How Small and Medium Enterprises (SMEs) Should Bid for Spot Instances of Amazon's EC2 Cloud

In cloud service provisioning, spot instances are spare slots for which it has no pre-booking, unlike reserved or on-demand instances for which a cloud service provider (CSP) has a priori booking. CSPs like Amazon prefer spot instance approach to sell their "idle" computing resources as and when these idle slots appear. Though they price the spot instances dynamically depending on supply-demand status, usually the spots instances are relatively cheap. Hence, Amazon's spot instances are an attractive option for IT managers in small and medium enterprises (SMEs) that normally have sporadic requirements for resources. However, SMEs have to win their desired spot instances through the auction mechanism conducted by Amazon. Since the IT manager always looks for finishing her job quickly within some specified budget, finding how to bid for spot instances in order to stay within its limited budget is a challenging task for her. She may continue to consume spot instances as long as her bid exceeds the current spot price. But, if she loses at any point, the unfinished task must be put on hold by some checkpointing mechanism so that the task may resume from the same point when she wins the spot next time. Using simulations for a very popular cloud, namely Amazon EC2, it has been found that, at a lower bid price, OPTIMAL checkpointing leads to a total cost higher than the total HOURLY checkpointing cost on a much higher bid value. Therefore, SMEs should go for higher bid prices when using OPTIMAL checkpointing and lower bid prices with HOURLY checkpointing. In the process, the author has observed some interesting correlation among checkpoint strategy, task reliability and completion time, which is reported here.