A Formal Temporal Log Data Model for the Global Synchronized Virtual Machine Environment Sean Thorpe 1 , Indrajit Ray 2 , Indrakshi Ray 3 , Tyrone Grandison 4 1 Faculty of Engineering and Computing, University of Technology Kingston, Jamaica sthorpe@utech.edu.jm 2. Department of Computer Science, Colorado State University, Fort Collins, USA indrajit@cs.colostate.edu 3 Department of Computer Science, Colorado State University, Fort Collins, USA iray@cs.colostate.edu 4 IBM Research Yorktown Heights, NY, USA, tyroneg@us.ibm.com Abstract: The use of timestamps is fundamental to the management of time varying information and arguably it may be even more important for the synchronization of the virtual machine (VM) log data sets. In the context of managing (VM) logs for transactional database activity, the consistency of its state can be evaluated by these timestamps. Temporal data models claim to be point based whereas other temporal models are interval based. Hence the premise for synchronization as a component of a time event has become critical to a distributed hybrid compute cloud. The contributions of this paper apply the use of formal temporal mechanisms to appreciate the behaviour of our case study deployment. In our study we design a software application called a global virtual machine log auditor. We use the auditor to synchronize virtual server log events across a suite of non native VM environments in distinct time-zones. This work is useful in managing cloud data migration and synchronization across these time zones. Our implementation uses a snapshot equivalent approach to monitor the synchronized log events on these VMs. In this context the paper precisely defines the notions of point based and interval based temporal data models as the application of the case scenario, thus providing a new and formal basis for characterizing such models within the cloud computing environment. This paper’s motivation is an adoption of earlier work done [1, 4 15, 21]. Keywords: timestamps, interval logs, point, cloud, temporal I. Introduction Temporal data models include timestamp attributes in their relation schemas and give special semantics to the values of these attributes in their query languages. Virtually all data models intended for practical use employ some form of intervals for their timestamp values. Unfortunately, It is generally impractical to record individually all the time points for a distributed virtual machine database. For the purposes of our ongoing work [1] we manage and archive system event logs over periodic intervals as a function of the timestamps. Intervals may simply be employed for reasons of practicality, i.e. as syntactical shorthands for time points [12]. Thus, referring to a data model as interval-based simply if it employs interval timestamps bears little real significance. It says little about the qualities of the data model. Rather for our synchronized VM log environment, the notion of point and interval based data model must be defined on a semantic level. The questions then what are the real defining properties of point and interval-based data models as a function of the synchronized temporal data model for the VM environment. This paper provides an answer to this question. To get a real feel for the range of possible semantics of time data models, it is instructive to consider a simple example adopted from Bohlen et. al. [21]. We assume that the two tuple time-stamped relations r 1 and r 2, below are given and consider possible definitions of the temporal difference of these two relations, r 1 – T r 2 . r 1 : A TS a [1,10] a [11,20] a [21,30] r 2 : A TS a [5,15] When we want to construct a difference operator between both relations, there are four possible definitions-: R 1 through to R 4. R 1 : A TS a [1,4] a [16,30] Journal of Information Assurance and Security. ISSN 1554-1010 Volume 6 (2011) pp. 398–406 © MIR Labs, www.mirlabs.net/jias/index.html Dynamic Publishers, Inc., USA