CN103618759A - Cloud service-based remote desktop computer wake-up method - Google Patents

本发明公布了一种基于云服务的远程桌面计算机唤醒方法，利用云服务供应商提供的在线云租赁服务，利用云服务维护内网中有限数目中介台式计算机（agent）的持续网络连接，当唤醒时通过中介计算机向目标台式机发送带有目的台式机MAC地址的Wake-on-Lan数据包进行唤醒操作。本发明提供了一种不需要增加新的硬件设备、不改变企业工作的环境并且在较少的资金花费和较少的CPU及网络带宽开销的情况下，能够有效的远程唤醒企业或大型组织内部台式计算机的技术，从而节省企业能源。台式机在空闲时能够自动睡眠并且在需要时被顺利唤醒。

The invention discloses a method for waking up a remote desktop computer based on a cloud service. The online cloud leasing service provided by a cloud service provider is used to maintain the continuous network connection of a limited number of intermediary desktop computers (agents) in the intranet. At this time, the intermediary computer sends a Wake-on-Lan data packet with the MAC address of the target desktop to the target desktop to perform a wake-up operation. The present invention provides a method that does not need to add new hardware devices, does not change the working environment of the enterprise, and can effectively remotely wake up the interior of the enterprise or large organization with less capital expenditure and less CPU and network bandwidth overhead. Desktop computer technology, which saves business energy. Desktops can automatically sleep when idle and wake up gracefully when needed.

一种基于云服务的远程桌面计算机唤醒方法A remote desktop computer wake-up method based on cloud service

技术领域technical field

本发明属于计算机网络技术领域，特别是涉及一种基于云服务的远程桌面计算机唤醒方法。The invention belongs to the technical field of computer networks, in particular to a method for waking up a remote desktop computer based on cloud services.

背景技术Background technique

当前台式机仍然占据着企业设备中的主要部分。由于远程办公的需要，用户有较大的需求在非办公时间或地点对位于企业内的台式机进行远程操作。为了达到这个目的，台式设备经常在闲置时仍然处于开机工作状态，耗费了大量的电能，同样增加了企业电力开支。为了能够避免这种情况，能够让台式机自动睡眠并且在需要的时候能够远程唤醒已经关机或睡眠台式机的系统就显得尤为重要。Today's desktops still occupy the majority of enterprise equipment. Due to the requirement of telecommuting, users have a greater demand to remotely operate desktop computers located in the enterprise during non-office hours or locations. In order to achieve this purpose, desktop devices are often still in the working state when they are idle, which consumes a lot of power and also increases the power consumption of enterprises. In order to avoid this situation, it is particularly important to be able to automatically sleep the desktop and remotely wake up the system that has shut down or sleep the desktop when needed.

目前已经有了很多方案用于解决远程唤醒问题，可以分为三类：At present, there are many solutions to solve the remote wake-up problem, which can be divided into three categories:

第一类引入了新的硬件，比如新的网卡、主板等，比如英特尔的远程唤醒技术（RWT）。RWT需要睡眠设备的网卡与一个认证服务器维持稳定的TCP连接。当网卡从该连接接收到唤醒信息后就会唤醒主机。然而RWT仅仅对英特尔的设备有效。The first category introduces new hardware, such as new network cards, motherboards, etc., such as Intel's Remote Wakeup Technology (RWT). RWT requires the sleeping device's NIC to maintain a stable TCP connection to an authentication server. When the network card receives a wake-up message from this connection, it will wake up the host. However RWT is only valid for Intel devices.

第二类解决方案采用了虚拟化技术。在此类方案中，用户的台式主机需要安装虚拟层（Hypervisor），用户实际使用的是安装在虚拟层上层的虚拟机。当用户机器睡眠时，虚拟机将被迁移至托管服务器。当用户返回时，虚拟机将再次迁移到台式设备。此类解决方案的弊端是所有台式机上需要安装虚拟层，并且需要将用户操作系统改为虚拟机，大部分用户不愿意接受这样的工作环境上的变化。The second type of solution uses virtualization technology. In this type of solution, the user's desktop host needs to install a virtualization layer (Hypervisor), and what the user actually uses is a virtual machine installed on the upper layer of the virtualization layer. When the user's machine goes to sleep, the virtual machine is migrated to the hosting server. When the user returns, the virtual machine is migrated to the desktop again. The disadvantage of this type of solution is that a virtualization layer needs to be installed on all desktops, and the user operating system needs to be changed to a virtual machine. Most users are unwilling to accept such a change in the working environment.

第三类依赖于代理设备用于唤醒睡眠的机器。唤醒时，代理设备发送一个带有目的主机MAC地址的Wake-on-Lan(WOL)数据包。由于WOL数据包仅对内网设备有效，所以应用此方案时要求每个子网中必须有个代理设备。在此类方案目前最主要且有代表性的有两种方法，其一为在每个子网中设置一个专用代理服务器，其二是子网中每个还在工作的台式机均承担代理任务，通过分布式的方案，只要有设备未睡眠就可以用来唤醒其他机器。然而这两种方法都存在着很大的不足，第一种方法需要特定的服务器，购置及管理费用颇高，并且子网内仅一台代理也构成单点故障，若增加备份服务器则会增加更多的费用；第二种方法需要活动的机器频繁的检测其他机器来获知其他设备的状态，进行状态同步，从而造成对本身CPU及网络的极大开销。The third category relies on proxy devices for waking up sleeping machines. When waking up, the proxy device sends a Wake-on-Lan (WOL) packet with the MAC address of the destination host. Since the WOL data packet is only valid for intranet devices, it is required to have a proxy device in each subnet when applying this scheme. Currently, there are two main and representative methods in this type of scheme. One is to set up a dedicated proxy server in each subnet, and the other is to assume the proxy task for every desktop computer that is still working in the subnet. Through the distributed scheme, as long as a device is not sleeping, it can be used to wake up other machines. However, these two methods have great disadvantages. The first method requires a specific server, and the purchase and management costs are high, and only one agent in the subnet also constitutes a single point of failure. If you add a backup server, it will increase More costs; the second method requires the active machine to frequently detect other machines to know the status of other devices and perform status synchronization, resulting in a huge overhead on its own CPU and network.

发明内容Contents of the invention

有鉴于现有技术的上述不利因素，本发明所要解决的问题是提供一种不需要增加新的硬件设备、不改变企业工作的环境并且在较少的资金花费和较少的CPU及网络带宽开销的情况下，能够有效的远程唤醒企业或大型组织内部台式计算机的技术，从而节省企业能源。台式机在空闲时能够自动睡眠并且在需要时被顺利唤醒。In view of the above-mentioned unfavorable factors of the prior art, the problem to be solved by the present invention is to provide a kind of environment that does not need to add new hardware equipment, does not change the enterprise work and in less capital expenditure and less CPU and network bandwidth overhead Under certain circumstances, it can effectively remotely wake up the technology of desktop computers in enterprises or large organizations, thereby saving energy for enterprises. Desktops can automatically sleep when idle and wake up gracefully when needed.

为了实现上述目的，本发明利用云服务供应商提供的在线云租赁服务，利用云服务维护内网中有限数目中介台式计算机（agent）的持续网络连接，当唤醒时通过中介计算机向目标台式机发送带有目的台式机MAC地址的Wake-on-Lan数据包进行唤醒操作。In order to achieve the above object, the present invention utilizes the online cloud leasing service provided by the cloud service provider, utilizes the cloud service to maintain the continuous network connection of a limited number of intermediary desktop computers (agents) in the intranet, and sends A Wake-on-Lan packet with the MAC address of the destination desktop performs a wake-up operation.

基于云服务的远程桌面计算机唤醒技术主要包括云服务端和客户机端。云服务端可以配置在租赁的云服务上或是私有云中，而企业中的每台台式计算机都可以安装客户机端并承担唤醒同子网下其他计算机的任务。The remote desktop computer wake-up technology based on cloud service mainly includes cloud server and client. The cloud server can be configured on the rented cloud service or in the private cloud, and each desktop computer in the enterprise can install the client and assume the task of waking up other computers under the same subnet.

云服务器及客户机端包括以下步骤：The cloud server and the client end include the following steps:

步骤1：当台式计算机启动或睡眠前会向云服务端发送时间信息。服务器端记录下台式计算机的睡眠及启动时间，并记录下来存储到数据库中。由于步骤1不会大量丢包，所以可以认为数据库中的状态记录基本可信。Step 1: When the desktop computer starts or sleeps, it will send time information to the cloud server. The server side records the sleep and startup time of the desktop computer, and records it and stores it in the database. Since step 1 will not lose a lot of packets, it can be considered that the state records in the database are basically reliable.

步骤2：当用户需要唤醒企业内的台式机时，用户使用家庭电脑或移动设备等登录云服务。当用户通过登录云服务端提供的认证系统进入服务想远程唤醒自己企业中的台式计算机时，触发云服务向所在内网中的代理设备发送唤醒要求。Step 2: When the user needs to wake up the desktop in the enterprise, the user logs in to the cloud service using a home computer or mobile device. When the user enters the service through the authentication system provided by the cloud server and wants to remotely wake up the desktop computer in his enterprise, the cloud service is triggered to send a wakeup request to the proxy device in the intranet.

步骤3：云服务根据数据库中的睡眠记录对内网中的台式机进行聚类分组，同组内的台式机有更大的概率在比较近的时间内睡眠。云服务基于分组结果选择一定数目（如三台）中介台式机。云服务同时记录下每天最后睡眠或一直工作的机器，同样选为一台中介台式机。这些中介台式机可被称为代理设备。用户登录后，唤醒命令需要的信息，如目的台式机的MAC地址会从云端发送至中介台式机。Step 3: The cloud service clusters the desktops in the intranet according to the sleep records in the database, and the desktops in the same group have a greater probability of sleeping in a relatively recent time. The cloud service selects a certain number (for example, three) of intermediary desktops based on the grouping result. At the same time, the cloud service records the last sleeping or working machine every day, which is also selected as an intermediary desktop. These intermediary desktops may be referred to as proxy devices. After the user logs in, the information required for the wake-up command, such as the MAC address of the destination desktop, will be sent from the cloud to the intermediary desktop.

步骤4：代理设备从云端接收到需要唤醒台式机的MAC地址后，就会立即发送WOL命令并测试目的台式机是否能够ping通，从而确定该台式机是否被唤醒。由于有多台代理设备发送WOL命令用于唤醒目的台式机，所以不需要重新发送。Step 4: After the proxy device receives the MAC address of the desktop computer that needs to be woken up from the cloud, it will immediately send the WOL command and test whether the target desktop computer can be pinged, so as to determine whether the desktop computer is woken up. Since there are multiple proxy devices sending WOL commands to wake up the destination desktop, there is no need to resend them.

步骤5：当ping的返回信息显示获知目的台式机已经被唤醒，代理设备会通过云服务显示唤醒成功。Step 5: When the return information of ping shows that the destination desktop has been woken up, the proxy device will display the wakeup success through the cloud service.

步骤6：为了进一步确保可用性，代理设备将要睡眠前需要监测其他代理设备状态，如果有其他代理设备不在工作状态，需要向云服务发出示警，报告代理设备缺失，从而云服务需要选择新的代理设备。Step 6: In order to further ensure availability, the proxy device needs to monitor the status of other proxy devices before going to sleep. If other proxy devices are not working, it needs to send an alert to the cloud service and report that the proxy device is missing, so that the cloud service needs to select a new proxy device .

步骤7：如果当前局域网内活动着的台式机数目较少（如少于总数目的三分之一）时，代理设备自动睡眠时会自我阻止，并向云端发送即将睡眠信息，得到云段确认后才能睡眠，从而使得代理设备更加可靠。Step 7: If the number of active desktops in the current local area network is small (for example, less than one-third of the total), the proxy device will stop itself when it automatically sleeps, and send an imminent sleep message to the cloud, and get the confirmation of the cloud segment to sleep, making the proxy device more reliable.

步骤8：如果当前活动设备数目少于所需的代理设备数目，直接唤醒睡眠了的台式机成为新的代理设备。Step 8: If the number of current active devices is less than the required number of proxy devices, directly wake up the sleeping desktop to become a new proxy device.

如上述基于云服务的远程桌面计算机唤醒技术，其中，步骤3中的聚类算法以及步骤6、7中的方法合称为“选择——保证机制”，该机制一直运行从而保证了总是有活动着的代理设备用于转发WOL包来唤醒目的台式机。For example, the cloud service-based remote desktop computer wake-up technology mentioned above, wherein the clustering algorithm in step 3 and the methods in steps 6 and 7 are collectively referred to as the "selection-guarantee mechanism". The active proxy device is used to forward WOL packets to wake up the destination desktop.

本发明，基于云服务的远程桌面计算机唤醒技术具有以下有益的技术效果：In the present invention, the remote desktop computer wake-up technology based on cloud service has the following beneficial technical effects:

（1）可有效为企业节能，远程唤醒技术使用户不需要让自己的台式计算机一直处于工作状态，从而节省了计算机空闲状态时耗去的电能。(1) It can effectively save energy for enterprises. The remote wake-up technology makes users not need to keep their desktop computers in working state all the time, thus saving the power consumed when the computer is idle.

（2）基于云服务的远程桌面计算机唤醒技术能够有效的保证远程唤醒的有效性，采用多代理设备、选择——保证机制等能有效保证至少有一个代理设备处于工作状态。(2) The remote desktop computer wake-up technology based on cloud services can effectively guarantee the effectiveness of remote wake-up, and the use of multi-agent devices and selection-guarantee mechanisms can effectively ensure that at least one agent device is in a working state.

（3）由于基于云服务的远程桌面计算机唤醒技术不需要更换新的硬件或增加多服务器，不需要增添额外费用，租赁的云服务又可以为多个子网共用，从而有效降低了成本。同时，管理工作集中在云服务端，管理成本同样有效降低。(3) Since the remote desktop computer wake-up technology based on cloud services does not need to replace new hardware or add multiple servers, no additional costs need to be added, and the rented cloud services can be shared by multiple subnets, thereby effectively reducing costs. At the same time, the management work is concentrated on the cloud server, and the management cost is also effectively reduced.

（4）相比于其他方案，基于云服务的远程桌面计算机唤醒技术还可以充分考虑安全问题。登录云服务系统需要进行身份认证，同时，由于唤醒操作都是通过云服务段开始发出，可以避免过多唤醒操作的同时进行从而产生重大的电力峰值。(4) Compared with other solutions, the remote desktop computer wake-up technology based on cloud services can also fully consider security issues. Logging in to the cloud service system requires identity authentication. At the same time, since the wake-up operations are all sent through the cloud service segment, it is possible to avoid excessive power peaks caused by simultaneous wake-up operations.

（5）由于基于云服务的远程桌面计算机唤醒技术的绝大部分工作都集中在云服务端，使桌面计算机本身的CPU及网络带宽占用率较低，也使得系统在用户办公电脑上运行不会对用户正常使用造成负面影响。(5) Since most of the cloud service-based remote desktop computer wake-up technology is concentrated on the cloud server, the CPU and network bandwidth occupancy rate of the desktop computer itself is low, and the system will not run on the user's office computer. Negatively affect the normal use of users.

附图说明Description of drawings

图1是现有最典型基于WOL技术的远程桌面唤醒技术的结构示意图。FIG. 1 is a schematic structural diagram of the most typical remote desktop wake-up technology based on WOL technology.

图2是本发明的远程桌面唤醒技术的结构示意图。Fig. 2 is a schematic structural diagram of the remote desktop wake-up technology of the present invention.

图3是远程桌面唤醒技术系统各部件间的工作过程示意图。FIG. 3 is a schematic diagram of the working process among components of the remote desktop wake-up technology system.

具体实施方式Detailed ways

以下将结合附图对本发明的构思、具体结构及产生的技术效果作进一步说明，以充分地了解本发明的目的、特征和效果。The idea, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings, so as to fully understand the purpose, features and effects of the present invention.

按照背景技术中的分类，现有的远程桌面唤醒技术可以分为三类：增加新硬件、虚拟环境以及睡眠代理。前两种方案或是由于需要为每台机器增加硬件从而带来更新和管理、部署成本，或是完全改变办公环境影响用户习惯并且桌面虚拟环境对一些终端设备的支持也同样是个问题。采用睡眠代理的方式成为一种选择。According to the classification in the background art, the existing remote desktop wake-up technologies can be divided into three categories: adding new hardware, virtual environment and sleep agent. The first two solutions are either due to the need to add hardware for each machine to bring update, management, and deployment costs, or completely change the office environment to affect user habits, and the desktop virtual environment’s support for some terminal devices is also a problem. The approach of using sleep proxies becomes an option.

图1是现有采用睡眠代理方式中具有代表性的方法，需要在每个局域网中增设一台服务器作为睡眠代理，对于远程唤醒的需求，由代理服务器发送Wake-On-Lan数据包，并最终唤醒终端台式机。由于睡眠代理方式的基础是Wake-On-Lan机制，所有局域网都需要增设代理服务器实现远程唤醒。每个局域网中都要增加新的服务器，这不仅在购买成本、配置成本和管理成本上增加了开销，也增加了新的用电设备。同时，考虑到单点故障问题，可能需要更多的服务器从而进一步增加问题的严重性。Figure 1 is a representative method in the existing sleep proxy mode. It is necessary to add a server in each LAN as a sleep proxy. For remote wake-up requirements, the proxy server sends Wake-On-Lan data packets, and finally Wake up the terminal desktop. Since the basis of the sleep proxy method is the Wake-On-Lan mechanism, all LANs need to add a proxy server to realize remote wake-up. A new server must be added in each LAN, which not only increases the cost of purchase, configuration and management, but also adds new power consumption equipment. Also, considering the single point of failure problem, more servers may be required to further increase the severity of the problem.

另一种采用睡眠代理的方案不需要在局域网中增加服务器作为睡眠代理，而是使局域网中的每台活动着的台式机都成为睡眠代理，为此，每台台式机需要随机访问其他机器来确定其他机器的状态来决定是否“接管”该台式机从而当需要唤醒该台式机时由“接管”机器发送Wake-On-Lan数据包。因为每台台式机都是独立的，要知道其他机器的状态需要不停的访问其他设备，当局域网中台式机总数较多时，不仅增加了每台台式机本身的工作负载，同时也对企业网络增加占用率。Another scheme using sleep proxy does not need to add a server as a sleep proxy in the LAN, but makes every active desktop in the LAN a sleep proxy. For this reason, each desktop needs random access to other machines to Determine the status of other machines to decide whether to "take over" the desktop so that the "takeover" machine sends Wake-On-Lan packets when the desktop needs to be woken up. Because each desktop computer is independent, to know the status of other machines, it is necessary to continuously access other devices. When the total number of desktop computers in the local area network is large, it not only increases the workload of each desktop computer itself, but also affects the corporate network. Increase occupancy.

图2是本发明的方法，针对上述方案的弊端，该方法增加了一个统一的云服务，但不需要在每个局域网中都增加代理服务器，从而没有了在每个局域网中增加服务器带来的各项成本及新的电力开销问题，同时利用统一的云服务获取局域网中各台式机的状态信息，并维护局域网中一定数量的代理，实现远程桌面唤醒而不需要代理不停的访问其他台式机来获取状态信息。Fig. 2 is the method of the present invention, for the drawback of above-mentioned scheme, this method has increased a unified cloud service, but does not need to all increase proxy server in each local area network, thus has not increased server in each local area network Various costs and new power consumption issues, and at the same time use a unified cloud service to obtain the status information of each desktop in the LAN, and maintain a certain number of agents in the LAN to achieve remote desktop wake-up without the need for agents to continuously access other desktops to get status information.

图3是远程桌面唤醒方法的运行过程。控制信息主要在云服务和局域网中台式机间进行，而台式机中有一定数目的特殊设备被称为代理设备，用来协助云服务完成最终唤醒操作。Fig. 3 is the running process of the method for waking up the remote desktop. The control information is mainly carried out between the cloud service and the desktop computer in the local area network, and there are a certain number of special devices in the desktop computer called proxy devices, which are used to assist the cloud service to complete the final wake-up operation.

本发明基于云服务的远程桌面计算机唤醒技术的具体步骤如下：The concrete steps of the remote desktop computer wake-up technology based on cloud service of the present invention are as follows:

步骤1：当台式计算机启动或睡眠前会向云服务端发送时间信息（如图3中的过程1）。服务器端维护一个哈希表用于记录下各台式计算机的睡眠及启动时间，这个哈希表被成为状态表。而状态表的历史信息均被存储在数据库中。由于步骤1不存在大量丢包，所以数据库中记录的各台式机的状态信息可以认为是基本可信的。Step 1: When the desktop computer starts or goes to sleep, it will send time information to the cloud server (process 1 in Figure 3). The server side maintains a hash table to record the sleep and startup time of each desktop computer, and this hash table is called a state table. The historical information of the state table is stored in the database. Since there is not a large amount of packet loss in step 1, the state information of each desktop computer recorded in the database can be considered to be basically credible.

步骤2：用户需要唤醒企业内的台式机时，利用从企业获取到的用户名和密码登录到云服务端（如图3中的过程2）。云服务端的登录和认证服务可以直接利用云本身提供的应用，如Openstack的dashboard和keystone服务。用户登录云端后，触发向对应内网中的代理设备发送唤醒要求。Step 2: When the user needs to wake up the desktop in the enterprise, log in to the cloud server with the user name and password obtained from the enterprise (process 2 in Figure 3). The login and authentication services of the cloud server can directly use the applications provided by the cloud itself, such as the dashboard and keystone services of Openstack. After the user logs in to the cloud, a wake-up request is sent to the proxy device in the corresponding intranet.

步骤3：云服务根据数据库中的睡眠记录使用聚类算法对内网中的台式机进行聚类分组，同组内的台式机有更大的概率在比较近的时间内睡眠。云服务根据分组结果选择一定数目（如三台）中介台式机，云服务同时记录下每天最后睡眠或一直工作的机器，同样选为一台中介台式机。这些台式机被称为代理设备。用户登录后，唤醒命令需要的信息（如目的台式机的MAC地址）会被从云端发送至代理设备（如图3中的过程3）。Step 3: The cloud service uses a clustering algorithm to cluster the desktops in the intranet according to the sleep records in the database. The desktops in the same group have a greater probability of sleeping in a relatively recent time. The cloud service selects a certain number (for example, three) of intermediary desktops according to the grouping results, and the cloud service also records the last sleeping or working machine every day, and also selects an intermediary desktop. These desktops are called proxy appliances. After the user logs in, the information required by the wake-up command (such as the MAC address of the destination desktop) will be sent from the cloud to the proxy device (process 3 in Figure 3).

具体聚类算法，以台式机启动时间作聚类条件的K-聚类算法为例：For the specific clustering algorithm, take the K-clustering algorithm as an example of the desktop startup time as the clustering condition:

(a)如已知分类数为k；选取k个样本点作为初始聚类中心，记为z₁(l),z₂(l),z₃(l),...,z_k(l),迭代序号l＝1；(a) If the number of classifications is known to be k; select k sample points as the initial clustering center, denoted as z ₁ (l), z ₂ (l), z ₃ (l),...,z _k (l ), iteration number l=1;

(b)使用近邻规则将所有样本分配到各聚类中心所代表的k类w_j(k)中，各类所包含的样本数为N_j(l)；(b) Use the nearest neighbor rule to assign all samples to k categories w _j (k) represented by each cluster center, and the number of samples contained in each category is N _j (l);

(c)计算各类的重心（均值向量),并令该重心为新的聚类中心，即 z j ( l + 1 ) = 1 N j ( l ) Σ X i ∈ w j ( l ) X i , j = 1,2 , . . . , k ; (c) Calculate the center of gravity (mean vector) of each type, and make the center of gravity the new cluster center, that is z j ( l + 1 ) = 1 N j ( l ) Σ x i ∈ w j ( l ) x i , j = 1,2 , . . . , k ;

(d)如果z_j(l+1)!＝z_j(l),则返回步骤(2),继续迭代计算。(d) If z _j (l+1)!=z _j (l), return to step (2) and continue iterative calculation.

(e)如果z_j(l+1)＝z_j(l)，迭代过程结束，此时的聚类结果就是最优聚类结果。(e) If z _j (l+1)=z _j (l), the iterative process ends, and the clustering result at this time is the optimal clustering result.

步骤4：代理设备从云端接收到需要唤醒台式机的信息（MAC地址）后，会在发送Wake-on-Lan数据包唤醒目的台式机的同时使用ping命令测试该台式机是否已经被唤醒（如图3中过程4）。并且由于代理设备不止一台，多台设备都会发送WOL数据包，所以没有必要重新发送WOl数据包来确保目的台式机能被成功唤醒。Step 4: After the proxy device receives the information (MAC address) that needs to wake up the desktop from the cloud, it will use the ping command to test whether the desktop has been woken up while sending the Wake-on-Lan data packet to wake up the destination desktop (such as Process 4 in Figure 3). And because there are more than one proxy device, multiple devices will send the WOL packet, so there is no need to resend the WOL packet to ensure that the destination desktop can be successfully woken up.

步骤5：当ping命令显示目的台式机已经被成功唤醒后，代理设备会通过云端告知用户唤醒操作成功（如图3中过程5、6、7），用户从而远程登录自己的台式机。Step 5: When the ping command shows that the target desktop has been successfully woken up, the proxy device will notify the user of the successful wake-up operation through the cloud (processes 5, 6, and 7 in Figure 3), and the user can log in to his desktop remotely.

步骤6：为了进一步确保可用性，代理设备将要睡眠前需要监测其他代理设备状态，如果有其他代理设备不在工作状态，需要向云服务发出示警，报告代理设备缺失，从而云服务需要选择新的代理设备。Step 6: In order to further ensure availability, the proxy device needs to monitor the status of other proxy devices before going to sleep. If other proxy devices are not working, it needs to send an alert to the cloud service and report that the proxy device is missing, so that the cloud service needs to select a new proxy device .

其具体步骤如下：The specific steps are as follows:

(a)代理设备检测到某个代理设备自动睡眠，向云端报告。(a) The agent device detects that a certain agent device automatically sleeps, and reports to the cloud.

(b)在云端，如果该缺失的代理设备是工作最晚的代理设备，重新遍历状态表，在仍然工作的台式机中找到新的工作到最晚的代理设备。(b) In the cloud, if the missing proxy device is the latest proxy device, re-traverse the state table to find a new proxy device that works the latest in the still working desktops.

(c)如果该缺失的代理设备是从聚类分组中选择的，则对剩下仍在工作的台式机进行重新聚类分组，选择出新的代理设备。(c) If the missing proxy device is selected from the cluster group, re-cluster and group the remaining working desktops to select a new proxy device.

(d)云服务通知新选择的台式机使其成为新的代理设备。(d) The cloud service notifies the newly selected desktop to become the new proxy device.

步骤7：如果当前局域网内活动着的台式机数目较少（如少于总数目的三分之一）时，代理设备自动睡眠时会自我阻止，并向云端发送即将睡眠信息，得到云段确认后才能睡眠，从而使得代理设备更加可靠。其具体步骤如下：Step 7: If the number of active desktops in the current local area network is small (for example, less than one-third of the total), the proxy device will stop itself when it automatically sleeps, and send an imminent sleep message to the cloud, and get the confirmation of the cloud segment to sleep, making the proxy device more reliable. The specific steps are as follows:

(a)当一个子网中活动的台式机数目少于总数的三分之一时，系统进入特殊状态。(a) When the number of active desktops in a subnet is less than one-third of the total, the system enters a special state.

(b)在特殊状态下，代理设备要睡眠时需保持工作状态并向云端发送信息。(b) In a special state, the proxy device needs to keep working and send information to the cloud when it wants to sleep.

(c)云端确认信息后重新选择新的代理设备替代原代理设备工作。(c) After the cloud confirms the information, a new proxy device is selected to replace the original proxy device.

(d)云端返回信息至原代理设备，该代理设备进入睡眠。(d) The cloud returns information to the original proxy device, and the proxy device goes to sleep.

(e)如果该子网中活动设备数目超过总数的三分之一后，系统退出特殊状态。(e) If the number of active devices in the subnet exceeds one-third of the total, the system exits the special state.

步骤8：如果当前活动设备数目少于所需的代理设备数目，直接唤醒睡眠了的台式机成为新的代理设备（如图3中过程8）。Step 8: If the number of current active devices is less than the required number of proxy devices, directly wake up the sleeping desktop to become a new proxy device (process 8 in Figure 3).