From the hot spot a few years ago to the seemingly cold one now, the market has quieted down a lot about SDN. But does this mean that it is on the wane?

Apparently not.

After a series of processes such as concept prototype, technical verification, and then return to production and value investment, SDN has truly stepped into the stage of commercial application. The focus of enterprises’ investment in network has shifted from hardware to software, and SDN has successfully promoted the network transformation. IHS Markit predicts that revenue from software-defined networks (including SDN hardware switches, SDN software controllers, etc.) will reach $15.8 billion by 2022. So this is a market with great commercial prospects. Virtual networks have played an important role in IT infrastructure, both public and private clouds, and Overlay has been widely used and deployed. However, although SDN improves the flexibility and programmability of network deployment, and reduces network investment and operation costs, it also brings new problems. Compared with physical network, IT operation and maintenance personnel for virtual network monitoring and management become difficult.

The monitoring problem of virtual network

Specifically, the virtual network challenges faced by enterprises are mainly reflected in two aspects: one is difficult to collect network status and traffic data, and the other is difficult to detect faults in advance and locate them quickly.

First of all, from the perspective of collection, the traditional operation and maintenance tools applied to the physical network cannot directly collect the traffic in the virtual environment, which leads to the blind spot of the virtual network traffic found by enterprises. In addition, in the cloud environment, the access and dependency relationships of containers, virtual machines, hosts, access, core, and network services are complex, making it more difficult to achieve fine-grained traffic discovery and analysis. As a result, IT is difficult for IT managers to perceive changes in the network. Co-founder and COO, spruce network sources in an interview with intellectual society described the change of the physical network to the virtual network: the physical network like highways between cities, the use of cameras or other sensing device can easily perception and analysis of traffic information, and the virtual network is turned into strips of abstract logic roads, traditional device is difficult to collect for traffic.

Secondly, from the point of view of fault location, it is precisely because of the acquisition obstacles of network data that network operation and maintenance problems are directly caused. For operation and maintenance personnel, the connection state, resource usage state and policy state of virtual network are a “black box”, and network operation has no way to start. Therefore, when the network failure occurs, the operation and maintenance personnel are difficult to quickly find and locate the problem, thus affecting the business.

To sum up, since the birth of virtual networks, the monitoring problem has been with us today, and with the further application of hybrid cloud, the challenge becomes more and more obvious. The Spruce Network DeepFlow® product is the first light in the “black box” of virtual network traffic that companies need to open, control, and deliver reliable, quality network services to their businesses.

Deepflow ® three steps to light up the cloud network black box

Spruce Network Deepflow ® is a product for traffic collection, visualization and analysis of virtual networks. It breaks the current situation of lack of monitoring and management of virtual networks in China. There are also products that look similar to Deepflow ® in the industry, such as operations tools for network vendors, third-party acquisition and analysis tools for NPM vendors, and data acquisition tools for cloud vendors. However, according to the sources, they all have their own limitations or focus in different directions. For example, the cloud itself is a pooled resource facing large-scale scenarios, and traffic acquisition, processing and analysis need systematic planning; Centralized processing is bound to face bottlenecks in a cloud environment; It is not enough to analyze only from the perspective of network; Enterprise hybrid cloud involves multiple public clouds and private clouds in multiple data centers. The overall network state needs to be visualized and can be associated with the mutual visit relationship of businesses and services.

DeepFlow ® can help companies in a variety of cloud and heterogeneous resource pool construction traffic data platform, unified collection and distribution network traffic between virtual machines, realize the key link of the overall business performance monitoring, and provides end-to-end path all virtual network diagnosis, solve the enterprise cloud in the virtual network monitoring, operational, security and other problems.

In addition to a comprehensive network traffic collection capability that supports a variety of cloud environments and deployment configurations, Deepflow ® is also characterized by its ability to preprocess telemetry data, packet data, and network flow data. The Spruce Network’s proprietary high-performance algorithm enables operator preprocessing. Through filtering, de-weighting, compression, truncation, encapsulation and other pre-processing methods, the transmission bandwidth of the collected information is reduced and the performance is improved.

In terms of traffic distribution, the DeepFlow® open API interface enables a variety of third-party tools to connect and send traffic, including NOC/SOC, intelligent operation and maintenance platform, business analysis, security analysis, audit tools, as well as specific tenant and specific traffic screening, thus providing a more comprehensive decision basis for enterprises. For the overall flow collection and distribution, what makes DeepFlow® special is its distributed collection computing capability, efficient pre-processing and unified management. “DeepFlow® provides the collection abstraction capability for various resource pools in the hybrid cloud, and establishes a time-series database to achieve high-performance time-series data reading and distribution capability. And then have the ability to build the entire cloud traffic platform, systematic management.” Source emphasis.

Based on this, Spruce Network can help enterprises to clearly draw cloud network panorama and business portrait from multiple dimensions, and provide performance analysis, security policy verification, retrospection analysis, and network diagnosis function between any two network elements, so as to provide support for stable operation of the business. The three steps of collection, distribution and analysis solve the difficulties of virtual network monitoring. After enterprises have accurate and comprehensive perception of virtual network, the last big obstacle of cloud in business is eliminated.

One supervision, spruce boost automation network upgrade

Of course, Deepflow ® is not the whole picture in the Spruce Network’s product planning. Based on a deep understanding of virtual networks, Spruce Networks can do much more.

The Spruce Network Services Platform (NSP) is a Network interconnection and service Platform, which is different from Deepflow ®. “NSP focuses on providing a unified virtual Network abstraction layer for hybrid cloud and cloud Platform, enabling virtual Network to extend across resource pools and available domains. It avoids the virtual network being limited to the physical network of a single brand, and is a “super” network controller for the construction of customer-oriented cloud data center network. Deepflow ® focuses on the construction of virtual network traffic data platform for hybrid cloud, providing a unified abstract layer for traffic collection for operation and maintenance system, and providing customer-oriented cloud operation guarantee. “For a data center network, NSP is the control and DeepFlow® is the monitoring. Together, the two enable a closed loop of operation and management of the entire virtual network.” “The source said.

Furthermore, NSP virtualizes the resource pool network and provides the cloud with the capability of hybrid orchestration for multi-resource pool and heterogeneous networks. In addition, it also provides a series of east-west and south-north data center network boundary service capabilities such as virtual firewall and load balancing. By integrating NFV and traditional network functions, NSP meets the business requirements of unified management of heterogeneous network resource pool, fast and flexible networking and flexible capacity expansion.


The ultimate goal of the Spruce Network is to enable cloud data center networks to be self-optimizing and dynamically aware, with both Deepflow ® and NSP monitoring and interacting. To that end, the NSP continues to iterate on technology and functionality, with major upgrades on the horizon. All in all, the combination of Deepflow ® and NSP provides more imagination for the development of the cloud network, whether to see the cloud network more clearly or to make the cloud network more free. It gives new vitality to SDN, the network transformation technology.