Network Working Group T. Suzuki Internet-Draft T. Tarui Intended status: Informational Hitachi, Ltd. Expires: April 18, 2013 October 15, 2012 Requirements for an Energy-Efficient Network System draft-suzuki-eens-requirements-00 Abstract Requirements concerning an energy-efficient network system such as a cloud system are presented. Specifically, a large-scale cloud system, which is composed of multiple data centers (DCs) and a wide area network (WAN) to connect these DCs, is focused on. The problems needed to be overcome in order to make the system energy efficient are presented. The requirements that must be satisfied in order to solve these problems are also presented. Status of this Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." This Internet-Draft will expire on April 18, 2013. Copyright Notice Copyright (c) 2012 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of Suzuki & Tarui Expires April 18, 2013 [Page 1] Internet-Draft Energy-Efficient Network System October 2012 the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Problem Statement . . . . . . . . . . . . . . . . . . . . . . 4 3. Use Case . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 3.1. Management System Structure . . . . . . . . . . . . . . . 5 3.2. Server Function . . . . . . . . . . . . . . . . . . . . . 6 3.3. Power-saving Method on a large scale cloud system . . . . 7 4. Requirements . . . . . . . . . . . . . . . . . . . . . . . . . 8 5. Security Considerations . . . . . . . . . . . . . . . . . . . 9 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 10 7. Informative References . . . . . . . . . . . . . . . . . . . . 11 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 12 Suzuki & Tarui Expires April 18, 2013 [Page 2] Internet-Draft Energy-Efficient Network System October 2012 1. Introduction This draft describes the requirements that an energy-efficient network system must satisfy. An example of the network system is a cloud system. Recently, cloud services, which provide various resources to distant data centers (DCs), such as processing power, storage, and applications, via a network have become wide-spread around the world. The users of cloud services have been increasing in proportion to the enhancement of the services. Besides, the number and the scale of cloud systems have also been increasing in accordance with the growth of the number of users. As a result of these trends, the power consumed by the systems has risen dramatically, and devising power- saving measures targeting the systems has become one of the biggest issues. Under those circumstances, the requirements concerning a power-saving cloud system to guarantee network-access quality to a virtual machine (VM) are described in this draft. A power-saving cloud system is created by reallocating all VMs appropriately in the system in order to save power on the basis of cooperation between a DC management server and a network management server. The requirements concerning power saving at the network level are described here while the requirements concerning the device level are still being discussed by the EMAN [EMAN] working group. In section 2, specific issues on the network level are discussed. In section 3, a usage case for a power-saving cloud system is described. In section 4, requirements that the system must satisfy are prescribed. Suzuki & Tarui Expires April 18, 2013 [Page 3] Internet-Draft Energy-Efficient Network System October 2012 2. Problem Statement One way to create a power-saving cloud system is to consolidate working physical servers by appropriate VM migrations. In a VM migration, many VMs are accommodated in the same physical server while the load of the CPU for each VM is low. Unnecessary physical servers are then shut down, and power consumption of the entire system is cut. In regard to a conventional power-saving method, reallocation of the VMs is determined on the basis of the load of only physical servers and VMs. Service quality is therefore guaranteed from the perspective of CPU load. However, network congestion is possible when excess consolidation based on VM migrations is executed without considering the load corresponding to the network resource (such as bandwidth) even if enough CPU resources are available. Power-saving control for a cloud system should therefore be done while service quality is guaranteed. To do that, power-saving control must be executed from the perspective of not only CPU load but also network load. However, a method or protocol for communication between a DC management server and a network management server has not been defined yet. Suzuki & Tarui Expires April 18, 2013 [Page 4] Internet-Draft Energy-Efficient Network System October 2012 3. Use Case 3.1. Management System Structure A target cloud system for power saving is shown in Figure 1. The system is composed of multiple DCs and a wide area network (WAN) connecting the DCs. An example of a management system for cloud- system power-saving is shown in Figure 2. In the example, there is a WAN management server on the WAN side and multiple DC sub-management servers and a DC main-management server on the DC side. +---------------------------------------------------+ | | | Wide Area Network (WAN) | | | +----------+-----------------------------+----------+ | | | | | | +----------+----------+ +----------+----------+ | | | | | Data Center (1) | - - - | Data Center (n) | | | | | +---------------------+ +---------------------+ Figure 1: Example of a target cloud system Suzuki & Tarui Expires April 18, 2013 [Page 5] Internet-Draft Energy-Efficient Network System October 2012 +---------------------------------------------------+ | WAN Mgmt Srv | +-------------------------+-------------------------+ | |(1) WAN-DC I/F | +----------+--------------+-------------------------+ | DC Main-Mgmt Srv | +----------+-----------------------------+----------+ | | |(2) DC Main-Sub I/F |(2) DC Main-Sub I/F | | +----------+----------+ +----------+----------+ | | | | | DC Sub-Mgmt Srv (1) | | DC Sub-Mgmt Srv (n) | | +-------+ | +---------------------+ +---------------------+ (3) DC-DC I/F Figure 2: Example of a management system 3.2. Server Function In this section, a function for each management server shown in Figure 2 is described in concise way. WAN Management Server: The WAN managemnet server monitors available bandwidth for each node and consumed bandwidth by each VM in the WAN and predicts future loads for each resource. In addition, it calculates the minimum necessary number of data-routing paths and bandwidth on the basis of the monitored resource conditions and predicted future load for each resource. The number of routing paths is then consolidated accordingly. DC Main-Management Server: The main-management sever controls the start and end times of the VM consolidation for power saving of all DCs by utilizing the huge difference in the CPU loads on the servers during business hours and during the night. In addition, it manages other DC sub-management servers. DC Sub-Management Server: The sub-management server monitors available bandwidth for each node and consumed bandwidth by each VM and predicts future loads for each resource. It then calculates minimum necessary Suzuki & Tarui Expires April 18, 2013 [Page 6] Internet-Draft Energy-Efficient Network System October 2012 number of physical servers to accommodate all VMs (some of which are under low load late at night). Next, it reallocates them appropriately on the basis of the monitored resource conditions and predicted future load for each resource. It then consolidates routing paths according to the minimum number of necessary routing paths and bandwidth. 3.3. Power-saving Method on a large scale cloud system The following four methods can be used for power saving. 1. The power consumption of the DC is saved by turning off unnecessary physical servers after appropriate VM reallocation for the physical servers based on VM migrations while the quality of service (QoS) of the cloud services provided by the DC is guaranteed. 2. The power consumption of the DC is saved by turning off unnecessary physical servers and nodes or their ports on unnecessary data-routing paths after appropriate VM reallocation in the DC while QoS of the cloud services is guaranteed. 3. The power consumption of the DC is saved by turning off unnecessary physical servers after appropriate VM reallocation for the physical servers based on VM migrations between DCs while QoS of the cloud services is guaranteed. 4. The power consumption of the DC and WAN is saved by turning off unnecessary physical servers and nodes or their ports on unnecessary data-routing paths after appropriate VM reallocation between DCs while QoS of the cloud services is guaranteed. Suzuki & Tarui Expires April 18, 2013 [Page 7] Internet-Draft Energy-Efficient Network System October 2012 4. Requirements The interfaces shown in Figure 2 are needed to create the power- saving cloud system based on the schemes described in the previous section. The requirements for each interface are briefly described below. WAN-DC Interface: The interface is used to transmit information concerning a change in the conditions of the network and server resources (including VMs) in the DC from the DC management server to the WAN management server. In addition, by accessing the WAN management server, the DC management server uses the interface to determine if any problems occur in the WAN when the planned VM migration is executed. DC Main-Sub Interface: The interface is used by the DC main management server to inform the DC sub-management server to start or stop power- saving control for the network and server resources (including VMs) in each DC. In addition, the DC sub-management server uses the interface to determine if any problems occur in the WAN when the planned VM migration over the DCs is executed via the WAN-DC interface. DC-DC Interface: The interface is used to determine appropriate reallocations of network and server resources by exchanging conditions of each resource (such as the load on the CPU of the physical server and VMs and availability of network bandwidth in the DC). Suzuki & Tarui Expires April 18, 2013 [Page 8] Internet-Draft Energy-Efficient Network System October 2012 5. Security Considerations This document describes problems and requirements for a power-saving large-scale network system. To achieve this power saving, it is necessary to exchange information on the resource conditions in the WAN and DC between management servers in the WAN and DC. It is therefore necessary to use a secure communication channel between management servers. Suzuki & Tarui Expires April 18, 2013 [Page 9] Internet-Draft Energy-Efficient Network System October 2012 6. IANA Considerations This document includes no request for IANA. Suzuki & Tarui Expires April 18, 2013 [Page 10] Internet-Draft Energy-Efficient Network System October 2012 7. Informative References [EMAN] "EMAN Working Group". Suzuki & Tarui Expires April 18, 2013 [Page 11] Internet-Draft Energy-Efficient Network System October 2012 Authors' Addresses Toshiaki Suzuki Central Research Laboratory, Hitachi, Ltd. 292 Yoshida-cho Totsuka-ku, Yokohama, Kanagawa 244-0817 Japan Phone: +81-45-860-2177 Email: toshiaki.suzuki.cs@hitachi.com Toshiaki Tarui Central Research Laboratory, Hitachi, Ltd. 292 Yoshida-cho Totsuka-ku, Yokohama, Kanagawa 244-0817 Japan Phone: +81-45-860-2177 Email: toshiaki.tarui.my@hitachi.com Suzuki & Tarui Expires April 18, 2013 [Page 12]