The goal of this paper is to provide a methodology for determining bandwidt
h requirements for various hardware components of a World Wide Web server.
The paper assumes a traditional symmetric multiprocessor (SMP) architecture
for the web-server, although the same analysis applies to an SMP node in a
cluster, The paper derives formulae for bandwidth demands for memory, proc
essor data bus, network adapters, disk adapters, I/O-memory paths, and I/O
buses. Since the web workload characteristics vary widely, three sample wor
kloads are considered for illustrative purposes: 1) standard SPECweb96, 2)
a SPECweb96-like workload that assumes dynamic data and retransmissions, an
d 3) WebProxy, which models a web proxy server that does not do much cachin
g and, thus, has rather severe requirements. The results point to a few gen
eral conclusions regarding Web workloads. In particular, reduction in memor
y/data bus bandwidth by using the virtual interface architecture (VIA) is v
ery desirable, and the connectivity needs may go well beyond the capabiliti
es of traditional systems based on the traditional PCI-bus. Web workloads a
lso demand a significantly higher memory bandwidth-than data bus bandwidth
and this disparity is expected to increase with the use of VIA. Also, the c
urrent efforts to offload TCP/IP processing may require a larger headroom i
n I/O subsystem bandwidth than in the processor-memory subsystem.