Lecture 30

• First, we finished the notes on protocols.
• Then, we began to look at a case study of a E-business service: an online auction site. A CBMG (Customer behavior model graph) was generated for the site, with nodes for Entry(e), Home Page (h), Search(s), Login(g), View Bids (v), Create Auction (c), and Place Bids (b).

  Workload characterization determined the existence of two user groups, A and B, and transition probabilites from one page (state) to another (p_xy, where x=source state and y = target state) were calculated. A spreadsheet containing these values can be found here:

  • Chap8-CBMG.xls
• A set of equations was generated to solve for each V_x (x = state):
• 1. V_e = 1
  2. V_h = V_e * p_eh = 1
  3. V_s = V_h * p_hs + V_s * p_ss
  4. V_v = V_s * p_sv
  5. V_g = V_h * p_hg + V_s * p_sg + V_v * p_vg
  6. V_c = V_g * p_gc
  7. V_b = V_g * p_gb

• We can manipulate eq. 2 above to solve for V_s = p_hs / (1 - p_ss), and then "plug-and-chug" to solve for all other values of V, which can also been seen in the spreadsheet.