maiakaat

VMWare install rev 3

Nov 23rd, 2011
34
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 3.60 KB | None | 0 0
  1. Host hardware:
  2.  
  3. 8GB DDR2 Memory scalable to 32-64GB
  4. (current memory throughput < 6000GB/s)
  5. (upgraded max memory throughput 12000GB/s)
  6.  
  7. 2 x 3Ghz Xeon Core 2 CPU's scalable to 3Ghz Quad Core
  8. (8 core total - no HT max)
  9. (SSE3, VT-x, x64)
  10.  
  11. 2 x SATA 500GB Disks Western Digital RE3
  12. (Scalable to 600GB SAS 15k disks or SSD)
  13. (Option to attach external SAS unit with up to 28 Direct Attached disks over 4 channels)
  14.  
  15. 1 x HP LSI 1078 P400 Raid with 256MB DDR2 64bit Battery backed RAID
  16. (Upgradeable to 512MB Max, or dual LSI 1078 with redundant batteries)
  17. (35000 IOPS per second max)
  18. (PCIe x8 1.x - 2.5Gbps x 8 = 2GB/s)
  19. (8 SAS 3Gbps channels = 2400GB/s)
  20. (Advanced data guard serial enabling raid 6 + 60 in addition to 5+50)
  21. (multiple paths to storage, redundant controller support)
  22.  
  23. 2x Broadcom onboard LAN 5721, with ipv6, tcp offload
  24. (LAN upgradeable to any PCI-e x8/x16 LAN card such as Dial 10Gbe or 4x 1Gbe Intel Pro/1000)
  25.  
  26. 1 x Switch is Netgear 748T with Jumbo frame, 96Gb backbone
  27.  
  28.  
  29. Partitioning Chosen:
  30.  
  31. 1 array created from 2 disks SATA 7.2k
  32. Physical disk 1 25GB for VMWare console and host
  33. Physical disk 2 240GB for Data store 1 (untamed + compatible VMs)
  34.  
  35. 64KB stripe (if applicable)
  36. The controller has acceleration enabled
  37. The physical disks have write cache enabled
  38. The logical disks have acceleration read and write enabled
  39. Acceleration can be disabled per logical disk,
  40. Possible read or write enablement only per logical disk
  41.  
  42. Partitons for VMWare:
  43.  
  44. / -reduced to 4GB
  45. /swap -increased to 1600MB
  46. /usr -added at 2048MB
  47. /var/log -added at 10GB
  48. /tmp -added at 4GB
  49.  
  50. SPACE REQUIRED FOR HP MANAGEMENT AGENTS
  51.  
  52. Untamed VM settings:
  53.  
  54. VIM used to create second data store
  55. 1MB block size
  56. Host as fixed size 60GB
  57. VM swap allocated on same partition
  58. Website root moved outside of VM on same partition with incremental allocation
  59. Temp moved outside of VM on same partition as website root
  60. Logs moved outside of VM on same partition as website root
  61. LSI SAS adaptor
  62.  
  63. 2 CPU's allocated with reserved 500MB
  64. 2GB RAM reserved with +4GB available
  65. 1 x ESXs VMXNet 3rd generation LAN with tcpip offloading?
  66.  
  67. Installation of VMWare tools
  68.  
  69. Performance Disk only:
  70.  
  71. VMWare write latency < 5ms
  72. VMWare read latency peak 25ms
  73.  
  74. bench32.exe against LSI 1068 hardware raid with identical disks with 64MB total size
  75. 2.0x RW 512kb bloack sizes ay 7MB/s
  76. 0.8x R below 4KB block sizes rising at 25MB/s
  77. 1.5x W below 4KB block sizes rising at 30-40MB/s
  78. 5x RW at 64KB block sizes at 450-550MB/s
  79. 8x-10x W peak above 128KB blocks 750MB/s
  80. 4-8x write above 128KB blocks at 800MB/s
  81.  
  82. bench32.exe against LSI 1068 hardware raid with identical disks with 256MB total
  83. peak at 2x write 150MB/s
  84. peak at 1.0x read <200MB/s
  85.  
  86. iometer 32KB against Promise TX4650 - comparable to LSI (read 100%)
  87. 5000 IOPS (Promise has 1500 IOPS)
  88. write latency <2ms (0.5ms typical) vs 8ms
  89. read latency (unknown) vs 8ms
  90. average latency (unknown)
  91. >90MB/s) vs <30MB/s (I think - ratio was about 3 to 1)
  92. iometer 4KB against Promise TX4650 - comparable to LSI (read 25% write 75%)
  93. 2500 IOPS (Promise has 500 IOPS)
  94. write latency <2ms (0.5ms typical) vs 8ms
  95. read latency (unknown) vs 8ms
  96. average latency (unknown)
  97. >40MBs (I think) vs <9MB/s (I think)
  98. iometer 16KB against Promise TX4650 - comparable to LSI (read 50% write 50%)
  99. 10000 IOPS (Promise has <1500 IOPS)
  100. write latency <2ms (0.5ms typical) vs 8ms
  101. read latency (unknown) vs 8ms
  102. average latency (unknown)
  103. >20MBs (I think) vs <6MB/s (I think)
  104.  
  105. *last two IOmeters may have been other way around (ie 16KB then 4KB)
Advertisement
Add Comment
Please, Sign In to add comment