Data centers often need to handle tens of thousands or even hundreds of millions of user requests in seconds during the “618” & “Double 11” shopping festivals and the “red envelope rain” interactions on the Spring Festival. Such a huge challenge requires access support with extremely low latency, high throughput, and high stability from back-office mission-critical systems.
The access requests and data volume faced by mission-critical applications will increase rapidly, and may even explode in a certain period of time. The underlying storage system should present stable performance, high reliability, and low latency. The impact of occasional performance jitter on the business may be amplified layer by layer, which will result in I/O timeout, and eventually affect user experience.
The cloud-based Business Support System (BSS) needs to display high performance, high concurrency and low latency, and be easy to extend and maintain. In the customer’s new IT system, the core system of BSS is deployed by adopting the non-shared disk array to build the database cluster—MySQL Cluster and by constructing a cluster access network with X86 server + PBlaze Series NVMe SSD.
The distributed solution based on MySQL database can improve the performance of the cluster via multiple nodes. The data can be stored in multiple copies to ensure the security.
NVMe SSD is used to improve the I/O capability of server nodes and reduce the dependence on memory capacity.
The TCO is reduced by over 50% compared with that of traditional IOE architecture.
Currently, there are nearly 1 billion short video users in China. As a strong performer in the short video industry, the customer builds a high-performance resource pool with thousands of PBlaze Series NVMe SSDs in its Hadoop distributed architecture, which not only guarantees the experience of every user, but also provides sound support for HFT businesses like red envelope.
User scale and activeness surge to new highs constantly. The new storage architecture allows every user to experience real-time response to video sharing and communication, red envelope collection and other operations.
Compatibility with different components of the system is achieved, giving full play to the advantageous high scalability of Hadoop distributed architecture.
High-quality customer services provided by Memblaze not only facilitate the smooth construction and deployment of the customer’s system, but also help realize stable operation during the application, which saves a considerable amount of operation and maintenance costs for the customer.
Increasing data requires a quicker analysis system and a faster decision-making system, as well as a disaster recovery system with faster response in backup and recovery operations. While enhancing system performance, maintenance costs need to be cut down as well. In the customer’s new solution, a new integrated database machine is built with the x86 server, IB network, and high-performance PBlaze Series NVMe SSD, which effectively addresses the above challenges.
The business performance and business simulation computing performance of the data warehouse have been improved in multiples. Particularly, the simulation computing ability based on business demand has been enhanced by up to 7 folds.
The time for backup and recovery of the whole database is shortened within 1/6 of the previous time.
The deployment of new integrated database machine no longer requires purchasing traditional microcomputers and SAN storage system, reducing the overall purchase and maintenance costs by 50%.