List of Figures

1. The graphic on the left depicts the basic STREAMS architecture. The figure on the right shows the STREAMS message.
2. This figure depicts a typical STREAMS-based TCP/IP stack with well-defined service interfaces.
3. This figure shows average round-trip time for TLI, Sockets, and DLPI on Solaris over Fast Ethernet.
4. This figure depicts the stacks used in the Solaris DLPI and TLI tests.
5. This figure shows the throughput achieved for TLI, Sockets, and DLPI on Solaris over Fast Ethernet. CPU usage is also shown.
6. This figure depicts the stacks used in the Windows NT DLPI and TLI tests.
7. This figure shows average round-trip time for TLI, Sockets, and DLPI on Windows NT over Fast Ethernet.
8. This figure shows the loopback STREAMS pseudo device driver resident in the kernel.
9. This figure shows the loopback throughput achieved on Solaris with the pseudo-driver and write/read, putmsg/getmsg system calls. CPU utilization is also included.
10. These figures show the Gigabit Ethernet sender and receiver throughput on Solaris for both DLPI and TLI.
11. This figure shows throughput achieved using TLI over Gigabit Ethernet up to the maximum Transport Service Data Unit (65507 bytes).
12. This figure shows the BSD and STREAMS stacks.
13. This figure shows Loopback TCP throughput for BSD vs. STREAMS and Socket Buffer sizes 4K and 8K.
14. This figure shows Loopback TCP throughput for BSD vs. STREAMS and Socket Buffer sizes 32K and 61K.
15. This figure shows network TCP Round-Trip Time for BSD vs. STREAMS and Socket Buffer size 4K and 8K.
16. This figure shows network TCP Round-Trip Time for BSD vs. STREAMS and Socket Buffer size 32K and 61K.
17. This figure shows Loopback TCP Round-Trip time for BSD vs. STREAMS and Socket Buffer size 4K and 8K.
18. This figure shows Loopback TCP Round-Trip time for BSD vs. STREAMS and Socket Buffer size 32K and 61K.
19. This figure shows how the STREAMS queue is organized.
20. This figure shows STREAMS stack used in the priority band tests.
21. This figure shows two clients sending to one receiver using the modified Fast Ethernet driver MHME, over a timeline. Message 1 is received by the server on band 2. However, Message 2 is not delivered to the application on band 1.
22. This figure shows two clients sending to one receiver using the modified Fast Ethernet driver MHME, over a timeline. A continuous stream of messages on band 1 is sent first; afterwards a stream of messages on band 2 is sent taking priority over band 1.
23. This figure shows two clients sending to two servers using the modified Fast Ethernet driver MHME, over a timeline. The server processes (P2,P3) are running on the same host, but listening on different DLSAP addresses.