"The Huawei OceanStor series unified management and disaster recovery solution has provided us with an advanced, complete service solution. Not only have the devices far surpassed our expectations, but through comprehensive testing, we've determined that its disaster recovery mechanisms can completely ensure continuous operation of our core services. The Huawei Storage solution has solved the difficulties our operations were facing and completely satisfied our service requirement. I believe it was exactly what we were looking for."
— Liu Jingjie, Manager, China Unicom (Shanghai) Wo-Store O&M Center
China Unicom Network Communications Group Co. Ltd., (China Unicom) possesses a telecommunication network that connects China to the world, has proactively promoted broadband deployment in both fixed and mobile networks, and has provided a full range of services and high-quality ICT services to its enormous customer base. Recently it created the "Wo" brand-name, which includes "Wo-3G," "Wo-4G," and "Wo-Residential," accumulating over 400 million users. Its role in the progress of ICT deployment both economically and in terms of social impact cannot be underestimated.
The "Wo-Base" core system primarily consisted of "Wo-Commercial," "Wo-Gaming," billing, big data, database, and other services, that employed Linux, Unix, Oracle and SQL server systems. At the time, all of its many services are deployed on low-performance storage equipment purchased years ago, while the storage performance of the download engine system was already sorely insufficient. Key databases (primary databases, billing database) had yet to receive data protection planning, and because device acquisition and deployments were sporadic and management was complex, CAPEX and OPEX remained persistently high.
"Through years of product development and service operation, we at the 'Wo-Base" possessed a faithful community of users. Because the number of our users had been growing steadily, our service system concurrent access requests increase daily. The devices purchased years ago were increasingly unable to handle user requests. We understood that along with improvements to hardware processing capability and core storage technology, storage devices capacity and performance had taken enormous leaps forward, accompanied by the development of new data protection technologies like snapshot, data mirroring, and remote disaster recovery. As a result, we'd had a long-standing interest in reconstructing the IT storage system at the 'Wo-Base" by purchasing high-end storage equipment that can provide new technologies, and constructing a system that could grow along with our core service development in the future. Of course, in addition to equipment with stable architecture and advanced data protection, we were also closely watching our CAPEX and OPEX during this reconstruction process," said Liu Jingjie, Manager of the Wo-Store O&M Center.
After their analysis, Mr. Liu Jingjie's O&M team concluded that a storage system with the capability to process over 1-million concurrent requests and provide local disaster recovery was essential.
At the same time, because the older storage system was built in multiple stages, the creation of "data islands" in each system was difficult to avoid and device utilization rates were lackluster. Each service system used independent storage devices, which although on one hand was beneficial in preventing services from affecting each other, also made data sharing and analysis services difficult if the devices were located far apart. In addition, managing the variety of different devices was difficult, while expansion and maintenance complexities led to persistently high TCO for the O&M department.
"We were facing serious challenges using these technologies. Besides the weak processing capability affecting our services, expansion and upgrade of our existing devices was extremely difficult. Also, we'd started to migrate some service systems to a virtualized platform, and wanted to continue to virtualize other services going forward. Our current storage system hadn't passed through cloud environment verification and couldn't deal with the necessary expansion. Because of this, building a new storage system with a virtualized environment and industry-leading architecture, which was also expandable and easily managed, had become an issue we needed to face," said Mr. Liu.
After the "Wo-Base" decided to purchase a new storage system, it began to consider the issues of data security and ensuring service continuity. Because China Unicom's "Wo" platform had such a large user base, with online users numbering in the tens of millions per day, the importance of its user account information data, configuration data, and private data was paramount. How to securely migrate all of this data from the old system to the new one while ensuring zero data loss was a problem that required careful consideration, as well as pre-operation testing and verification of the data migration solution to be employed.
Early in the second half of 2013, Mr. Liu Jiangjie's O&M department opened communication and discussion with all relevant solution suppliers in the hope they could provide a complete, mature storage solution that would not only fulfill their service requirements, but was cost effective, and could ensure absolute security and service continuity.
Huawei created an 'OceanStor series centralized management and disaster recovery" solution for the customer that targeted the issues they were facing. Based on Huawei solution, one OceanStor 18500 and two OceanStor S5500T devices were deployed in the "Wo-Base" equipment room. The main database and downloaded engine data was centralized on the high-end OceanStor 18500, along with two sets of both Oracle RAC production system and system data. In the OceanStor 18500 system, the SmartQoS function ensures the quality of key services, while the HyperReplication data synchronization software was used to automatically replicate key service data from the OceanStor 18500 to one of the S5500Ts. These techniques help ensure real-time availability of service data and removed the danger of important data not having a disaster protection plan, which had been a concern. Later, another NAS device could be added to change the existing file server and array service model of the download engine system into a unified, directory-based, file-level NAS service model. Offline protection was provided for the download engine system to integrate the online disaster recovery system into a complete multi-tier data protection solution.
"Through test results we discovered that the OceanStor 18500 was able to effectively solve the problem of slow concurrent reads/writes during peak access periods that we'd been experiencing with its SmartTier data-tiering feature. The algorithm automatically sent frequently accessed data to SSDs for processing, and if the number of access requests fell, it would automatically migrate this data to SAS disks to allow new hotspot data to utilize SSD resources. The ability of Huawei storage's automatic data tiering ability to improve services and data read/write efficiency during peak hours was obvious, even better than we had previously thought," said Mr. Liu.
"In addition, our download engine contained almost ten million small files that varied in size from a few KB up to a few MB. Previously, our storage system's ability to process access to these files was far from ideal and disk utilization rates often hit 100%, meaning that end users had a poor download experience. We migrated out download engine data to the Huawei OceanStor 185000 for testing and discovered that download speed increased by 6-7 times without reaching 100% disk utilization."
"We verified the efficiency of the synchronous data replication between the OceanStor 18500 and S5500T multiple times to simulate its recovery ability after a supposed service outage and discovered that their RTO and RPO met our earliest requirements for service operation continuity, ending the problem we'd had for a long time of a lack of disaster protection for core data."
A second OceanStor S5500T, specifically targeting virtual environments was deployed and optimized, providing storage services in the "Wo-Base" service system virtual platform, completely fulfilling virtual platform application requirements for fast read/write.
This solution was unique from solutions provided by other vendors in that it didn't assume that the highest hardware specification was most ideal, but instead was tailored to the customer needs. By analyzing the situation from the customer's point-of-view, and helping the "Wo-Base" face not merely its inability to satisfy the customer technologically, but also its problems of cost-effectively planning for the future and data protection, the solution delivered a strong price-performance ratio, and won the "Wo-Base" O&M team's approval.
Because of the important services that the "Wo-Base" provides, to guarantee safe operation, Huawei created a meticulous data migration solution for its client, and to date has accomplished the migration of 200TB of data without a single occurrence of data loss or affect on service continuity.
Today, the "OceanStor series unified management and disaster recovery" solution that Huawei proposed and implemented has already shown its great benefit. Not only has its solved the problems of no disaster recovery, difficulty in providing virtual service deployment, and high-investment and O&M costs that had plagued the "Wo-Base," but also meets the needs of the base's future development. The benefits that the "Wo-Base" has reaped by deploying a Huawei data storage solution are clear: