Dynamic Pricing Trade Market or Shared Resources in IIU Federated Coud Tongrang Fan 1, Jian Liu 1, Feng Gao 1 1Schoo o Inormation Science and Technoogy, Shiiazhuang Tiedao University, Shiiazhuang, 543, China Fantr29@126.com, iuian331316@hotmai.com,.gao@ive.com Abstract. A trade market or shared resources comprises o ederation members and market agent was proposed with introducing o market mechanism in the process o resources sharing. The dynamic pricing mechanism o shared resources in this market was studied, market agent adusts the proit rate o resources dynamicay according to the statistica data o the suppy-demand reationship, thus to achieve dynamic change o resource prices aong with the suppy-demand reationship. The market is easy to impement in IIU ederated coud, can promote open sharing o resources and improve resource utiization. Keywords: IIU, ederated coud, resource sharing, dynamic pricing, market 1 Introduction In order to sove the present probem in China Internet appication innovation such as Internet experiment resources distribution uneven, the cooperation in Internet technoogy innovation team is not uy etc., the ministry o education science and technoogy deveopment center took the ead ounding the Internet Innovation Union (IIU). IIU use the technica means o ederated coud to achieve integration and sharing o the Internet technoogy equipment resources among dierent experimenta patorms in dierent regions. Now at the preiminary stage o the union construction, the ministry o education science and technoogy deveopment center oers an introductory inancia support to encourage members to provide shared resources and maintain the deveopment o the union. The members o the union can use a the resources in certain conditions. Though this way can provide a good environment and saeguard or sharing resources, it negects the initiative o the subect o resource sharing, and aso ignores the obective aws in the process o resources sharing. Athough coud computing technoogy [1-2] is deveopment rapidy, resources suppy reasonaby or coud providers is sti a great chaenge. On the one hand the providers can provide service or most o the requests ony need a imited number o resources, but this woud resut in reusing cients in the peak time. On the other hand, ow utiization rate o resources in o-peak time eads the providers to improve the CCSC 213, ASTL Vo. 22, pp. 17-175, 213 SERSC 213 17
Proceedings, The 2nd Internationa Conerence on Coud-Computing and Super-Computing service price to maintain proit. The ederation overcomes these imitations by dynamic resource sharing among the providers. Eduardo et a. [3] studied the mechanism o resource sharing based on the genera equiibrium theory to cacuate the equiibrium price to coordinate between providers in ederated coud. Marian et a. [4-5] proposed a strategy-proo dynamic resource pricing strategy o mutipe resource types based on reverse auction. Buyya et a. [6-7] designed and impemented a resources market in which coud users trade resources by consutations and negotiations. In this paper, a dynamic pricing trade market or shared resources was proposed according to the use characteristics o shared resources in IIU ederated coud and the market mechanism in the process o resource share, which aimed to promote the sustainabe deveopment o the open sharing o resources. 2 Shared resource market o IIU This paper proposes a trade market comprises o ederation members and market agent or shared resources to coordinate the resources trading among members in IIU ederated coud. When a ederation member puts their ide resources into the trade market, the member is a resource provider in the market. Whie a ederation member need to access to resources rom the market the member is a consumer. The ide time o the resources shared by are the commodity in the market. The market agent is responsibe or coecting the situation o ide resources and resources request submitted by the ederation members, pricing the resources, processing resources trade, reeasing resources price and sending to the ederation members. As the IIU ederation members are in dierent regions and edera shared resources are distribution cross-regiona, to reduce the transmission cost o experimenta data, the distance nearby principe is used by the market agent to process resources requests in the process o resources trading. The market agent dea with the resources requests received in accordance with irst come irst served strategy. 2.1 Ide time o shared resources The IIU ederation members can determine the condition o resources ide and absent in advance because the experimenta resources are used ony by themseves and usuay they woud have cear pans or the use o resources. Assume there are L members in the IIU ederated coud, use S i to indicate i th ederation member, where i [ 1, L] ; Assume the number o resources shared by ederation member S i is m i, use a ik to indicate k th resource shared by the member S i, where k [ 1, mi ] ; Assume there are n types shared resources in the IIU ederated coud, use D to indicate th recourse in the ederation, where [1,n ]. Use n to indicate the number o resource D. Use set S i to indicate resources D owned by member S i, where S i ={a ik a ik is resource D, k 1, m ] }.So that [ i 171
Dynamic Pricing Trade Market or Shared Resources in IIU Federated Coud L n = S i i= 1. (1) The unit time or trading the resources ide time is 1 hour, each Shared resources have 24 units a day time. In each unit time resources can be one o the oowing three states: Nontradabe, Tradabe and Traded. Using array A ik to indicate the status o k th resource o IIU member S i in one day, thereore recourse is Nontradabe in time sot t [ t] = 1 recourse is Tradabe in time sot t t [,23] 2 recourse is Traded in time sot t A ik. (2) Ater the ederation members determined the pan o resources using in one day, the date and the array A ik o ide resources in that day is submitted to the market agent. I a ederation member needs some resource at a certain time, the member coud send resources request the market agent to buy the ide time o the resource beore that time coming. The market agent processes the request according to the resource price at the request received time. 2.2 Dynamic pricing o resources In the IIU ederated market a transactions are processed by the market agent, so the market agent coud obtain the suppy-demand reationship according to the resources transactions. Thereore, it is very easy to achieve the dynamic change o resource prices according to the suppy-demand reationship. The dynamic pricing strategy presented in this artice cacuates resources price by cost-pus pricing method, and dynamicay adusts the proit rate o resources according to the suppydemand reationship o resources, thus to achieve the dynamic change o resource prices aong with the suppy-demand reationship. Assume the cost o resource D need to purchase and insta is B in the IIU ederation, the depreciation rate per hour is β, the operating cost per hour is M, then the cost o resource D per hour is C = β B + M. (3) Assume the initia proit margin o resource D is R R R min max, and then the initia price o resource D is R, where P = (1 + R ) C. (4) To reduce the inuence o abnorma change o suppy and demand to the price at next time sot and to avoid arge uctuations o the resources price, in this paper the resources proit rate is adusted according to the tota suppy-demand reationship o 172
Proceedings, The 2nd Internationa Conerence on Coud-Computing and Super-Computing time sot d and time sots beore d. With I d1 h and I d2 h and E d h indicating the amount o tradabe and traded and requested o time sot d-h respectivey, the tota suppydemand reationship o time sot d and time sots beore d is Using h E d h= d = 1. (5) h h ( I d1 + I d 2 ) h= d R to indicate resources proit rate o time sot d, we have min min d 1 R d < ( R 1 d max max d 1 R = R d > ( R 1. (6) d 1 R (1 + d /1) others Then the transaction vaue o shared resources o time sot d is P d = (1 + R d ) C min (1 + R ) C max = (1 + R ) C d 1 P + C R d 1 d d d /1 < ( R > ( R min max others d 1 d 1 1 1. (7) The market agent cacuates the transaction price o resource D at the beginning o each time sot, and pubishes the price in the resources market. The market agent processes a requests o resource D received in this time sot according to this price. 3 Simuation experiment In this section, we present the experimenta evauation o the proposed dynamic pricing mechanism or shared resources, compare with the ixed pricing. The experiment simuates the change o the amounts o resources suppy and demand and resources price. There are three situations o suppy-demand reationship o resources in the market: short suppy and oversuppy and baanced. The three suppy-demand reationships above are origina states o the experiment. The experimenta resuts are shown in Fig. 1, 2. Fig. 1 shows when using dynamic pricing, resource price increase and decrease when short suppy and oversuppy respectivey to a stabe vaue. Fig. 2 shows when using ixed pricing suppy-demand reationship is uctuate in the vicinity o the initia vaue sighty and the suppy and demand condition wi not change, whie using dynamic pricing suppy-demand reationship is tend to be, i.e. deveop to equiibrium state. The experiment resuts indicates that using dynamic pricing in 173
Dynamic Pricing Trade Market or Shared Resources in IIU Federated Coud the shared resources market, increase resource price when short suppy to inspire the members to share more resources and decrease price when oversuppy to promote resource demand, can achieve a baance between suppy and demand and thus to promote Eicient sharing o resources. Fig. 1. Resource price Fig. 2. Suppy-demand reationship 174
Proceedings, The 2nd Internationa Conerence on Coud-Computing and Super-Computing 4 Concusions The resources sharing approach so ar in IIU ederated coud ignores the market mechanism in the process o resources sharing, it is adverse to the sustainabe deveopment o resources open sharing. This artice anayses the use characteristics o resources in the IIU ederated coud, ederation members can predict whether the resources are ide or acking according to the experiment pan. A dynamic pricing trade market or shared resources comprises o ederation members and market agent was proposed, to achieve eicient open sharing o resources. The ederation members are both resource providers and resource consumers, and the ide time o shared resources is traded commodity in the trade market. The proit rate o resources is adusted dynamicay according to the statistica data o the suppy-demand reationship, so as to achieve dynamic change o resource prices aong with the suppy-demand reationship. The market is easy to impement in the IIU ederated coud, and it is beneicia to improve resource utiization and promote open sharing o resources. The dynamic pricing mechanism proposed in this paper does not consider the cost variance o same resources and consumption eve in dierent regiona and ignores the impact o non-working time on the suppy-demand reationship. The next research woud consider the impact o the actors above and other actors on resource prices in the market. Reerences 1. Armbrust, M., Fox, A., Griith, R., et a.: A view o coud computing. Communications o the ACM. 53(4), 5--58(21) 2. FOSTER, I., ZHAO, Y., RAICU, I., et a.: Coud computing and grid computing 36-degree compared. In: Grid Computing Environments Workshop, GCE 28, pp.1 1. IEEE, NJ(28) 3. Gomes, E. R., Vo, Q. B., Kowaczyk, R.: Pure exchange markets or resource sharing in ederated couds. Concurrency and Computation.: Practice and Experience. 24(9), 977-991(212) 4. Mihaiescu, M., Teo, Y. M.: Dynamic resource pricing on ederated Couds. In Custer, Coud and Grid Computing (CCGrid), 21 1th IEEE/ACM Internationa Conerence on. pp.513- -517. IEEE, NJ(21) 5. Marian, M., Yong, M. T.: Strategy-Proo Dynamic Resource Pricing o Mutipe Resource Types on Federated Couds.: In Hsu, C. H., Yang, L. T., Park, J. H., Yeo, S. S. (eds) ICA3PP 21. LNCS, vo. 681, pp. 337--35. Springer, Heideberg (26) 6. Garg, S. K., Vecchioa, C., Buyya, R.: Mandi: a market exchange or trading utiity and coud computing services. The Journa o Supercomputing. 1--22(211). 7. Caheiros, R. N.,Toosi, A. N.,Vecchioa, C.,Buyya, R.: A coordinator or scaing eastic appications across mutipe couds. Future Generation Computer Systems. 28(8), 135-- 1362(212) 175