Mylex AcceleRAID150 MegaRAID Express Plus Version 0.90 August 199
|
|
|
Top Searches for this datasheet
3HUIRUPDQFH %HQFKPDUNV
Mylex AcceleRAID150 MegaRAID Express Plus
Version 0.90 August 1998
34551 Ardenwood Blvd. Fremont, 94555-3607 (510) 796-6100
Table Contents
Introduction Performance Analysis Benchmark Configuration.4 Server Configuration Controller Configuration.4 Mylex AcceleRAID 150: MegaRAID Express Plus 466:.4 Benchmark Test Program.4 Benchmark Results OLTP (SQL Environment) Application Performance Comparison:.5 OLTP Application (SQL Environment) Summary.7 File Server Application Performance Comparison:.8 File Server Application Summary Server Application Performance Comparison:.10 Server Application Summary.11 Video Server Application.12 Performance Comparison:.12 Video Server Application Summary Appendix IOMeter Overview
Bm030299.doc
Dqpv
This document presents performance comparison between Mylex AcceleRAID MegaRAID Express Plus RAID controllers. AcceleRAID MegaRAID Express Plus similar products competing same server market. Both RAID adapters have SCSI chip card provide SCSI channel. Both RAID adapters Intel i960RP3V33 processors Symbios Logic 53C895 SCSI chips adapter. However, when compared with industrial standard IOMeter, AcceleRAID clearly shows overall better performance compared with similar designed counterpart, MegaRAID Express Plus 466. SCSI channels controllers were used testing Intel N440BX motherboard. Since both controllers targeted server segment market, where RAID level predominately used high performance fault tolerance, comparison this document will concentrate controllers' performance RAID level environment, except video sever environment.
This paper focuses performance comparison based parameters that simulate real world application environment. Every effort been made ensure optimal configuration both controllers. Application environments modeled this benchmark follows
OLTP-Online Transaction Processing (SQL Environment) File Server Server Video Sever
Each above environments specific requirements with varying queue depths utilization criteria. Test Program used industry standard Intel's IOMeter, version 1998-01-05. details these set-ups test results discussed this document.
Bm030299.doc
Server Configuration
System Intel N440BX with dual Pentium 300MHz CPU, 64MB SDRAM Drives Quantum Viking 4.5WSE channel running mode Operation System Windows Server with Service Pack Benchmark IOMeter Intel's Architecture (SAL), version Programs 1998.01.05
Mylex AcceleRAID 150:
Firmware 4.06-0-52 BIOS DAC960 4.10-33 RAID Configuration RAID stripe size, cache segment Write back enabled EDO/Parity memory Miniport Driver DAC960NT.sys, version 5.10
MegaRAID Express Plus 466:
Firmware Version T40A BIOS Version 1.23 Configuration* stripe size Read policy: read normal, Cache policy: direct Write back enabled DRAM Miniport Driver MRAID35X.sys, version 2.12
Note: MegaRAID Express Plus 466, different combinations read policy with stripe size ranging from (default value) 128KB were carefully tested, above controller configuration were chosen best performance results.
IOMeter, developed Intel's Server Architecture (SAL) used generate disk access pattern workload based application characteristics. Appendix further information about IOMeter IOMeter result.
Bm030299.doc
OLTP environment characterized many users acting small blocks data throughout more databases. mixture read-only updateintensive transactions. profiles mostly small random block read requests byte size, with moderate small block writes. read/write ratio approximately 67:33. OLTP application, critical factors affecting overall subsystem performance throughput rate (IO/sec), utilization, effective rate (IO/sec divided utilization), average response time. these factors presented performance comparison these controllers.
Performance Comparison:
IO/sec Rate
Queue Accle150 AMI466
Raid OLTP Rate
IO/sec
Queue Depth
Accle150
AMI466
Bm030299.doc
tiliz atio
ueue I466
IO/sec
0.88 1.38 1.76 2.01 2.07 0.98 1.57 1.48 1.76
Effec tive
ctive
ueue I466
cle15 I466
Average Response Time (msec)
Queue Accle150 AMI466
12.7 20.2 36.7 64.5 122.7 243.1 468.9
10.5 15.4 23.5 58.4 148.3 290.7 557.1 1092.7
Response Time Raid
1200.0 1000.0 800.0
msec
600.0 400.0 200.0
Queue Depth
Accle150
AMI466
Bm030299.doc
IO/Sec Avg.Resp. Time
Accle150
IO/sec
Resp. Time (ms)
1000
1200
AMI466
OLTP Application (SQL Environment) Summary
MegaRAID Express Plus peaks IO/sec when queue depth equals then drops significantly (down IO/Sec) queue depth increases. AcceleRAID continues improve queue depth increases reaches IO/sec queue depth times greater IO/sec than that delivered MegaRAID Express Plus 466. normal OLTP application load, i.e. outstanding controller, AcceleRAID delivers times more IO/sec than MegaRAID Express Plus does. AcceleRAID much faster response time times faster than MegaRAID Express Plus 466) while delivering highest host queue depth. This very critical DBMS application.
Bm030299.doc
File servers predominantly RAID level when available, since RAID provides best efficiency performance capacity with full protection from single drive failures. benchmark system configuration reflects typical file server with drives channel with 27GB capacity normal operation mode, failed drive degraded mode. general, requests vary from bytes bytes mostly they around bytes, with read/write ratio 4-to-1. Although single client access sequentially, from server's perspective, accesses random. Here, performance measured data transfer rates (MB/sec).
Performance Comparison:
Normal Mode Operation (MB/sec)
Queue Accle150 AMI466
MB/sec Host orkload (Normal Mode)
1.21 1.94 2.74 3.42 3.75 4.31 4.55 4.58 4.84
1.11 1.64 2.30 2.81 3.22 2.28 2.30 2.33 2.38
6.00 5.00 4.00 MB/sec 3.00 2.00 1.00 0.00 Queue Depth
Accle150
AMI466
Bm030299.doc
Degraded Mode Operation (MB/sec)
Queue Accle150 AMI466
1.03 1.54 1.95 2.23 2.48 2.52 2.56 2.63 2.78
0.91 1.24 1.57 1.86 1.97 1.36 1.32 1.33 1.36
MB/sec Host Workload (Degraded Mode)
3.00 2.50 2.00
MB/sec
1.50 1.00 0.50 0.00
Queue Depth
Accle150
AMI466
File Server Application Summary
normal mode operation, MegaRAID Express Plus 466's bandwidth peaks 3.22 MB/sec when queue depth equals then drops significantly (down 2.28 MB/sec) queue depth increases. AcceleRAID continues improve reaches 4.84 MB/sec queue depth 256, which times greater IO/sec than that delivered MegaRAID Express Plus 466. AcceleRAID outperforms MegaRAID Express Plus degraded operation again similar pattern normal mode operation. MegaRAID Express Plus peaks 1.97 MB/sec queue depth while AcceleRAID 150's bandwidth continues improve when queue depth increase reaches 2.78 when queue depth 256, which about times wider than that MegaRAID Express Plus 466.
Bm030299.doc
profile Server similar that File Server. Different from ECommerce, server applications similar regular internet/intranet browser accesses that predominately "read" with minor exceptions. server behavior simulation, mixed request sizes range from bytes 512K bytes, mostly between bytes bytes with read/write ratio 100:0. From server's aspect, access pattern random. Both normal degraded mode (with failed drive) performances measured transfer rates (MB/sec).
Performance Comparison:
Normal Mode operation (MB/sec)
Queue Accle150 AMI466
MB/sec
1.60 2.81 4.50 6.18 7.42 9.87 11.99 12.92 13.06
1.57 2.88 4.76 6.70 8.70 9.75 10.17 10.27 10.67
Transfer rate host workload (Normal Mode)
14.00 12.00 10.00 8.00 6.00 4.00 2.00 0.00
Queue Depth
Accle150 AMI466
Bm030299.doc
Degraded Mode Operation (MB/sec)
Queue Accle150 AMI466
1.40 2.18 3.20 3.80 4.08 4.41 4.36 4.45 4.74
1.26 2.04 2.85 3.63 4.40 4.70 4.79 4.74 4.78
Degraded Performance
6.00 5.00
MB/sec
4.00 3.00 2.00 1.00 0.00
Queue Depth
Accle150
AMI466
Server Application Summary
pure read operation, AcceleRAID comes ahead 20%) MegaRAID Express Plus host loading increases. degraded mode, controllers' performance similar.
Bm030299.doc
video server environment, multiple data streams access different regions drive space. These accesses mostly sequential with byte block sizes. single worker accessing region disk space, multiple worker activities single RAID volume generating semi-randomness between regions simulates data stream. Here, worker threads total regions RAID volume. this benchmark, both RAID (with cache write back disabled) RAID (with cache write back enabled) compared. Performance measured data transfer rates (MB/sec).
Performance Comparison:
Read Sequential Access (MB/sec) orkers Accle150 AMI466 31.37 35.48 11.80 12.23 12.88 12.51 40.00 35.00 15.85 15.84 30.00 18.84 18.52 25.00 20.00 20.95 20.93 15.00 22.95 22.60 10.00 23.76 23.61 5.00 0.00 23.53 23.77 23.45 24.30
IO/sec
Raid Read Throughput
workers
Accle150
AMI466
Bm030299.doc
Write Sequential Access (MB/sec) orkers Accle150 AMI466 12.81 8.19 17.94 8.99 25.00 20.17 9.46 21.11 9.62 20.00 21.46 9.73 15.00 21.70 9.82 21.96 9.90 10.00 21.81 9.97 21.77 10.01 5.00 21.58 10.05
IO/sec
0.00
Raid Write Throughput
workers
Accle150 AMI466
Read Sequential Access (MB/sec) orkers Accle150 AMI466 23.70 30.70 11.09 12.18 35.00 13.28 12.59 30.00 15.14 15.96 25.00 17.08 19.16 20.00 18.87 21.90 15.00 20.16 23.89 19.94 24.87 10.00 19.78 25.46 5.00 19.83 25.55
IO/sec
0.00
Raid Read Throughput
workers
Accle150 AMI466
Bm030299.doc
Write Sequential Access (MB/sec) orkers Accle150 AMI466 9.94 6.31 10.17 6.35 12.00 10.39 6.35 10.00 9.22 6.35 8.00 5.69 6.35 6.00 4.19 6.34 3.48 6.34 4.00 3.31 6.34 2.00 3.26 6.33 0.00 3.20 6.33
IO/sec
Raid Write Throughput
workers
Accle150
AMI466
Video Server Application Summary
large sequential read operations, MegaRAID Express Plus ahead AcceleRAID small margin. write operations, AcceleRAID outperforms MegaRAID Express Plus times greater transfer rates RAID RAID AcceleRAID some abnormal behavior which will improved with future firmware upgrade.
Bm030299.doc
results this document were obtained using IOMeter version 1998.01.05, copyright 1998 Intel Corporation. Intel does endorse IOMeter results. More information about IOMeter obtained following address:
Specific questions about IOMeter sent iomter@intel.com.
Bm030299.doc
Other recent searches
NTMD4N03R2 - NTMD4N03R2 NTMD4N03R2 Datasheet
MMSZ5221B - MMSZ5221B MMSZ5221B Datasheet
MMSZ5259B - MMSZ5259B MMSZ5259B Datasheet
LPC210x - LPC210x LPC210x Datasheet
LPC2102 - LPC2102 LPC2102 Datasheet
HT36A4 - HT36A4 HT36A4 Datasheet
2CTD432259F1701 - 2CTD432259F1701 2CTD432259F1701 Datasheet
Privacy Policy | Disclaimer
© 2012 Datasheet Archive
