Shulou(Shulou.com)06/02 Report--

So-called edge computing

The concept of edge computing Edge Computing comes into being. In 2014, the European Telecommunication Standards Institute (ETSI) established the Mobile Edge Computing Specification working Group (ETSI Mobile Edge Computing Industry Specification Group) to promote related standardization work. In 2016, ETSI extended this concept to Multi-access Edge Computing (Multi-Access Edge Computing,MEC), taking into account the scenario requirements of FMC (fixed network / mobile convergence). In April 2016, 3GPP SA2* also formally accepted MEC as a key technology of the 5G architecture. The simple understanding is that in the data source (client side). ) and the ability to calculate at the edge of the system. According to Cisco, the background of edge computing, namely fog computing, is to solve the problem that more and more applications are migrating to the "cloud", relying on a powerful centralized control system to centrally process washing data, as shown in the figure. However, with the increase in the amount of data, our computing power needs to be greatly improved. Therefore, the rise of edge computing is different from the traditional centralized thinking. His main computing nodes and applications are distributed in the data center near the terminal, which makes the service response and reliability higher than the traditional centralized cloud computing concept. According to the definition of ETSI, multi-access edge computing is on the edge side of the network close to people, objects or data sources, and provides edge intelligence services nearby through an open platform that integrates network, computing, storage, application and other core capabilities to meet the key needs of industry digitalization in agile connection, real-time business, data optimization, application intelligence, security and privacy protection.

Specifically,

Edge computing is different from the traditional centralized thinking, its main computing nodes and applications are distributed in the data center near the terminal, which makes the response performance and reliability of the service higher than that of traditional centralized cloud computing. Specifically, edge computing can be understood as a computing program brought by the edge near the data source.

If you use a more general term, that is, proximity computing or proximity computing (Proximity Computing)

The data center that distinguishes cloud computing is closer to the data source in the core network, and the edge computing data center is far away from the data source (high number of network hops). From the schematic diagram of Cisco company above, we can find that compared with the terminal devices, it takes layers of devices to reach the core network, and the devices at the edge layer are in this path, on the contrary, some cloud computing is centralized big data processing, and edge computing can be understood as edge big data processing. The difference is, only this time, the data can be solved on the edge side instead of being sent to the remote cloud. Edge computing is more suitable for real-time data analysis and intelligent processing, compared with simple cloud computing is also more efficient and secure edge computing and cloud computing are actually a way to deal with big data's computing operation. More accurately, edge computing should be a supplement to cloud computing and optimize the application practice.

Originally, because the project requires us to use K8s for cluster management, the pain of installing K8s is vivid, and the performance cost and computer performance requirements of K8s limit my imagination. It is undeniable that K8s is unique in cluster management, but it is still a long way to be applied to other fields. However, thanks to rancher's new product, K3s, which is 5s less than K8s, it has a high position on K8s. It brings infinite possibilities for edge computing.

Edge computing is deployed through K3s because the product is so lightweight that it claims to require only 40m of hard disk space and 512m of memory to deploy

Next, we will experience the practice of ultra-lightweight cluster control technology through a series of deployments. At present, I only use two nodes to deploy the system, one is the cluster, the other is agent. These two nodes are docker containers with a specification of 4g 4cpu. Download the operating software of K3s with the following command

$wget https://github.com/rancher/k3s/releases/download/v0.3.0/k3s

Connecting to github.com (192.30.253.112)

Connecting to github-production-release-asset-2e65be.s3.amazonaws.com (52.216.112.11)

K3s 100% | * | 38.7m 0:00:00 ETA

As it describes, the system is only 38.7m. Modify the operation property to gain execution permission, and then run it in the background.

$sudo chmod + x k3ssudo. / K3s server &

The following is the run log:

[1] 247

INFO [0000] Preparing data dir / var/lib/rancher/k3s/data/56c346dbd23e9399b9ccb062cc8d71110f96eed4cd8d138a7c097e4c651d12a6

[node1] (local) root@192.168.0.23 ~

$INFO [2019-04-13T08:14:05.448051228Z] Starting k3s v0.3.0 (9a1a1ec)

INFO [2019-04-13T08:14:06.398250769Z] Running kube-apiserver-watch-cache=false-cert-dir / var/lib/rancher/k3s/server/tls/temporary-certs-allow-privileged=true-authorization-mode Node RBAC-- service-account-signing-key-file / var/lib/rancher/k3s/server/tls/service.key-- service-cluster-ip-range 10.43.0.0 advertise-address 16-- advertise-port 6445-- advertise-address 127.0.0.1-- insecure-port 0-- secure-port 6444-- bind-address 127.0.0.1-- tls-cert-file / var/lib/rancher/k3s/server/tls/localhost.crt-- tls-private-key-file / var/lib/rancher/k3s/server/tls/localhost.key-service-account-key-file / var/lib/rancher/k3s/server/tls/service.key-service-account-issuer K3s-api-audiences unknown-basic-auth-file / var/lib/rancher/k3s/server/cred/passwd-kubelet-client-certificate / var/lib/rancher/k3s/server/tls/token-node.crt-kubelet-client-key / var/lib/rancher/k3s/server/tls/token-node.key

INFO [2019-04-13T08:14:08.761712689Z] Listening on: 6443

INFO [2019-04-13T08:14:09.608524354Z] Node token is available at / var/lib/rancher/k3s/server/node-token

INFO [2019-04-13T08:14:09.608562854Z] To join node to cluster: K3s agent-s https://172.18.0.6:6443-t ${NODE_TOKEN}

INFO [2019-04-13T08:14:10.774070472Z] Connecting to proxy url= "wss://localhost:6443/v1-k3s/connect"

INFO [2019-04-13T08:14:09.718944649Z] Run: K3s kubectl

INFO [2019-04-13T08:14:09.719024250Z] K3s is up and running

INFO [2019-04-13T08:14:09.769336884Z] Logging containerd to / var/lib/rancher/k3s/agent/containerd/containerd.log

INFO [2019-04-13T08:14:10.779764787Z] Handling backend connection request [node1]

From the log, we can find some interesting information, such as

Node token is available at / var/lib/rancher/k3s/server/node-token We are going to find the path to token To join node to cluster: K3s agent-s https://172.18.0.6:6443-t ${NODE_TOKEN} this instruction that will be added as agent And this IP:172.18.0.6 is due to the network isolation between mirrors caused by the namespace (namespace) isolation of docker container I used. in order to make the traffic interconnection, port 6443 has been opened as a listening port K3s kubectl from IPNAT to 192.168.0.23Listening on: 6443. You can run the kubectl command line waiting for node node1 CIDR not assigned yet, and the management node is ready to wait for agent to join.

All right, let's use the following command to find token

There is no node access at present.

But don't worry, we are about to connect to the new device:) in our agent node, we still have to download the K3s software in the form of wget

Verify the connectivity of the two nodes

Add the node to the server side $. / K3s agent-s https://192.168.0.23:6443-t K10dff6042746f035482131dcc04299b24b2aa06b801585519d49a8741e164f95a7::node:246085e87b9e0ff4e85cfc4d5bf7cea6 using the following command

Similarly, we can still see the process of agent joining step by step from the log.

Let's look back at the server log to clearly reflect the process of joining agent and the statistical results after joining.

The addition of images is simple and convenient. With the help of docker technology, we can quickly manage images. For example, according to the Internet of things scenario, installing agent on IOT devices will collect some environmental parameters according to business requirements. Although it will face a large number of nodes in IOT environment, it will not hinder our management of these nodes. The minimalist server design completely enables us to monitor data on mobile phones. It can even be said that our mobile phone is a monitoring center of the Internet of things computer room. With the development of 5G environment, this design brings us unlimited possibilities. Just like the picture below, it is an eye-catching application in Shenzhen Machinery Exhibition.

Calculate the edge of the data through massive sensors, and then upload it to the cloud service to analyze the phenomenon. If our docker image computing is fast enough, we will be more comfortable in dealing with emergencies. In large Internet enterprises, this kind of monitoring center architecture is very common, but there is still a certain distance from the specific landing to the industrial sector. With the help of edge computing, we can completely get through the last kilometer and realize industry + Internet.

It is not difficult to find the innovative potential and excellent value that edge computing provides to industry participants and the general public: end users can get a better and more personalized application experience through it, whether they connect to the network or access to content. Operators will open up their enhanced network capabilities to third-party OTT providers or application developers through edge computing platforms, and provide their applications and services to mobile users, enterprises and vertical industries. Operators are also expected to redefine their role in the entire industry chain, thereby achieving new revenue, providing more high value-added services, and opening up new market opportunities. OTT providers and independent application developers can also quickly develop new applications and shorten the development cycle on the open standard platform of edge computing, creating higher and faster returns for themselves while providing end users with an extremely fast experience with near zero latency. Edge computing will become the combination of cloud service providers and operators, where countless cross-border innovations are brewing the explosive power of "blockbuster".

Hey, hey, the future is promising.

Reference:

Edge Computing Research report, Hong Xuehai × × Guo Shusheng, Information Technology Strategy Research Center of Institute of Computing Technology, Chinese Academy of Sciences. What is Edge Computing: https://makerpro.cc/2018/12/what-is-edge-computing/ faces 5G Edge Computing, Intel leads Intelligent Network Transformation, Intelk3s official website: https://k3s.io/docker Environment, play with docker: https://labs.play-with-docker.com/

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.