zjem13@trc.amoco.com (James E. McAnelly) (06/05/91)
Has anyone tested the AGS+ to see what the aggregate thru-put is. I'm interested in ethernet traffic using the six port C bus cards and four of the cards in the AGS+. zjem13@trc.amoco.com
pugh@hprnd.rose.hp.com (Rex Pugh) (06/12/91)
Here's an excerpt of some results from a performance evaluation that we've done. I know you may think its a crock since we are a competitor of cisco, but the evaluation was not done for marketing ammunition. Our corporate network is built with cisco routers, and the evaluation was done to aid in their planning and procurement for future growth. ======================================================================= ROUTER PERFORMANCE ================== TABLE OF CONTENTS ================= I. Purpose II. Router Equipment and Configurations III. Test Equipment and Methodology IV. IP Routing Tests and Results V. Bridge Forwarding Tests and Results VII. Analysis/Conclusions II. ROUTER EQUIPMENT AND CONFIGURATIONS ======================================== cisco AGS+ ----------- Hardware: 1 - CSC/3 with a 30MHz 68020 and 4M bytes of memory 1 - Environmental controller 64K bytes multibus memory 64K bytes NOVRAM 1 - cBus controller with 128K bytes 35ns SRAM (FW vers 1.0) 1 Cypress 7C1901 30ns 16-bit slice processor 3 - MEC six-port ethernet interface cards FW vers 1.6 1 Cypress 7C1901 30ns 16-bit slice processor 6 SEEQ DQ8003 8 bit ethernet controller chips 6 XILINX PGA's (probably filter assist chips) 1 - MEC two-port ethernet interface card FW vers 1.1 1 Cypress 7C1901 30ns 16-bit slice processor 6 SEEQ DQ8003 8 bit ethernet controller chips 6 XILINX PGA's (probably filter assist chips) Note: They don't de-populate the 2 and 4 port MEC cards!? No Multibus cards. Software: Bridging software GS Software (GS3-BF) Version 8.2(3) Compiled Tues 12-Feb-91 12:04 by mlb System bootstrap version 4.3(2) IP Routing configuration IP network number 1 (Class A) IP subnet mask 255.255.255.0. fast switching enabled, no keepalive on any interface. no RIP, no IGRP, or any other routing protocols no bridging enabled All other interface parameters left at the defaults 60 static arp entries of the following form: arp 1.{slot}.{connector}.{host} {mac address} {arpa|iso1|snap} where {slot} is the cBus slot number of the subnet, {connector} is the MEC connector of the subnet, {host} is the host number on the subnet 1 = router, 2 = ethernet host, 3 = iso1(802.3) host, and 4 = snap host The following is a pictorial view of the cBus slots and the ports on the MEC cards. This shows the Ethernet interfaces and their subnet addresses. ___________________________________________________________ | Ether12 | Ether13 | Ether14 | Ether15 | Ether16 | Ether17 | Slot2 | 1.2.1.1 | 1.2.2.1 | 1.2.3.1 | 1.2.4.1 | 1.2.5.1 | 1.2.6.1 | |_________|_________|_________|_________|_________|_________| | Ether0 | Ether1 | Ether2 | Ether3 | Ether4 | Ether5 | Slot0 | 1.0.1.1 | 1.0.2.1 | 1.0.3.1 | 1.0.4.1 | 1.0.5.1 | 1.0.6.1 | |_________|_________|_________|_________|_________|_________| | cBus controller slot | |___________________________________________________________| | Ether6 | Ether7 | Ether8 | Ether9 | Ether10 | Ether11 | Slot1 | 1.1.1.1 | 1.1.2.1 | 1.1.3.1 | 1.1.4.1 | 1.1.5.1 | 1.1.6.1 | |_________|_________|_________|_________|_________|_________| | Ether0 | Ether1 | | Slot3 | 1.3.1.1 | 1.3.2.1 | | |_________|_________|_______________________________________| III. TEST EQUIPMENT AND METHODOLOGY =================================== Test Equipment 2 - HP 25C Vectra PC's (25MHz 386) 1 - Zenith 33MHZ 386 PC 1 - Dell 486 PC 2 - HP 4972 LAN analyzers 1 - Network General Ethernet Sniffer Test Software The four PC's were used to generate the packet streams. The software used was a special driver developed in-house called LANSTUB. It provides a simple interface for editing the packet to be transmitted and will then either transmit the selected packet continuously, or a specified number of times. The latter mode was used since it creates the fastest packet stream. Methodology The basic idea of each test is to generate an input stream(s) with the PC(s), measure the input packet rate to the router (4972 stats package), and then measure the packet rate output from the router. The performance of the router is measured by how well its output rate matches the input rate. Although it may be more effective to generate a discreet number of packets on the input side and measure the number of packets forwarded by the router, synchronizing the start of each packet stream in a multiple stream test was nearly impossible. The result of not starting each stream simultaneously. was that for a short period of time at the beginning and end of the test the router would not be receiving all of the streams. A router that could not keep up with the input of all the streams simultaneously might be able to keep up with the streams at the beginning and end. Thus the drop rate would be understated. Furthermore we could not ensure that the test would be exactly the same for each router, and the results may not be comparable. Each PC sits on a separate LAN segment generating a near media-speed stream of homogeneous packets all destined for a host on some other LAN segment. Input streams were never set up to be forwarded back onto an input stream LAN segment or some other stream's output LAN segment. We used static ARP table entries in the routers for the destination hosts, so there weren't any systems receiving the streams. The packet streams were designated by color for ease of reference (and the thinlan cables used for the streams were colored). A given stream's input and output segments were connected to the router ports listed in the test. Here is a list of the streams and their rates: Packet Generating Packet Stream PC Rate ---------------------------------- Red Dell 486 14,530 pps Yellow HP 25C 14,444 pps Black Zenith 14,533 pps Green HP 25C 14,444 pps You may wonder why we only used 4 packet streams when we had up to 20 subnets (10 streams) which could be used. Please refer to the test results to see why we didn't need more packet sources! Sample Packet: All packets were 64 bytes long. No IP options, fragmentation, or layer 2 translation needed (i.e. ethernet to 802.3). Ethernet: 00000c00e93c 02608c321740 0800 4500002e 00000000 0806aec5 01000102 01020102 00010355 55555533 3333330f 0f0f0f00 ff00ff00 00ffff25 2627 802.3: 00000c00e93c 02608c321740 002e 0606 03 45000029 00000000 0b06abc9 01000102 01020103 55555555 33333333 0f0f0f0f 00ff00ff 0000ffff 25 SNAP: 00000c00e93c 02608c321740 002e aaaa 03000000 0800 45000024 00000000 0806aece 01000102 01020103 3333330f 0f0f0f00 ff00ff00 00ffff25 Convention for Test Description The best way to describe the flows of the packet streams is with pictures. For the most part, we ignore the source and destination addresses in the test descriptions. The MAC and IP addresses of the packets generated by each of the PC's were edited to get the stream flow described by the test. For example, the single stream port-to-port test for the AGS+ can be described with the following picture: R ___ ^ v Slot 0 =0===1===2===3===4===5= which shows that the Red stream input port is port 0 of slot2 and the output port is port 1 of slot 2. The stream was forwarded from the subnet on port 0 to the subnet on port 1. IV. IP ROUTING PERFORMANCE RESULTS ================================== 1. Single stream port-port forwarding test (theoretical max of 14880 pps) ---------------------------------------------------------------------------- Single stream of packets in one port, routed to another port on same card. For both of the single stream tests, the input rate was varied by increasing the inter-packet gap. This was done to find the rate at which no packets were dropped. __________________________________ | AGS+ | |_________________________________| | R | | ___ | | ^ v | | Slot0 =0===1===2===3===4===5= | | | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 12388 12388 0% | | Red 14530 12029 17% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14530 1517 90% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14530 1517 90% | |_________________________________| Router Summary Ethernet 802.3 ----------------------------- cisco AGS+ 14488 1350 cisco IGS 5600 800 Wellfleet 12646 ??????? HP router ER 9000 8300 2. Single stream card-card forwarding test (theoretical max of 14880 pps) ---------------------------------------------------------------------------- Single stream of packets in one port, routed to another port on a different card (i.e. across the backplane). For both of the single stream tests, the input rate was varied by increasing the inter-packet gap. This was done to find the rate at which no packets were dropped. __________________________________ | AGS+ | |_________________________________| | | | Slot0 =0===1===2===3===4===5= | | v | | | R | | v | | Slot1 =6===7===8===9==10==11= | | v | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 12305 12305 0% | | Red 14531 12330 15% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 1508 1508 0% | | Red 14531 1496 90% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 1508 1508 0% | | Red 14531 1496 90% | |_________________________________| 3. Two stream port-port throughput test (theoretical max of 29760 pps) ---------------------------------------------------------------------------- Two streams each entering two ports on one card, routed to two other ports on the same card. (Note that the Bradner tests listed wellfleet at 25K pps, but they tested using two cards, and one stream per card routed port to port on the card). __________________________________ | AGS+ | |_________________________________| | _____ | | / Y \ | | ^ v | | Slot0 =0===1===2===3===4===5= | | v ^ | | \__R__/ | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 9023 38% | | Yellow 14454 9015 38% | | ------------------- | | 28985 18038 38% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 749 95% | | Yellow 14447 742 95% | | ------------------- | | 28978 1491 95% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 749 95% | | Yellow 14447 742 95% | | ------------------- | | 28978 1491 95% | |_________________________________| 4. Two stream card-card throughput test (theoretical max of 29760 pps) ---------------------------------------------------------------------------- Two streams each entering two ports on one card, routed to two other ports on a different card. __________________________________ | AGS+ | |_________________________________| | | | v v | | Slot0 =0===1===2===3===4===5= | | v v | | R| |Y | | v v | | Slot1 =6===7===8===9==10==11= | | v v | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 9034 38% | | Yellow 14454 9026 38% | | ------------------- | | 28985 18060 38% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 740 95% | | Yellow 14447 741 95% | | ------------------- | | 28978 1481 95% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 740 95% | | Yellow 14447 741 95% | | ------------------- | | 28978 1481 95% | |_________________________________| 5. Three stream port-port throughput test (theoretical max of 44640 pps) ---------------------------------------------------------------------------- Three streams each entering three ports on one card, routed to three other ports on the same card __________________________________ | AGS+ | |_________________________________| | _____ | | / Y \ | | ^ v | | Slot0 =0===1===2===3===4===5= | | v ^ v B ^ | | \__R__/ \_/ | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 5998 59% | | Yellow 14444 5998 58% | | Black 14532 5996 59% | | ------------------- | | 43507 17992 59% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 499 97% | | Yellow 14445 493 97% | | Black 14532 485 97% | | ------------------- | | 43508 1477 97% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 499 97% | | Yellow 14445 493 97% | | Black 14532 485 97% | | ------------------- | | 43508 1477 97% | |_________________________________| 6. Three parallel stream throughput test (theoretical max of 44640 pps) ---------------------------------------------------------------------------- Three streams each entering three ports on one card, routed to three other ports on a different card. All three output ports are on the same card. __________________________________ | AGS+ | |_________________________________| | | | v v v | | Slot0 =0===1===2===3===4===5= | | v v v | | R| |Y |B | | v v v | | Slot1 =6===7===8===9==10==11= | | v v v | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 6031 58% | | Yellow 14444 6028 58% | | Black 14532 6030 59% | | ------------------- | | 43507 18089 58% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 485 97% | | Yellow 14445 485 97% | | Black 14532 489 97% | | ------------------- | | 43508 1459 97% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 485 97% | | Yellow 14445 485 97% | | Black 14532 489 97% | | ------------------- | | 43508 1459 97% | |_________________________________| 7. Four stream card-card 2 card throughput test (theoretical max of 59520 pps) ---------------------------------------------------------------------------- Four streams each entering four ports on one card, routed to four other ports on a different card. All four output ports are on the same card. __________________________________ | AGS+ | |_________________________________| | | | | v v v v | | Slot0 =0===1===2===3===4===5= | | v v v v | | |R |Y |B |G | | v v v v | | Slot1 =6===7===8===9==10==11= | | v v v v | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 4525 69% | | Yellow 14446 4522 69% | | Black 14443 4523 69% | | Green 14532 4523 69% | | ------------------- | | 57952 18093 69% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 368 97% | | Yellow 14444 367 97% | | Black 14532 364 97% | | Green 14443 360 98% | | ------------------- | | 57950 1459 97% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 368 97% | | Yellow 14444 367 97% | | Black 14532 364 97% | | Green 14443 360 98% | | ------------------- | | 57950 1459 97% | |_________________________________| 8. Four stream card-card 4 card throughput test (theoretical max of 59520 pps) ---------------------------------------------------------------------------- A four card setup (cards 1-4) each card with two ports (1 & 2). ____________________________________ | AGS+ | |___________________________________| | __________ | | / \ G | | / v | | | Slot2 =12==13==14==15==16==17= | | | v | | | \ R | | | \ | | | v | | | Slot0 =0==1===2===3===4===5= | | | v | | | \ Y | | | \ | | | v | | | Slot1 =6==7===8===9==10==11= | | | v | | | \ B | | | \ | | | v | | | Slot3 =18=19=============== | | | v | | \_______/ | |___________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 4525 69% | | Yellow 14462 4523 69% | | Black 14532 4522 69% | | Green 14463 4523 69% | | ------------------- | | 57988 18093 69% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 362 97% | | Yellow 14462 369 97% | | Black 14532 367 97% | | Green 14463 361 97% | | ------------------- | | 57988 1459 97% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 362 97% | | Yellow 14462 369 97% | | Black 14532 367 97% | | Green 14463 361 97% | | ------------------- | | 57988 1459 97% | |___________________________________| 9. Four stream port-port 4 card throughput test (theoretical max of 59520 pps) ---------------------------------------------------------------------------- A four card setup (cards 1-4) each card with two ports (1 & 2). Four streams, each stream enters port one and exits via port 2 on the same card. Thus 4 parallel streams none crossing the backplane. __________________________________ | AGS+ | |_________________________________| | ___ | | ^ R v | | Slot2 =12==13==14==15==16==17= | | ___ | | ^ Y v | | Slot0 =0===1===2===3===4===5= | | ___ | | ^ B v | | Slot1 =6===7===8===9==10==11= | | ___ | | ^ G v | | Slot3 =18==19================ | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 4526 69% | | Yellow 14446 4526 69% | | Black 14443 4525 69% | | Green 14532 4517 69% | | ------------------- | | 57952 18094 69% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 368 97% | | Yellow 14444 367 97% | | Black 14532 364 97% | | Green 14443 360 98% | | ------------------- | | 57950 1459 97% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 368 97% | | Yellow 14444 367 97% | | Black 14532 364 97% | | Green 14443 360 98% | | ------------------- | | 57950 1459 97% | |_________________________________| 10. Four stream card-card 4 card throughput test (theoretical max of 59520 pps) ---------------------------------------------------------------------------- __________________________________ | AGS+ | |_________________________________| | | | v v | | Slot2 =12==13==14==15==16==17= | | v v | | |R |Y | | v v | | Slot0 =0===1===2===3===4===5= | | | | | | v v | | Slot1 =6===7===8===9==10==11= | | v v | | |B |G | | v v | | Slot3 =18==19================ | | v v | |_________________________________| | | | Ethernet | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 4485 69% | | Yellow 14446 4482 69% | | Black 14443 4480 69% | | Green 14532 4480 69% | | ------------------- | | 57952 17927 69% | | | | 802.3 | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 364 97% | | Yellow 14444 364 97% | | Black 14532 361 98% | | Green 14443 357 98% | | ------------------- | | 57950 1446 98% | | | | SNAP | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 364 97% | | Yellow 14444 364 97% | | Black 14532 361 98% | | Green 14443 357 98% | | ------------------- | | 57950 1446 98% | |_________________________________| V. BRIDGING PERFORMANCE RESULTS ================================== All bridging tests were run with no IP configured. Spanning tree was on, however. 1. Single stream port-port bridging test (theoretical max of 14880 pps) ---------------------------------------------------------------------------- __________________________________ | AGS+ | |_________________________________| | R | | ___ | | ^ v | | Slot0 =0===1===2===3===4===5= | | | |_________________________________| | | | Bridged | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 11957 18% | | | |_________________________________| Router Summary Bridging --------------------- cisco AGS+ 11957 cisco IGS 5900 HP router ER 12000 2. Single stream card-card bridging test (theoretical max of 14880 pps) ---------------------------------------------------------------------------- __________________________________ | AGS+ | |_________________________________| | | | v | | Slot0 =0===1===2===3===4===5= | | v | | | R | | v | | Slot1 =6===7===8===9==10==11= | | v | |_________________________________| | | | Bridged | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 12310 15% | |_________________________________| 3. Four stream card-card 2 card throughput test (theoretical max of 59520 pps) ---------------------------------------------------------------------------- __________________________________ | AGS+ | |_________________________________| | | | v v v v | | Slot0 =0===1===2===3===4===5= | | v v v v | | |R |Y |B |G | | v v v v | | Slot1 =6===7===8===9==10==11= | | v v v v | |_________________________________| | | | Bridged | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 3638 75% | | Yellow 14447 3640 75% | | Black 14535 3638 75% | | Green 14444 3637 75% | | ------------------- | | 57957 14561 75% | |_________________________________| 4. Four stream card-card 4 card throughput test (theoretical max of 59520 pps) ---------------------------------------------------------------------------- ____________________________________ | AGS+ | |___________________________________| | __________ | | / \ G | | / v | | | Slot2 =12==13==14==15==16==17= | | | v | | | \ R | | | \ | | | v | | | Slot0 =0==1===2===3===4===5= | | | v | | | \ Y | | | \ | | | v | | | Slot1 =6==7===8===9==10==11= | | | v | | | \ B | | | \ | | | v | | | Slot3 =18=19=============== | | | v | | \_______/ | |___________________________________| | | | Bridged | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 3638 75% | | Yellow 14447 3638 75% | | Black 14535 3635 75% | | Green 14444 3636 75% | | ------------------- | | 57957 14547 75% | |___________________________________| 5. Four stream port-port 4 card throughput test (theoretical max of 59520 pps) ---------------------------------------------------------------------------- __________________________________ | AGS+ | |_________________________________| | ___ | | ^ R v | | Slot2 =12==13==14==15==16==17= | | ___ | | ^ Y v | | Slot0 =0===1===2===3===4===5= | | ___ | | ^ B v | | Slot1 =6===7===8===9==10==11= | | ___ | | ^ G v | | Slot3 =18==19================ | |_________________________________| | | | Bridged | | Stream Input Output Drop% | | ----------------------------- | | Red 14531 3640 75% | | Yellow 14447 3640 75% | | Black 14535 3641 75% | | Green 14444 3639 75% | | ------------------- | | 57957 14560 75% | |_________________________________|
eckert@immd4.informatik.uni-erlangen.de (Toerless Eckert) (06/15/91)
From article <13670001@hprnd.rose.hp.com>, by pugh@hprnd.rose.hp.com (Rex Pugh): > __________________________________ > | AGS+ | > |_________________________________| > | _____ | > | / Y \ | > | ^ v | > | Slot0 =0===1===2===3===4===5= | > | v ^ | > | \__R__/ | > |_________________________________| > | | > | Ethernet | > | Stream Input Output Drop% | > | ----------------------------- | > | Red 14531 9023 38% | > | Yellow 14454 9015 38% | > | ------------------- | > | 28985 18038 38% | > | | > | 802.3 | > | Stream Input Output Drop% | > | ----------------------------- | > | Red 14531 749 95% | > | Yellow 14447 742 95% | > | ------------------- | > | 28978 1491 95% | > | | > | SNAP | > | Stream Input Output Drop% | > | ----------------------------- | > | Red 14531 749 95% | > | Yellow 14447 742 95% | > | ------------------- | > | 28978 1491 95% | > |_________________________________| What i don't understand is what the relevance of the numbers is. Is this meant to measure how fast the box can drop packets ? Why are there no numbers for which the box had a maximum throughput WITHOUT dropping packets, you only had this information for the first two or so diagrams ? The number of packets on the outputs will most surely be higher for the maximum number of input packets that the box can forward without dropping packets. From my understanding (and measurements on different equipment) you can have quite a sharp negative response onto packet overload on input, so the output rate can drop rapidly if you overload the box. I don't see this to be a big problem, because upper layers have to cope with overload anyway, so a negative characteristic in the router might even speed that process up ? --- Toerless.Eckert@informatik.uni-erlangen.de /C=de/A=dbp/P=uni-erlangen/OU=informatik/S=Eckert/G=Toerless bandwidth - the final frontier.
robelr@ucs.indiana.edu (Allen Robel) (06/18/91)
> > | 802.3 | > > | Stream Input Output Drop% | > > | ----------------------------- | > > | Red 14531 749 95% | > Why are there no numbers for which the box had a maximum throughput > WITHOUT dropping packets, you only had this information for the first Is anyone even *close* to this amount of traffic on one interface anyway? At 64 bytes/packet that's ~90% of an ethernet. I would question the relevance of this test. allen
robelr@ucs.indiana.edu (Allen Robel) (06/18/91)
> anyway? At 64 bytes/packet that's ~90% of an ethernet. I would
Opps, actually only about 75% (14531 X 64 X 8 bits) but still...
allen
kre@cs.mu.oz.au (Robert Elz) (06/18/91)
robelr@ucs.indiana.edu (Allen Robel) writes: >> anyway? At 64 bytes/packet that's ~90% of an ethernet. I would >Opps, actually only about 75% (14531 X 64 X 8 bits) but still... Your 90% was closer, if not an under estimate, add 9 us per packet for inter-frame gap, and you get another 13%, then add the preamble bits, and you get more ... I think the original implied that this was the maximum rate that they could transmit at, so its probably as close to 100% actual utilisation of the wire as you can get with packets that small. As to its relevance ... none at all, while complete ethernet utilisation is possible, its not rational with packets that small. You want your router to be able to handle a full ethernet at 1024 byte (and greater) packet sizes, but if a net is flodded with baby packets its probably a feature that most of the things are discarded. kre