With the implementation of China's B&R (The Belt and Road Initiative) strategy and the establishment of the AIIB. Making the best opportunity for many of the Chinese infrastructure construction giants to expand their business oversea. As a leading company in China's water conservancy and electric power infrastructure construction industry, POWERCHINA has many branches, hydropower stations and factories in Asia Pacific and the Middle East. With the digital transformation of enterprise management and production, POWERCHINA has built its own digital platform, using IT systems and tools to improve the efficiency of operations and production. However, due to the internet performance issue (jitter and packet loss), the network connectivity between global sites and the data center have become bottlenecks of the global digital platform. The low performance of the network connectivity slow down the data transmission between clients and servers, and even caused the IT service unavailable incident frequently for remote site which area without proper telecom infrastructure.
Based on the facts, POWERCHINA reevaluated the network solution, and try to find out a perfect way to solve the connectivity issues. Leasing Line (MPLS) was an option, but also has some limitations:
1. Overseas sites are mostly located in geographically remote areas. Local telecom operators don’t have sufficient resources, and leasing line implementation is very difficult, costly and time consuming. Many sites will be relocated or rebuilt with the progress of the construction project, and the contract period of the dedicated line is more than 12 months. Both of the cost and the flexibility are not matched with the business needs.
2. There are no full-time IT staff at overseas sites, and local staff do not have the competence to maintain professional network equipment (switches/routers).
3. The leasing line (MPLS) can only provide a consistent network performance (very small jitter and low packet loss rate), but the traditional transmission network does not have the TCP optimization function. Under high latency conditions, the transmission efficiency of TCP could be limited. For example, the latency of MPLS between Shandong and Dubai is about 200ms, and the maximum default value of the TCP window of the customer's server is 64KB, then the maximum transmission bandwidth of the single TCP session is 64KB/200ms*8 = 2.56Mbps. In POWERCHINA’s OA system user case, OA user needs to upload large files from client to the OA server every day, the MPLS has a limited performance improvement on this OA user case because of lacking TCP optimization function.
For the TCP window and its scaling principle, please refer to: http://packetlife.net/blog/2010/aug/4/tcp-windows-and-window-scaling/
According to the customer's pain points, we propose a solution which use AXESDN NaaS as the networking solution to quickly deploy the connectivity between global sites and the data center, then activate the AXESDN TCP optimization, network monitoring and management features on AXESDN Cloud for all the other network functions that customer need.
Network topology
Deployment lead time: hours
1. Before the AER is delivered to the customer site, AXESDN will work with the customer to finalize an installation solution and pre-configure it in both the cloud and AER.
2. When the local customer receives the AER, they will receive an email and also a hard copy about the “AER Installation Guide”. Local customer does not need to have an IT maintenance background. Just following the instructions to complete the connection for power cable and the network cable.
3. After AER is powered on and the network cable is connected, the installation and configuration work has been completed theoretically. For further configuration and maintenance activity, AXESDN can do remotely.
Site relocation: hours
AXESDN supports on-demand billing for both bandwidth and duration. During the relocation process, the network services of the new site and the old site can work in parallel. Customer can decide when to disconnect the old site depend on their relocation progress, and don’t need to set a deadline to relocation process from network service contract perspective, because the old site can be billed per day.
TCP optimization
Since the customer needs to maximize the bandwidth utilization of a single TCP session, we turned on TCP segmentation optimization on AXESDN Cloud. For “Segmentation optimization”, please refer to the following picture:
Unoptimized TCP Session
Optimized TCP Session
AXESDN TCP Performance Optimization technical method:
o TCP retransmission optimization
o TCP ACK mechanism optimization
o TCP congestion control optimization
o TCP real transparent proxy
Implementation result:
Comparison of Network performance indicator (Latency/Packet loss)
User console: Monitor and management
Comparison of transmission efficiency
The following transmission efficiency comparison is based on the actual data transmission between the customer’s Qingdao site and the Bahrain site with following conditions: (The internet is unstable, so the transmission efficiency of long-term testing on internet will be much lower than the following results)
· Qingdao-Bahrain Internet delay 300ms, no packet loss
· Qingdao-Bahrain AXESDN delay 200ms, no packet loss
· Qingdao site Internet bandwidth is 100Mbps
· Bahrain site internet bandwidth is 10Mbps
· AXESDN allocates 4Mbps for Bahrain site access
· Transferred file size is 50MB
As the figure, under the condition that the network end-to-end latency is 200ms, With AXESDN TCP optimization service, the file transmission can reach the cap of the bandwidth subscription (4Mbps)
Customer feedback