reuse, your services require less time for processing.
