SDDC: A Software Defined Datacenter Experimental Framework PDF

Preview

Summary

SDDC: A Software Defined Datacenter Experimental Framework Ala’ Darabseh1 , Mahmoud Al-Ayyoub1 , Yaser Jararweh1 , Elhadj Benkhelifa2 , Mladen Vouk3 , and Andy Rindos4 , 1 Jordan University of Science and Technology, Irbid, Jordan 2 Mobile Fusion Applied Research Centre, Staffordshire University, St...

Description

Accelerat ing t he world's research.

SDDC: A So ware Defined Datacenter Experimental Framework Yaser Jararweh 2015 3rd International Conference on Future Internet of Things and Cloud

Cite this paper

Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers

Download a PDF Pack of t he best relat ed papers 

Cent ralized Packet Rout ing and Congest ion Cont rol in Soft ware Defined Net working Abiodun At olani

SDSecurit y: A Soft ware Defined Securit y experiment al framework Yaser Jararweh Soft ware-Defined Net working: A Comprehensive Survey MARIA AT RÍCIA SABINO MACIEL alu.ufc.br

SDDC: A Software Defined Datacenter Experimental Framework Ala’ Darabseh1 , Mahmoud Al-Ayyoub1 , Yaser Jararweh1 , Elhadj Benkhelifa2 , Mladen Vouk3 , and Andy Rindos4 , 1 Jordan University of Science and Technology, Irbid, Jordan 2 Mobile Fusion Applied Research Centre, Staffordshire University, Stafford. UK 3 North Carolina State University, Raleigh, North Carolina, USA. 4 IBM Corporation, Research Triangle Park, North Carolina, USA. Abstract—The rapid growth and the distributed sites of the datacenters increase the complexity of control and management processes. A new paradigm which is called Software Defined Systems (SDSys) comes as a solution to reduce the overhead of datacenters management by abstracting all the control functionalities from the hardware devices and setting it inside a software layer. These functionalities are responsible for organizing and controlling the main blocks of the datacenter; network, storage, compute and security. The Software Defined Datacenter (SDD) integrates the software defined concepts into all of these main blocks. Transferring these concepts into workable systems requires checking several performance aspects and testing its correctness before building real systems. In this paper we introduce a novel experimental framework (SDDC) to provide a novel virtualized testbed environment for SDD systems. This work builds on the Mininet simulator, where its core components, the host, the switch and the controller, are customized to build the proposed experimental simulation framework for SDD. This simulator lets the users develop and test their own SDD solutions, and at the same time gives the researchers an experimentation tool for benchmarking purposes. The developed simulator could also be used as an educational tool to train students and novice researchers.

I.

I NTRODUCTION

In datacenters, especially large scale datacenter which adapt the cloud computing in their work, arise the need to find a standardize way to control and manage their infrastructure. The distributed locations of the data centers and the varied infrastructure providers make the control of the datacenters a complex process. Software Defined Systems (SDSys) is a very recent paradigm, proposed to address control and management challenges known in traditional platforms by hiding their complexities from the end users. This is achieved by isolating the data plane from the control plane. SDSys technology has grown very rapidly to encompasses a number of disciplines such as Networking (SDN), Storage (SDStor), Security (SDSec), Compute (SDCompute), Datacenters (SDD), etc. A technical report for a cloud networking company named CohesiveFT [1] discusses SDD as being “defined in Software”, i.e., all the needs of the datacenters (storage, security, compute and network) are defined by the software. The efficiency, cost savings and other benefits are fully capitalized by the SDD by deploying, controlling and migrating the application to the cloud. It is important to differentiate between the software defined concept and another related concept which is “software

deployed.” In the former concept, the APIs and software are used to control and manage the resources and devices. On the other hand, the software deployed concept means that the functionality of the service is deployed in a computer hardware object. Using a software to manage and control the resources is not a new concept. The essential difference, which was brought by software definedness, is the ability of the control layer to control all the underlying resources regardless of their vendor variations by physically isolating them from the hardware resources in the data layer [2]. The concept of abstraction in SDSys is similar to the idea of Object Oriented (OO) paradigm, where the implementation is separated from the interface representing the data layer and the control layer, respectively, in SDSys. The reason behind this separation is to simplify the modification process, so that any change in the implementation will not affect the interface and vice versa [2]. Transferring the concepts derived from SDSys into real workable system is not a simple task. Implementing new concepts or solutions directly in operational systems before testing them in simulated environments are considered inefficient, costly and risky; especially, when dealing with security related solutions. Though, apart from SDN, there exist no simulation environments for most SDSys, including SDStore, SDSec and SDCompute. The Mininet [3] simulator is the one of most widely used simulators by the SDN research community due to its simplicity and usability. In our previous researches [4], [5], we built experimental frameworks for SDStore and SDSec systems to allow the researchers to build their own topologies and test it across different performance aspects. The datacenter is considered a comprehensive system built from several main blocks which are the network, storage, security and compute. The same vein, the SDD is also integrate all the software defined systems of these main blocks and ingrate their functionalities. In this work, we endeavour to extend the features and functionalities of the Mininet simulator to enable the simulation of SDD systems. This framework is called the SDDC experimental framework. Virtualization has been widely adopted across different sectors following the rapidly increasing adoption of Cloud Computing technology and services due to its advantages over traditional computing provision. For that, in this paper we build SDDC by integrating the SDN (Mininet), SDStor (SDStorage), SDSec (SDSurity) experimental frameworks plus building another software defined block (SDCompute) which takes virtualization and the computing resources into

prepared by thr control plane in the controller to forward the messages, flow-packets [8]. Several research works have been published on SDN. In [9], the authors discussed most of SDN aspects and illustrated how this paradigm can support the Software Defined Environments (SDE). In addition, they showed the IBM vision to consolidate the SDN idea by integrating their IBM SDN virtual environments (SDN-VE) product with the Neutron, OpenStack network platform [10] to extend SDN-VE features. Software Defined Storage (SDStore). SDStore is one of the most important subsystems in SDSys. It takes the responsibility of managing huge data in storage systems by isolating the data control layer from the data storage layer. The control layer refers to the software component that manages and controls the storage resources, whereas the data layer refers to the underlying infrastructure of the storage assets[4].

Fig. 1.

The architecture of SDD and its main components [1]

consideration. The developed simulator, will not only support the development and testing of SDD solutions, it will also serve as an experimentation tool for researchers and for benchmarking purposes. It could also be used as an educational tool to train students and novice researchers. The rest of this paper is structured as follows. In section II we explain the idea of SDD in more details including its architecture and components and discuss some real SDD systems. After that, a brief introduction about Mininet is given in section III, whereas, our experimental SDDC framework is explained in section IV. Finally, we conclude this paper and present our future plans in section V. II.

S OFTWARE D EFINED DATACENTER (SDD)

SDD is a new paradigm which simplifies the control and management of the system [6]. This concept is a comprehensive component to different software defined systems which are the Software Defined Network (SDN), Software Defined Storage (SDStore), Software Defined Security(SDSec) and Software Defined Compute (SDCompute) [7]. Figure 1 shows these main components of the SDD. Software Defined Network (SDN). SDN is the latest innovation in network environment. As any software defined system, the SDN simplifies the network management by separating the control plane from the data plane. In network communication, the messages created by the user present the data plane; these messages need to be transferred to an appropriate destination. Network management is responsible for finding the best path to send the message by using the control messages while maintaining any information related to network like network traffic. In SDN the data plane and the control plane are separate, where the data plane uses the forwarding tables

Many corporations realize the benefits of SDStore and apply it in their storage centers. Examples include EMC Corporation, which launched ViPR software as an implementation for SDStore [11], IBM with its Storwize software [12], and many others. In addition to Storwize, ViPR, and NexentaStor SDStore solutions, Atlantis USX [13] proposed and implemented another SDStore solution to accelerate and enhance the performance and capacity of the system with a minimum cost. Actually, there are many proposed, implemented, or deployed SDStore solutions like Maxta [14], HITACHI [15], Datacore [16], CloudBytes [17], IBM SmartCloud [18], etc. We just highlight some of them to explain how the new storage architecture handles the limitations of classical storage system with higher performance and dramatically lower cost. Software Defined Security (SDSec). Similar to the SDN architecture, the architecture of SDSec separates the data plane from the control plane. The general view of this architecture is organized into three main layers: the physical layer, the control layer and the application layer[5]. The idea behind the SDSec concept appeared at the Cloud Security Alliance (CSA)1 as they sought to find a new approach for security with lower costs [19]. To transfer their vision into reality they launched the Software Defined Perimeter (SDP) project as new security architecture in order to keep secure systems against network attacks [20]. SDP was designed to complement SDN in order to reduce the attacks on the network applications by disconnecting them until the users and devices are authenticated. Another security company launched its own SDSec solution, called vArmour, for SDNbased and cloud data center systems to fully exploit the benefits of virtualization environments [21]. Software Defined Compute (SDCompute). Pishro SDS [22] argue that SDC is the first step toward SDD. Moreover they called it the server virtualization. The virtualization is used to accelerate the run time and let more than one simulation, user, task to work at the same time by encapsulating and isolating each one on a separate VM and let many VMs to work simultaneously [23]. In such a way, the system resources will be exploited very well and critical decisions will be 1 https://cloudsecurityalliance.org/

taken faster. Moreover, the system can scale up to cover many locations and users. III.

M ININET: A N ETWORK E MULATOR

Mininet is an OpenFlow-based SDN simulator which gives researchers an efficient way to evaluate their SDN frameworks by studying their behaviors and measuring their performances. Mininet is an open source simulator written in the python programming language and is built over the Ubuntu Linux distribution. The elements of Mininet are organized into three main components: the host, which sends and receives the packets, the switch, which stores all the required rules to forward the packets to its destinations, and a central controller, which handles the functionality of control and management operations in the network. Mininet supports different types of virtualized hosts, switches and controllers. Furthermore, it provides two essential tests, ping and iperf, to check the reachability and the network bandwidth, respectively. The minimum topology supported by Mininet consists of two hosts (h1 and h2), one switch (s1) and a central controller (c0). However, the users are given the flexibility to extend this topology and build their own customized topologies to test the performance of their algorithms/solutions. One of the drawbacks of Mininet is its inability to handle large scale networks [24]. MaxiNet is an extension of Mininet to emulate a large scale environment by building a distributed emulation environment [24]. The growing interest in Software Defined Systems (SDSys) in general and especially SDSec is the motivation behind our proposal to build an experimental framework for the SDSec system with the ambition to help advance SDSec research similar to what Mininet did for SDN. Moreover, the simplicity and flexibility of the Mininet help us to embody this ambition and transform it into reality by customizing the existing elements in Mininet. This gives researchers the chance to test their new systems before transferring them into real systems. IV.

SDDC: A S OFTWARE D EFINED DATACENTER E MULATOR

In this section we explain in details our Software Defined Datacenter experimental framework, SDDC. The proposed experimental platform is based on Mininet. Starting with the environment and characteristics of Mininet, we build SDDC by customizing and extending the elements of Mininet to facilitate building a virtualized environment to emulate the different Datacenter topologies test their performance under different scenarios. The idea of SDDC is to capture the basic idea of SDDC architecture, which is to abstract the data plane from the control plane. All the mechanisms; network mechanisms, storage mechanisms, compute mechanisms and security mechanisms are abstracted from the physical devices and set inside the software defined datacenter controller inside the control layer (centralized control) in the SDDC. In such a way, when the host sends data traffic to another host in the network, the forwarding process, authentication process, reservation process and all the datacenter controls occur at the control layer not at the device level (physical layer).

Fig. 2. The main components of SDDC and their integration to control the network.

To explore how this happens, we show next the main elements of the SDDC and how these elements interact with each other to provide an experimental framework for testing the efficiency and effectiveness of SDDC systems and determine different performance measurements level of their topologies. A. Elements of SDDC SDDC is divided into three main customized components: the host, the switch and the controller. Figure2 presents the general view of the SDDC framework and its main elements. SDDC Host. Mininet gives the users the ability to customize the host based on their requirements. By this flexibility we extend the basic “Host” element in Mininet to become SDDC Host. It is meant to be a simple host which may send or receive data traffic through the network. This host is a virtualized host like other Mininet hosts with some extra parameters, which are required for SDDC purposes. We classified these parameters into four main Hosts as follow (This host inherits all of the following Hosts types):: •

SDN Host[24]: This is the main host for mininet emulator, which is used to send the packets over the network. Moreover, this is cornerstone for our experimental SDDC framework host.

•

SDStorage Host [4]:This host storage present any type of the storage arrays where the user can store his data inside it. It is S host like any another SDN Host instance with some extra parameters related the

When the user ask for a number of RAMs then the SDSecurity switch and controller check this request if it is OKAY then it forward this request to the SDCompute switch. Then, the switch check its Compute Table and see the First available space (Packing mode) and at the same time guarantee that the number of the RAMs in the requests is not greater than the Max allowed RAMs for that host (SLA condition #1), if this condition success then the switch send a packet to the controller to reserve this amount of RAMs, then the controller Update the Compute table and the Map Table.

storage emulation. As an example , the number of directories, the number of files inside each directory, the size of the file and many other preferred parameters can be created inside this host to build your customized test. •

SDSecurity Host [5]: This host is also like the two previous hosts with some parameters realted to the security issues such as “Trust”, “zone ID” (which specifies the trust zone for the host), “Per” (which defines the permissions for this host: read, write, both or none), “resources consumption” and the “scope” parameter.

•

SDCompute Host: This host is working as any SDN, SDStorage,SDsecurity Hosts plus new features related to the compute part like a number of CPUs , RAMs and so on. For our experimental SDDC framework we only focus on the RAMs component. We set 32 RAMs for every hosts. Each User can request a set of RAMs and reserve it to do their needs. Each host can allow for a certain number RAMs for each request. Range from (1-8).

SDDC Switch. Mininet supports different types of switches: UserSwitch, KernelSwitch and OVSSwitch. For our experimental framework, we choose to extend the first one due to its simplicity. In addition this switch is do the functionality of network, storage, security and compute switches. It is a combination of all of these switches (This switch inherits all of the following switches types): •

SDN Switch(User Switch in our SDDC framework)[24]: This switch is used by the network to store the Forwarding Table inside it. This Table keep information about the path to the destinations for all the Hosts in the network to simplify the communication process.

•

SDStorage Switch[4]: The SDStorage Switch inherits all the functions and parameters of the UserSwitch and implements extra functionality related to the SDStore emulation. Inside this switch, a Function Table is created to keep an up-to-date information about all the SDStorage Hosts related to its status and keeps this information available to other hosts in the system.

•

SDSecurity Switch[5]: The SDSec Switch do extra functionality related to the SDSec emulation. Inside this switch a “Policies Table” is created to store all the IPs of the authenticated SDSec Host(s) and the related access policies for each one.

•

SDCompute Switch: This switch contains two Tables; Compute Table and Map table. ◦ Compute Table: Which keep information about the Compute hosts like; available RAMs, Max allowed RAMs, Power key statues; on, off, and so on. ◦ Map Table: Inside this Table Information about the User requests are kept; Sender IP, Destination IP, Start position and End position of the RAMs, since each host kept the RAMs in array.

SDDC Controller. This controller inherits the features of the “Controller” element in the Mininet which is the super class Controller for the all OpenFlow controllers. Mininet provides users the ability to choose the best controller implementation to control their topologies whether it is implemented locally inside the Mininet VM or externally by linking the topology to a remote controller. In our SDDC emulator, we use the basic controllers for Mininet like the super class Controller itself and POX controller [25] to implement our techniques by customizing the controllers for our purposes. This controller is entirely software-bas...