Most current web application use small objects to increase the data transfer speed between server and client. This leads to completion of the object transfer while the TCP is still in the initial slow-start phase, which starts with an initial window size. SCTP, a new transport layer protocol, uses congestion control mechanism that is similar to that of TCP but complements TCP's deficiencies in web applications. This paper presents an analytical model of object transfer latency for HTTP over SCTP as a function of the initial window size during the initial slow-start phase. Validation of our model using experimental testbed shows that our model results are within 4% of experimental results.