Honam Petrochemical Corporation (HPC) of Korea wants a more robust disaster-recovery system for its warehouse database servers. The company is concerned that line failures or natural disasters could block database access and force it to halt
production and shipment of its products. HPC has determined that SQL Server 2012 AlwaysOn functionality can mitigate risks of data outages. The same technology can also improve system efficiency through load balancing.
HPC, based in Seoul, Korea, operates large petrochemical facilities at complexes in several locations, including Yeosu and Daesan. Products—such as polyethylene and polypropylene—are stored at the plant in automated warehouses, each of which has 30,000–40,000
||With SQL Server 2012 AlwaysOn, we can make our existing disaster-recovery configuration more robust simply by using basic database functionalities without needing to purchase a separate backup and recovery solution.
| Daewon Kang
Information Strategy Team
Inventory, orders, and shipping are tracked and controlled by a warehouse management system (WMS) based on the Microsoft Windows Server 2008 R2 operating system. When the sales team enters a delivery order into the WMS, the system quickly locates the inventory
in the warehouses and determines which warehouse is most efficient to ship from.
Failure of the WMS and its database can have expensive consequences. If inventory data is lost, a visual inventory of a warehouse can take a few months. If products cannot be located or shipped, production might have to be delayed. After a day-long data
outage, a plant might have to shut down, costing the company huge amounts of money.
Because a database outage is potentially so expensive, HPC has developed an extensive disaster-recovery system. Each plant has its own WMS application server, and each of these servers is linked with a dedicated line to the company’s main database in Seoul.
HPC has mirrored the Seoul database on a backup server in Daejeon so that data could be immediately recovered in the event of a database failure in Seoul.
Despite these disaster-recovery efforts, HPC is still concerned about long-term database disconnections caused by an unexpected natural disaster. “If there is a failure in the integrated database, it can be recovered through the mirror server located in
Daejeon,” says Daewon Kang, Assistant Manager, Information Strategy Team, HPC. “But if the local plants and the main data center were disconnected by a flood, there is nothing we can do about it.”
While investigating how to further strengthen its disaster-recovery system in late 2011, HPC heard from Microsoft Korea about the AlwaysOn functionality that is built into Microsoft SQL Server 2012 data management software. HPC learned that AlwaysOn can
enhance mirroring capabilities by supporting use of multiple synchronous secondary databases to back up data on primary servers at the plants.
“SQL Server 2012 AlwaysOn functionality enables primary and secondary servers to be placed not only in the Daejeon center, but also at many plants,” says Mr. Kang. “With SQL Server 2012 AlwaysOn, we can make our existing disaster-recovery configuration more
robust simply by using basic database functionalities of SQL Server 2012 without needing to purchase a separate backup and recovery solution or using other backup methods like log shipping.”
The company also determined that it could use AlwaysOn technology to synchronize geographically distributed servers at the plants with the integrated database at the main center in Seoul. “SQL Server 2012 allows configuration of multiple secondary databases,
which is ideal for the WMS application servers that access the WMS integrated database in Seoul,” Mr. Kang says.
In addition, HPC discovered that AlwaysOn technology could improve WMS load balancing. The WMS system is used by about 500 people nationwide to enter and view data. HPC determined that about 90 percent of system usage is simply for viewing data, not for
entering it. Since data views don’t require changing data, they could be performed on each site’s active secondary servers. Performing data views on local secondary servers would reduce loads on the WMS integrated database, which would improve WMS performance.
Satisfied that AlwaysOn functionality offers many benefits, HPC decided to implement SQL Server 2012 Enterprise. “We plan to apply it to our existing WMS environment in 2012 and are considering expanding it to other business systems,” Mr. Kang said.
With SQL Server 2012 AlwaysOn, HPC expects to reduce the risk of plant shutdowns, simplify disaster recovery, and improve load balancing.
Less Risk of Plant Shutdowns
HPC expects that SQL Server 2012 AlwaysOn will help assure business continuity by avoiding the database outages that could cause plant shutdowns. “WMS downtime of more than one day forces the entire plant to be stopped,” Mr. Kang says. “In general, the cost
of a plant shutdown is estimated to be more than 3 billion won” (U.S.$2.5 million). “The failure of WMS could make that occur,” he adds.
Simpler, More Robust Disaster Recovery
SQL Server 2012 AlwaysOn functionality will make the WMS database recovery procedure simpler and more robust. In the past, if a failure of the main WMS database caused data loss, related information would have to be entered manually—a lengthy process. But
with AlwaysOn disaster recovery, use of multiple secondary servers should prevent any need for manual entry in case of data loss. That’s because the main database would be synchronized to servers in the Yeosu and Daesan plants—not just in Daejeon—so if servers
simultaneously shut down or line failure occurred, data could be recovered immediately.
Efficient Use of Network Resources
With AlwaysOn load-balancing improvements, HPC will be able to use network resources more efficiently. The company will be able to offload 90 percent of the network’s load onto the active secondary servers that would also be used for disaster recovery. That
would free up bandwidth on the main WMS database server.
The increased WMS capacity could be important for future growth. “If the production scale increases or a new plant is constructed, warehouse demand will also increase, which makes the issue of increasing system capacity come to the fore,” Mr. Kang says.
This case study is for informational purposes only. MICROSOFT MAKES NO WARRANTIES, EXPRESS OR IMPLIED, IN THIS SUMMARY.