DBS: Distributed Benchmark System

Go to DBS Top Page

Measurement Results by DBS

We show you the results of several experiments by DBS : Ethernet, VSAT satellite link, ATM, FDDI in our campus network and WAN which consists of T1 exclusive line running as the WIDE Internet. You will find that DBS has the capability of measuring and analyzing the network performance in more detail, and their property cannot be realized by other tools.

Comparing DBS with Netperf and ttcp
Ethernet
This environment is that two Ethernets are connected by a router.
VSAT Satellite Link Part I
It seems that TCP Window Scale option works well on the long-delay satellite link when there is one data stream. But TCP Slow-Start is really slow.
VSAT Satellite Link Part II
TCP may not work well on a long-delay satellite link when it was transfered in multiple data streams.
VSAT Satellite Link Part III
Can 16 data streams fill the satellite link whose bandwidth is 2Mbps?
FDDI
You will see that FDDI is familiar with TCP/IP.
ATM
You will see that ATM is unfamiliar with TCP/IP.
HIPPI over ATM Part I
You will see that the catalog spec may not trust.
HIPPI over ATM Part II
The controller chip of HIPPI-ATM code/decode machine are remade by NEC referred the result of ' HIPPI over ATM Part I '. Under the same condition, performance is improved. But when there are many data streams, those throughput will become unfair.
HIPPI over ATM Part III
The controller chip are remade again referred the result of ' HIPPI over ATM Part II'. It seems that the performance is improved by this fix. If there was not DBS, the tunning of the controller would be difficult.
WAN (T1 Link) Part I
The packet loss rate is about 1% when it was measured under practical network
WAN (T1 Link) Part II
The packet loss rate is about 30% when it was measured under practical network.