User Story COOP
Coop Efficiently Increases Error Identification
The Coop Cooperative based in Basel is one of the largest retail and wholesale organizations in Switzerland. The company has around 2.6 million members. Under its own name, Coop operates supermarkets, department stores, pharmacies and much more.
Coop offers the largest variety of brands in the Swiss retail trade and is characterized by a broad ecologically and socially profiled assortment as well as sustainability services along the entire value chain. With Transgourmet Holding AG, Coop is the number two in European pick-up and delivery wholesaler. Overall, the Coop Group sales outlets are located in the most densely populated areas in the Swiss retail trade with around 2,300 stores and offers its catering customers 142 cash & carry stores in Switzerland and other European countries.
The Coop Group employs over 86,000 people, including more than 50,000 in Switzerland. In addition, the online store coop@home enables the ordering of items from the Coop range with home delivery. The deadline-driven shipments require customized, individual transport solutions and must be carried out with the utmost precision and care. After all, this fundamental link between logistics and IT is already anchored in the definition of logistics: "Logistics is the holistic planning, management, coordination, execution and control of all internal and cross-company flows of information and goods."
The central task of logistics is to manage efficient and economic material movement in logistics networks. Logistics is primarily organized and controlled via its digital mapping.
However, logistics is no longer just the domain of IT. Technological trends such as the Internet of Things (IoT), Cloud Computing, Big Data and Industry 4.0 are information technology innovation fields in which the competitiveness of production and trading companies is decided. As a link between information, production and trade, logistics plays a central role in this. It is driven by IT and is itself becoming a driver of IT.
For this reason, operational process disruptions must be reduced to a minimum. If errors occur in the IT infrastructure despite all redundancy and precautionary measures, rapid reaction is required. To ensure smooth logistics operations, 15 mobile Allegro 500s and two Allegro 1000s are flexibly deployed at Coop's 30 locations. In addition, one of the most powerful Allegro appliances, the Allegro 5300, is used centrally for analysis.
Problems with a Java application
At Coop Logistics, a Java application was not working smoothly for inexplicable reasons. This interrupted cross-company collaboration and fast automated data exchange, resulting in many delays and unnecessary waiting times. Due to the complexity and dynamics of the processes, it was debatable whether the network or another source was causing the problem. Initial research suggested that the problem was not in the network, but in the supplier's application. A clear indication of the cause of the error or proof of the source of the error was not easy to determine.
In this situation, the Allegro Network Multimeter from Allegro Packets came into play and quickly identified the problem. The mobile Allegro 200 installation took only a few minutes. The device was inserted into the network path in front of the server in bridge mode. When the Allegro Network Multimeter operates in bridge mode, all data traffic between neighbouring monitor interfaces is transparently forwarded and analyzed simultaneously. In this case, the device acts as a network bridge and there is no need to have a separate Tap (test access point) to transparently interconnect the line.
One of the many advantages of the Allegro Network Multimeter is that it can be quickly deployed to quickly isolate problems in a targeted manner. The device can also be shipped to a system administrator who can quickly install it. The network in question can then be analyzed remotely. In many cases, this saves costly and time-consuming business trips.
In this example, following the initial data stream checks, the Allegro Network Multimeter signalled long delays in the Coop proprietary Layer 7 application. The connections affected and the data transmitted over the network were extracted via packet capture (pcap). The first analysis showed that on the connections
- no retransmissions or similar and,
- only short TCP responses could be detected.
From the information obtained, the first prognosis (‘the error is in the network’) was proven to be incorrect.
For this reason, the Layer 7 protocol data streams recorded in the pcap were sent to the external supplier’s support department. Thanks to the detailed information provided by the Allegro Network Multimeter, the support engineer or the developer responsible was able to understand the problem within minutes and work out an appropriate solution. In principle, the Jumbo Frames received from the network in the ‘Socket Selector’ of the Java framework library caused the application to reset. As a result, the application had to be restarted which led to the delays recorded by the Allegro Network Multimeter.
Advantages for Coop at a glance
- Efficiency increases
- Cross-site analyses
- Easy use of the Allegro Network Multimeter in bridge mode
- Quick help and suggestions through support
Allegro Network Multimeter
Based on the data streams recorded by the Allegro Network Multimeter, it quickly became clear that the Java application always ‘hiccupped’ during the transmission of so-called Jumbo Frames, and that the Layer 7 application experienced high latency in some situations. The term 'Jumbo Frames' refers to non-standardized and oversized packets in an Ethernet network. A standard Ethernet packet is defined as 1,518 bytes in the IEEE 802.3 standard. Frames that are larger are called Jumbo Frames.
Jumbo Frames can be useful for some applications, since they reduce protocol overhead and improve efficiency. In addition, the processing overhead of the nodes involved may be reduced because fewer frames need to be processed. Such frames are not standard, and it is important that all network elements can handle these Jumbo Frames if implemented.
By updating the Java library, the problem in the Java-based Layer 7 application was quickly resolved and the Coop servers were back up and running smoothly. For Coop, the rapid error diagnosis based on the Allegro Network Multimeter meant real added-value and increased efficiency, which manifested itself in the rapid identification of errors and reduced troubleshooting time due to the tool-supported analysis of the data.
"The results of the first analyses in the Coop networks were clear after the first use of the Allegro Network Multimeter. Our investment in the product has already paid for itself in within a few weeks of operation," said Roberto Abeledo Alonso, system engineer at Coop's headquarters in Basel. "I also have to make comments about the support. The competence of the Allegro Packets support team is of extremely high quality. Not only did we get the information we needed immediately, but we also receive additional ideas for our daily work."